CN103036839B - MU-MAS, wireless client device and the method implemented in MU-MAS - Google Patents

Abstract

Describe the wireless client device compensating MU-MAS communication, dynamically adapting MU-MAS communication system.The wireless client device that this compensation MU-MAS communicates comprises: one or more RF unit, for receiving the signal sent from one or more MU-MAS transmitter unit, and described signal down is transformed into base band; One or more modulus (A/D) converting unit, for receive frequency reducing conversion after signal and this signal from analog signal is converted to digital signal; Frequency/phase bias estimation/compensating unit, for estimated frequency and/or phase deviation and by information feed back to transmitter for precompensation; One or more OFDM unit, for removing Cyclic Prefix and performing fast Fourier transform (FFT) to report the signal in frequency domain in described digital signal; Channel estimating unit, also responsively calculates channel evaluation data for receiving the signal exported from described one or more OFDM unit during cycle of training; And feedback generator unit, carry out using in precoding to signal before being sent to described wireless client device at signal for described channel evaluation data being sent to base station.

Description

MU-MAS, wireless client device and the method implemented in MU-MAS

The divisional application that the application is the applying date is on 08 20th, 2008, application number is 200880102933.4, name is called the application for a patent for invention of " system and method for distributed input distributed output wireless communications ".

Priority request

The application is the application NO.10/902 submitted to July 30 in 2004, the continuation application of 978.

Technical field

The present invention relates generally to field of wireless communications.Especially, the present invention relates to the system and method for the radio communication of the distributed input distributed output for using Space-time coding techniques.

Background technology

The space-time code of signal of communication

Space multiplex (MUX) and space-time code are newer development known in wireless technology.Owing to there being several antenna to be used in each terminal, so a kind of space-time code of specific type is called as " multi-input multi output " (MIMO).By using multiple antenna to send and receiving, multiple independently radio wave can transmit in identical frequency range simultaneously.Article below provides the general introduction of MIMO.

IEEE member David Gesbert, IEEE member Mansoor Shafi, IEEE member Da-shanShiu,, IEEE member Peter J.Smith and IEEE senior member Ayman Naguib IEEEJOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL.21, NO.3, APRIL 2003: " From theory to Practice:An Overview of MIMO Space-TimeCoded Wireless Systems ".

The IEEE TRANSCTIONS ONCOMMUNICATIONS of IEEE member David Gesbert, IEEE member Helmut Bolcskei, Dhananijay A. Gore and IEEE member Arogyaswami J.Paulraj, VOL.50, NO.12, DECEMBER 2000: " Outdoor MIMOWireless Channels:Models and Performance Prediction ".

Substantially, MIMO technology is the application based on the spatially distributed antenna for producing coordinate spaces data flow in common band.Radio wave is propagated by this way, thus can be separated and demodulation individual signals at receiver, even if they are at identical transmitted in band, this can cause the communication channel of independent on multiple statistical significance (being namely effectively separated).Therefore, the standard wireless communication system of multipath signal is suppressed to compare (namely with making great efforts, multiple delay time signals of same frequency, and there is amendment in amplitude and phase place), MIMO can depend on irrelevant or weak relevant multipath signal, realizes higher throughput and the signal to noise ratio of improvement in given frequency band.Example shows, under the condition that power is suitable with signal to noise ratio (snr), MIMO technology achieves higher throughput (throughput), and traditional non-mimo system only can realize lower throughput.Qualcomm's (high pass is maximum wireless technology supplier) website http:// www.cdmatech.com/products/what_mimo_delivers.jsp:the page that subscript is entitled as " What MIMO Delivers " describes this function: " MIMO is the onlymultiple antenna technique that increases spectral capacity by delivering two ormore times the peak data rate of a system per channel or per MHz of spectrum.To be more specific, for wireless LAN or Wi- applications QUALCOMM'sfourth generation MIMO technology delivers speeds of 315Mbps in 36MHz ofspectrum or 8.8Mbps/MHz.Compare this to the peak capacity of 802.1 1a/g (even with beam-forming or diversity techniques) which delivers only 54Mbps in17MHz of spectrum or 3.18Mbps/MHz ".

Usually on, due to several reason, mimo system is less than actual property restriction (improvement therefore in network is less than 10 × throughput) of 10 antennas facing to each device:

1. physical restriction: enough intervals must be had between the mimo antenna on setter, thus each signal receiving statistical iteration.Even if although still can see the improvement of MIMO throughput when the antenna spacing of wavelength mark, when antenna more close to time efficiency worsen rapidly, which results in lower MIMO throughput multiplication device.

See such as below with reference to document:

[1]D.-S.Shiu，G.J.Foschini，M.J.Gans，and J.M.Kahn，“Fadingcorrelation and its effect on the capacity of multielement antenna systems，”IEEE Trans.Comm.，vol.48，no.3，pp.502-513，Mar.2000.

[2]V.Pohl，V.Jungnickel，T.Haustein，and C.von Helmolt，“Antennaspacing in MIMO indoor channels，”Proc.IEEE Veh.Technol.Conf.，vol.2，pp.749-753，May 2002.

[3]M.Stoytchev，H.Safar，A.L.Moustakas，and S.Simon，“Compactantenna arrays for MIMO applications，”Proc.IEEE Antennas and Prop.Symp.，vol.3，pp.708-711，July 2001.

[4]A.Forenza and R.W.Heath Jr.，“Impact of antenna geometry onMIMO communication in indoor clustered channels，”Proc.IEEE Antennasand Prop.Symp.，vol.2，pp.1700-1703，June 2004.

In addition, for little antenna spacing, coupling effect each other may reduce the performance of mimo system.

See such as below with reference to document:

[5]M.J.Fakhereddin and K.R.Dandekar，“Combined effect ofpolarization diversity and mutual coupling on MIMO capacity，”Proc.IEEEAntennas and Prop.Symp.，vol.2，pp.495-498，June 2003.

[7]P.N.Fletcher，M.Dean，and A.R.Nix，“Mutual coupling in multi-element array antennas and its influence on MIMO channel capacity，”IEEEElectronics Letters，vol.39，pp.342-344，Feb.2003.

[8]V.Jungnickel，V.Pohl，and C.Von Helmolt，“Capacity of MIMOsystems with closely spaced antennas，”IEEE Comm.Lett.，vol.7，pp.361-363，Aug.2003.

[10]J.W.Wallace and M.A.Jensen，“Termination-dependent diversityperformance of coupled antennas：Network theory analysis，”IEEE Trans.Antennas Propagat.，vol.52，pp.98-105，Jan.2004.

[13]C.Waldschmidt，S.Schulteis，and W.Wiesbeck，“Complete RFsystem model for analysis of compact MIMO arrays，”IEEE Trans.on Veh.Technol.，vol.53，pp.579-586，May 2004.

[14]M.L.Morris and M.A.Jensen,“Network model for MIMO systemswith coupled antennas and noisy amplifiers，”IEEE Trans.AntennasPropagat.，vol.53，pp.545-552，Jan.2005.

And when antenna is crowded with time together, antenna must do less usually, and this also can affect antenna efficiency.

See such as below with reference to document:

[15]H.A.Wheeler，“Small antennas，”IEEE Trans.AntennasPropagat.，vol.AP-23，n.4，pp.462-469，July 1975.

[16]J.S.McLean，“A re-examination of the fundamental limits on theradiation Q of electrically small antennas，”IEEE Trans.Antennas Propagat.，vol.44，n.5，pp.672-676，May 1996.

Finally, with lower frequency and longer wavelength, the physical size of MIMO device just becomes and is difficult to process.An extreme example is at HF wave band, and MIMO device antenna must 10 meters or larger distance separated from each other here.

2. noise limit.Receiver/the transmitter subsystem of each MIMO produces the noise of certain level.When this subsystem increasing closes on mutually placement, background noise will rise.Meanwhile, when needs identify more unlike signals from multiple-antenna MIMO system time, lower background noise is just required.

3. cost and Power Limitation.Although cost and power consumption are not focuses in some MIMO application, in typical wireless product, when developing a kind of successfully product, cost and power consumption are all vital restraining factors.For each mimo antenna, need the RF subsystem be separated, comprise mould-number (A/D) and number-Mo (D/A) transducer of separation.Unlike weigh with Moore's Law scale digital system a lot of aspects (observed result of the experience aspect done by cofounder Gordon of Intel mole, the transistor size on the integrated circuit of microdevice approximately just can quadruple every 24 months; Source: http://www.intel.com/technology/mooreslaw/), analog subsystem intensive like this has certain physical structure size and power requirement usually, its size and cost and power linear proportional.Therefore, compare with single antenna devices, multiple antennas MIMO device will become extremely expensive and have surprising energy consumption.

As result above, today expection most of mimo systems be in the grade of 2 to 4 antennas, some the SNR(signal to noise ratios causing the rising of throughput 2 to 4 times and cause due to the diversity benefit of multiaerial system) rising.Anticipate the mimo system (particularly due to shorter wavelength and nearer antenna spacing in higher microwave frequency) of 10 antennas, but except special for some and to except the insensitive application of cost, be very unpractiaca more than 10 antennas.

Virtual antenna array

A kind of special applications of the technology of MIMO type is virtual antenna array.This system is suggested in the research file that Euroscience technical field research cooperation tissue proposes, EURO, Barcelona, Spain, 15-17 day in January, 2003: Center for Telecommunication Research, King ' s CollegeLondon, UK: " A step towards MIMO:Virtual Antenna Arrays ", Mischa Dohler & Hamid Aghvami.

As described in file, virtual antenna array is cooperation wireless device system (such as cell phone), it communicates mutually in the communication channel be separated (if when they enough close on mutually), instead of in their main communication channels with their base station communication, work (such as with making collaborative, if they are the GSM cell phones in UHF waveband, so this can be industrial scientific medical (ISM) the wireless wave band of 5GHz).By forwarding information between the several devices in mutual relaying scope (except in base station range), to just look like them are, and to have a device job of multiple antenna physically the same, makes the throughput hoisting that single antenna devices realizes as the MIMO potentially.

But in fact, such system is extremely difficult to be realized and use is limited.First, must keep each device now the rarest two different communication paths promote to realize throughput, the availability of its second repeated link is often uncertain.And have second communication subsystem because they are minimum and have larger computation requirement, therefore this device is more expensive, physical size is larger, and consumes more power.In addition, potentially through multiple communication link, this system depends on very complicated systematic live collaboration.Finally, because simultaneous channel usage increases (such as, use the simultaneous phone call transmission of MIMO technology), computation burden for each device also just adds (with the linear increase of channel usage, exponentially increases usually), and this is very unpractiaca to the portable unit with strict power and size restrictions.

Summary of the invention

The invention provides a kind of wireless client device used in systems in which for the frequency and phase deviation compensating multi-user multi-aerial system MU-MAS communication, this wireless client device comprises: one or more RF unit, for receiving the signal sent from one or more MU-MAS transmitter unit, and described signal down is transformed into base band; One or more modulus (A/D) converting unit, for receive frequency reducing conversion after signal and this signal from analog signal is converted to digital signal; Frequency/phase bias estimation/compensating unit, for estimated frequency and/or phase deviation and by information feed back to transmitter for precompensation; One or more OFDM unit, for removing Cyclic Prefix and performing fast Fourier transform (FFT) to report the signal in frequency domain in described digital signal; Channel estimating unit, also responsively calculates channel evaluation data for receiving the signal exported from described one or more OFDM unit during cycle of training; And feedback generator unit, carry out using in precoding to signal before being sent to described wireless client device at signal for described channel evaluation data being sent to base station.

Present invention also offers a kind of inphase quadrature (I/Q) unbalanced wireless client device used in systems in which for compensating multi-user multi-aerial system MU-MAS communication, this wireless client device comprises: one or more RF unit, for receiving the signal sent from one or more MU-MAS transmitter unit, and described signal down is transformed into base band; One or more modulus (A/D) converting unit, for receive frequency reducing conversion after signal and this signal from analog signal is converted to digital signal;

One or more OFDM unit, for removing Cyclic Prefix and performing fast Fourier transform (FFT) to report the signal in frequency domain in described digital signal; I/Q channel perception estimation unit, receives the signal exported from described one or more OFDM unit and also responsively calculates channel evaluation data during cycle of training; And feedback generator unit, carry out using in precoding to signal before being sent to described wireless client device at signal for described channel evaluation data being sent to base station.

Present invention also offers a kind of wireless client device using communication characteristic for dynamically adapting multi-user multi-aerial system MU-MAS communication system in systems in which, this wireless client device comprises: one or more RF unit, for receiving the signal sent from one or more MU-MAS transmitter unit, and described signal down is transformed into base band; One or more modulus (A/D) converting unit, for receive frequency reducing conversion after signal and this signal from analog signal is converted to digital signal; One or more OFDM unit, for removing Cyclic Prefix and performing fast Fourier transform (FFT) to report the signal in frequency domain in described digital signal; Channel estimator, also responsively calculates link-quality matrix for receiving the signal exported from described one or more OFDM unit during cycle of training; And feedback generator unit, carry out using during modulation/coding, precoding and user select to signal before being sent to described wireless client device at signal for described link-quality matrix being sent to base station.

Describe a kind of system and method for compensating the frequency had in the multiaerial system (MAS) of multi-user (MU) transmission (" MU-MAS ") and phase deviation.Such as, method according to one embodiment of the present invention comprises: the training signal from each antenna of base station is sent in multiple wireless client device or each wireless client device, one in this customer set up or each customer set up analyze each training signal with generated frequency bias compensation data, and in base station receive frequency bias compensation data; MU-MAS precoder weight is calculated to eliminate the frequency shift (FS) at transmitter place in advance based on described frequency offset compensation data; Described MU-MAS precoder weight is used to carry out precoding to training signal, to generate the precoding training signal for each antenna of base station; Training signal after the precoding of each antenna from described base station is sent to each wireless client device in described multiple wireless client device, each customer set up analyzes each training signal to generate channel characteristics data, and receives described channel characteristics data in described base station; Calculate multiple MU-MAS precoding weight based on these channel characteristics data, this MU-MAS precoder weight is by the interference calculated for eliminating in advance between frequency and phase deviation and/or user; MU-MAS precoder weight is used to carry out precoding to data, to generate the data-signal after for the precoding of each antenna of base station; And the pre-coded data signal after described precoding is sent to its each client device by each antenna of base station.

Accompanying drawing explanation

By reference to the accompanying drawings, below detailed description can obtain the present invention is better understood, wherein:

Fig. 1 shows the mimo system of prior art.

Fig. 2 shows the N antenna base station carrying out with multiple single antenna customer set up communicating.

Fig. 3 shows the base station of carrying out three antennas communicated with three single antenna customer set ups.

Fig. 4 shows the training signal technology used in one embodiment of the present of invention.

Fig. 5 shows the channel characteristics data being transferred to base station according to an embodiment of the invention from customer set up.

Fig. 6 shows the distributed output of multiple according to an embodiment of the invention input (" MIDO ") downlink transfer.

Fig. 7 shows multi-input multi according to an embodiment of the invention and exports (" MIMO ") uplink.

Fig. 8 show according to an embodiment of the invention by the circulation of different customers with the base station distributing throughput.

Fig. 9 shows the custom partitioning based on closing on according to an embodiment of the invention.

Figure 10 shows the embodiments of the invention used in NVIS system.

Figure 11 shows the execution mode of the DIDO transmitter with I/Q compensate function unit.

Figure 12 shows the DIDO receiver with I/Q compensate function unit.

Figure 13 shows a kind of execution mode with the DIDO-OFDM system that I/Q compensates.

Figure 14 shows a kind of execution mode of the DIDO 2 × 2 performance (performance) when having and do not have I/Q and compensating.

Figure 15 shows a kind of execution mode of DIDO 2 × 2 performance when having and do not have I/Q and compensating.

Figure 16 shows when having and do not have I/Q and compensating for the SER(symbol error rate of different Q AM planisphere) a kind of execution mode.

Figure 17 shows a kind of execution mode and do not have in different user devices position with DIDO2 × 2 performance I/Q compensates.

Figure 18 shows at desirable (i.i.d.(independent and with distribution)) there is and not have in channel a kind of execution mode of SER I/Q compensates.

Figure 19 shows a kind of execution mode of the transmitter architecture of self adaptation DIDO system.

Figure 20 shows a kind of execution mode of the receiver architecture of self adaptation DIDO system.

Figure 21 shows a kind of execution mode of the method for self adaptation DIDO-OFDM.

Figure 22 shows a kind of execution mode of the antenna arrangement measured for DIDO.

Figure 23 shows the execution mode of the array configurations for different stage (order) DIDO system.

Figure 24 shows the performance of different stage DIDO system.

Figure 25 shows a kind of execution mode of the aerial array measured for DIDO.

Figure 26 shows 4-QAM and a kind of execution mode of the functional relation of DIDO 2 × 2 performance that leads of 1/2FEC and location of user equipment.

Figure 27 shows a kind of execution mode of the antenna arrangement measured for DIDO.

Figure 28 shows DIDO 8 × 8 in one embodiment and how to produce the SE larger than the DIDO 2 × 2 with low TX power demand.

Figure 29 shows at a kind of execution mode with DIDO 2 × 2 performance in day line options situation.

Figure 30 shows the different DIDO pre-coding schemes average BER in i.i.d. channel (BER) performance.

Figure 31 shows the functional relation between the quantity of extra transmitting antenna in the snr gain of ASel and i.i.d. channel.

Figure 32 to show when to have 1 and 2 exterior antenna in i.i.d. channel SNR threshold value and for the functional relation between block diagonalization (BD) and the number of users (M) of ASel.

Figure 33 shows for being positioned at equal angular direction and having the BER of two users of different angles expansion (AS) value and every user's average SNR.

Figure 34 shows the result similar with Figure 33, but has higher angle intervals between user.

Figure 35 depicts the different value of the mean angle of arrival (AOA) for user, the functional relation between AS and SNR threshold value.

Figure 36 shows the SNR threshold value of the exemplary cases for 5 users.

Figure 37, for the situation of 2 users, provides when having 1 and 2 additional antenna, the comparison of SNR threshold value BD and ASel.

Figure 38 shows the result similar with Figure 37, but for the situation of 5 users.

Figure 39 shows the SNR threshold value for the BD scheme with different AS value.

Figure 40 shows BD and ASel for having 1 and 2 additional antenna, the SNR threshold value in the space correlation channel with AS=0.1 °.

Figure 41 shows the calculating of the SNR threshold value for two other channel conditions of AS=5 °.

Figure 42 shows the calculating of the SNR threshold value for two other channel conditions of AS=10 °.

Figure 43-Figure 44 respectively illustrates when 1 and 2 additional antenna, the functional relation between the angle spread (AS) of SNR threshold value and number of users (M) and BD and ASel scheme.

Figure 45 shows the receiver being equipped with frequency offset estimator/compensator;

Figure 46 shows DIDO 2 × 2 system model according to one embodiment of the present invention.

Figure 47 shows the method according to one embodiment of the present invention.

Figure 48 shows when having and do not have frequency shift (FS), the SER result of DIDO 2 × 2 system.

The SNR threshold performance of different DIDO scheme compares by Figure 49.

Amount of overhead needed for distinct methods execution mode compares by Figure 50.

Figure 51 shows at f _maxthe small frequency skew of=2Hz and emulation when not having integer offset correction.

Figure 52 shows the result when closing integer offset estimator.

Embodiment

In the following description, in order to the object explained, in order to provide, the present invention be understood thoroughly, illustrate multiple specific details.But, it is obvious that, for one of ordinary skilled in the art, even without some specific details, still can the present invention be realized.In addition, known construction and device is shown as block diagram format, to avoid ultimata obfuscation of the present invention.

Fig. 1 shows the prior art mimo system with transmitting antenna 104 and reception antenna 105.The throughput of such system can realize 3 times of the throughput usually realized in available channel.Have multiple diverse ways to realize the details of this mimo system, it had description in about the published document of this theme, and explanation is below by method such for description one.

Before data are transmitted in the mimo system of Fig. 1, channel is by " characterization ".This is by realizing starting that " training signal " is transferred to each receiver 105 from each transmitting antenna 104.Training signal has coding and mod subsystem 102 to generate, and is converted to analog signal by D/A converter (not shown), is then converted to RF signal by each transmitter 103 from baseband signal.Each reception antenna 105 being coupled to its RF receiver 106 receives each training signal and is converted into baseband signal.Baseband signal is converted to digital signal by D/A converter (not shown), then this training signal of signal processing subsystem 107 characterization.The feature of each signal can comprise several factors, and such as, it comprises, relative to phase place and amplitude, absolute reference signal, relative reference signal, characteristic noise or other factors of the reference signal of receiver inside.The feature of each signal is normally defined and shows the phase place of the several aspect of signal and the vector of amplitude variations when signal is transmitted by channel.Such as, in the modulation signal of quadrature amplitude modulation (" QAM "), described feature may be several phase place of multipath reflection and the vector of amplitude excursion of signal.Another one example is, in the signal that OFDM (" OFDM ") is modulated, it may be several in OFDM frequency spectrum or all phase places of single component signal (sub-signal) and the vector of amplitude excursion.

The channel characteristics received by each reception antenna 105 and corresponding receiver 106 stores by signal processing subsystem 107.After three all transmitting antennas 104 complete their training signal transmission, signal processing subsystem 107 will store three for each channel characteristics in three reception antennas 105, this results in the matrix 108 of 3 × 3, it is expressed as channel characteristics matrix " H ".Each independent matrix element H _{i, j}it is the channel characteristics of the training signal transmission of the transmit antenna 104i that reception antenna 105j receives.

In this, matrix H 108 inverts to produce H by signal processing subsystem 107 ^-1, and wait for the transmission from the real data of transmitting antenna 104.Note, the multiple existing MIMO technology described in available document can be used for guaranteeing that H matrix 108 is reversible.

In force, the content (payload) of the data transmitted delivers to data input subsystem 100.Then, before delivering to coding and mod subsystem 102, it is assigned with device (splitter) 101 and is divided into three parts.Such as, if content is the ASCII bit of " abcdef ", it just can be assigned with device and be divided into three sub-contents " ad ", " be " and " cf ".Then, every sub-content sends to separately coding and mod subsystem 102.

By using the applicable statistical independence of each signal and the coded system of error correcting capability, individually every sub-content is encoded.These comprise, and are not limited only to, Reed-Solomon coding, Viterbi coding (Viterbi coding) and enhancing coding (Turbo Codes).Finally, the modulator approach to channel is suitable is used to modulate each in the sub-content after these three codings.Exemplary modulator approach is differential phase keying (DPSK) modulation (" DPSK "), 64-QAM modulation and OFDM.Here it should be noted, diversity gain that MIMO provides allows the modulation constellation of higher plate number, and described modulation constellation is using the SISO(single-input single-output of same channel) be also feasible in system.Then, each coding and modulation after signal transferred out by its antenna 104, described transmission follow D/A converting unit (not shown) D/A conversion and each transmitter 103 RF generation after.

Suppose have enough space diversitys to be present between transmission and reception antenna, each reception antenna 105 will receive the various combination of three signal transmissions from antenna 104.Each RF receiver 106 by each Signal reception to and convert them to baseband signal, then A/D converter (not shown) carries out digitlization to signal again.If y _nthe signal received by the n-th reception antenna 105, x _nbe the signal sent by the n-th transmitting antenna 104, N is noise, and so this just can describe with following equalities.

y ₁=x ₁H ₁₁+x ₂H ₁₂+x ₃H ₁₃+N

y ₂=x ₁H ₁₂+x ₂H ₂₂+x ₃H ₂₃+N

y ₃=x ₁H ₁₃+x ₂H ₃₂+x ₃H ₃₃+N

Suppose that this is a system with three equatioies of three unknown quantitys, so Here it is, and signal processing subsystem 107 derives x ₁, x ₂and x ₃the problem (suppose that N is in enough low level, allow to decode to signal) of linear algebra:

x ₁=y ₁H ^-1 ₁₁+y ₂H ^-1 ₁₂+y ₃H ^-1 ₁₃

x ₂=y ₁H ^-1 ₂₁+y ₂H ^-1 ₂₂+y ₃H ^-1 ₂₃

x ₃=y ₁H ^-1 ₃₁+y ₂H ^-1 ₃₂+y ₃H ^-1 ₃₃

Once derive three signal x transmitted _n, they are just by signal processing subsystem 107 demodulation, decoding and error correction, to recover three bit streams originally separated by distributor 101.These bit streams merge in combiner unit 108, and export as single data stream from data output 109.Supposing the system robustness can overcome noise induced damage, and so data export 109 bit streams produced by the same with the bit stream be incorporated in data input 100.

Although described prior art systems is usually effectively until four antennas, perhaps until the antenna of 10 more than, owing to describing in the background parts of the disclosure, when having a large amount of antenna (such as 25,100 or 1000), it becomes very unactual.

Usually, such prior art systems is two-way, and return path realizes in an identical manner, but conversely, all has transmission and receiving subsystem in every side of communication channel.

Fig. 2 shows one embodiment of the present of invention, and wherein, base station (BS) 200 is configured with wide area network (WAN) interface (such as being connected at a high speed by T1 or other) 201 and provides (N number of) antenna 202 of some.We use term " base station " to refer to any wireless site carrying out radio communication with one of fixed position group of client for the time being.The example of base station can be the access point in wireless lan (wlan), or WAN antenna or aerial array.Have some customer set ups 203-207, each have single antenna, and base station 200 is wirelessly served them.Although for the object of this example, be very easy to the base station expecting being positioned at office environment, in this environment, it provides service for the user's set 203-207 being equipped with the personal computer of wireless network, but this structure will apply to a large amount of applicable cases, indoor and outdoors, here wireless client is served in base station.Such as, described base station can be positioned on cellular tower, or is positioned in television broadcast towers.In one embodiment, base station 200 is placed in ground, for (frequency of such as 24MHz) up transmission of HF frequency, so that signal is returned from ionospheric reflection, as on April 20th, 2004 proposes, sequence number is No.10/817,731, what while exercise question is SYSTEM AND METHOD FOR ENHANCING NEAR VERTICALINCIDENTCE SKYWAVE (" NVIS ") COMMUNICATION USINGSPACE-TIME CODING, pending application described is the same, it is by the agent of dispensing the application, here as a reference.

Some details be associated with base station 200 and illustrated customer set up are only used to the object of illustration, instead of cardinal principle according to the present invention is required.Such as, this base station can be connected to multiple dissimilar wide area network via wan interface 201, and it comprises private wide area network, such as those wide area networks sent for digital video.Similarly, customer set up can be wireless data processing and/or the communicator of any kind, and it comprises, and is not only confined to, cell phone, personal digital assistant (" PDA "), receiver and wireless camera.

In one embodiment, n antenna 202 of base station spatially separates, thus each sends and receives the relevant signal of non-space, and the transceiver of to just look like described base station be prior art MIMO is the same.As described in the introduction, antenna is with λ/6(i.e. 1/6 wavelength) interval place experiment make, it successfully achieves the throughput hoisting from MIMO, but in general, these antenna for base station are more separated, the performance of system is better, and λ/2 are gratifying minimum ranges.Certainly, cardinal principle of the present invention is not limited to any specific separation between antenna.

Note, its antenna can be positioned over far distance by single base station 200 well.Such as, in HF frequency spectrum, antenna can have 10 meters or farther (such as, in NVIS above-mentioned realizes).If use the antenna that 100 such, the aerial array of this base station just can occupy the area of several square kilometres.

Except space diversity reception to communicate, in order to improve the effective throughput of system, one embodiment of the present of invention are by polarizations.Improving channel capacity by polarization is a kind of known technology, and it is employed a lot of year by satellite television providers.Use Polarization technique, multiple (such as three) base station or user antenna can be made with each other closely, and it is relevant to remain non-space.Although traditional RF system mostly just benefits from the two dimension of polarization, (such as x and y) diversity, structure described herein can benefit from three-dimensional (x, y and z) diversity of polarization further.

Except space and polarization diversity, one embodiment of the present invention adopt and are close to orthogonal antenna pattern (pattern), to improve link performance via directional diagram diversity.Directional diagram diversity can improve capacity and the bit error rate performance of mimo system, and its advantage compared to other diversity antenna technologies can see following article:

[17]L.Dong，H.Ling，and R.W.Heath Jr.，“Multiple-input multiple-output

wireless communication systems using antenna pattern diversity，”Proc.IEEE Glob.Telecom.Conf.，vol.1，pp.997-1001，Nov.2002.

[18]R.Vaughan，“Switched parasitic elements for antenna diversity，”IEEE

Trans.Antennas Propagat.，vol.47，pp.399-405，Feb.1999.

[19]P.Mattheijssen，M.H.A.J.Herben，G.Dolmans，and L.Leyten，“Antenna-pattern diversity versus space diversity for use at handhelds，”IEEETrans.on Veh.Technol.，vol.53，pp.1035-1042，July 2004.

[20]C.B.Dietrich Jr，K.Dietze，J.R.Nealy，and W.L.Stutzman，“Spatial，polarization，and pattern diversity for wireless handheld terminals，”Proc.IEEE Antennas and Prop.Symp.，vol.49，pp.1271-1281，Sep.2001.

[21]A.Forenza and R.W.Heath，Jr.，“Benefit of Pattern Diversity Via2-element Array of Circular Patch Antennas in Indoor Clustered MIMOChannels″，IEEE Trans.on Communications，vol.54，no.5，pp.943-954，May2006.

By using directional diagram diversity, can make multiple base station or user antenna each other closely, and however also can not spatially be associated.

Fig. 3 provides the additional detail of an embodiment of the base station 200 shown in Fig. 2 and customer set up 203-207.In order to the object simplified, this base station 300 is only shown as three antennas 305 and three customer set up 306-308.It is to be noted, however, that embodiments of the invention described herein can with the antenna 305(of almost unlimited amount namely, only by can space and noise limit) and customer set up 306-308 realize.

Prior art MIMO similar shown in Fig. 3 and Fig. 1, wherein, both have three antennas in every one end of communication channel.Significant difference is, in the mimo system of prior art, be fixed range (such as, being integrated in single device) between three antennas 105 on the right side of Fig. 1 are mutual, the signal received from each antenna 105 is processed together signal processing subsystem 107.By contrast, in figure 3, three antennas 309 on figure right side each be coupled on different customer set up 306-308, in the scope that each described customer set up can be distributed in base station 305 Anywhere.Like this, the signal that each customer set up receives can be encoded at it, modulate, processed independent of other two signals received in signal processing subsystem 311.Therefore, compared with " MIMO " system of multiple input (i.e. antenna 105) multiple output (i.e. antenna 104), Fig. 3 shows multiple input (i.e. antenna 305) distributed output (i.e. antenna 305) system, below refers to " MIDO " system.

Note, the application uses the term usage different from application before, to meet academia and industrial practice better.At the application NO.10/817 being entitled as the common pending trial of " SYSTEM ANDMETHOD FOR ENHANCING NEAR VERTICAL INCIDENCE SKYWAVE(" NVIS ") COMMUNICATION USING SPACE-TIME CODING " that 20 days quoted before April in 2004 submits to, the application NO.10/902 that on July 30th, 731 and 2004 submits to, 978(the application is the continuation application of this application) in, the meaning of " input " and " output " (in environment of SIMO, MISO, DIMO and MIDO) and this term expressing the meaning in this application is contrary.In application before, " input " refers to the wireless signal inputing to reception antenna (such as, the antenna 309 in Fig. 3), and " output " refers to the wireless signal that transmitting antenna (such as, antenna 305) exports.In academia and wireless industry, the antisense of usual use " input " and " output ", wherein " input " refer to input to the wireless signal of channel (namely, wireless signal from antenna 305 sends), and " output " refers to the wireless signal (that is, antenna 309 receive wireless signal) that exports from channel.The application adopts this term usage, and this usage is contrary with the usage in the application of quoting before this section.Therefore, the term usage equivalent form of value between several application is below depicted:

10/817,731 and 10/902,978 the application

SIMO = MISO

MISO = SIMO

DIMO = MIDO

MIDO = DIMO

MIDO structure shown in Fig. 3 achieves for giving the transmitting antenna of determined number the capacity boost being similar to MIMO and realizing in SISO system.But, a difference of the specific MIDO embodiment shown in MIMO and Fig. 3 is, for realizing the capacity boost that multiple antenna for base station provides, each MIDO customer set up 306-308 only requires single receive antenna, and for MIMO, each customer set up at least requires and the as many reception antenna of capacity multiple of wishing to realize.Suppose the restriction usually having an implementation, its restriction can place how many antennas (as explained in the introduction) at customer set up, and on typical case, mimo system is just limited between 4 to 10 antennas (capacity of 4 times to 10 times) by this.Because base station 300 is usually from fixing and fill dynamic location-based service in a lot of customer set up, expanded to far more than 10 antennas, and be very actual by suitable distance separate antenna with implementation space diversity.As described in, each antenna arrangement has the disposal ability of transceiver 304 and part coding, modulation and Signal Processing Element 303.It should be noted that, in this embodiment, no matter base station 300 expands how many, and each customer set up 306-308 only will require an antenna 309, therefore the cost for single user customer set up 306-308 will be very low, and the cost of base station 300 can be shared in the user of large cardinal.

In Fig. 4 to Fig. 6, show and how to complete from base station 300 to the example that the MIDO of customer set up 306-308 transmits.

In one embodiment of the invention, before MIDO transmission starts, channel is characterized.For mimo system, each antenna 405 pairs of training signals transmit one by one.Fig. 4 only show the transmission of first training signal, but for three antennas 405, has three transmission separated.Each training signal, by encoding, modulating and signal processing subsystem 403 generates, is converted to analog signal by D/A converter, and is sent by each RF transceiver 404 as RF signal.Available various different coding, modulation and signal processing technology comprise, and are not limited to those above-described technology (such as, Reed Solomon, Viterbi coding (Viterbi Coding); QAM, DPSK, QPSK modulation etc.).

Each customer set up 406-408 receives training signal by its antenna 409 and converts this training signal to baseband signal by transceiver 410.A/D converter (not shown) is encoded at this signal, modulate and place that signal processing subsystem 411 processes converts thereof into digital signal.Then signal characteristic logical block 320 identifies the feature (such as, identifying above-mentioned phase place and amplitude distortion) of gained signal and this feature is stored in memory.This characteristic processing process is similar to the processing procedure of the mimo system of prior art, and a significant difference is, each customer set up only calculates an one antenna, instead of the characteristic vector of n antenna.Such as, the described training signal of known mode is by the coding of customer set up 406, modulation and signal processing subsystem 420 initialization (when producing by receiving it in the information sent, or by other initialization process).When antenna 405 sends this training signal in a known pattern time, coding, modulation and signal processing subsystem 420 use correlation method to find the strongest training signal receiving mode, phase place and amplitude excursion save by it, and then this pattern cuts by it in the middle of the signal received.Next, it finds the last the second receiving mode relevant to described training signal, phase place and amplitude excursion is saved, and then the second strong mode cuts by it from the described signal received.This process is carried out always, drops under given background noise until save the phase place of certain fixed qty and amplitude excursion (such as, 8) or detectable training signal pattern.The vector of this phase/amplitude skew becomes the element H of vector 413 ₁₁.Meanwhile, the coding of customer set up 407 and 408, modulation and signal processing subsystem perform same process, produce their vector element H ₂₁and H ₃₁.

The memory that channel characteristics is deposited can be nonvolatile memory, such as flash memory, or hard disk, and/or volatile memory, such as random access memory (such as, SDRAM, RDAM).In addition, different user's sets can use dissimilar memory to store characteristic information (such as PDA uses flash memory, and notebook computer uses hard disk) simultaneously.On various customer set up or base station, ultimata of the present invention is not limited to the storing mechanism of any particular type.

As mentioned above, according to used scheme, because each customer set up 406-408 only has an antenna, each 1 × 3 row 413-415 only storing H matrix.Fig. 4 shows the stage after the first training signal transmission, and here, the first row of 1 × 3 row 413-415 stores the channel characteristic information of first antenna of three antenna for base station 405.All the other two row store the channel characteristics of ensuing two training signals transmission from all the other two antenna for base station.Note, in order to the object of illustration, the time tranfer that described three training signal patterns are being separated.If have selected three training signal patterns thus uncorrelated mutually, so they can transmit simultaneously, therefore reduce the training time.

As shown in Figure 5, after all three pilot transmission complete, 1 × 3 row 513-515 of the matrix H stored is sent it back base station 500 by each customer set up 506-508.In order to the object simplified, only show a customer set up 506 in Figure 5 and transmit its characteristic information.In conjunction with suitable error correction coding (such as Reed Solomon, Viterbi coding (Viterbi Coding) and/or enhancing coding (TurboCodes)), suitable modulator approach (such as DPSK, 64QAM, OFDM) can be used to guarantee, and base station 500 receives the data in row 513-515 exactly.

In Fig. 5, although all three antennas 505 demonstrate Received signal strength, for the transmission receiving every 1 × 3 row 513-515, single antenna and single transceiver of base station 500 are enough.But, under certain condition, use a lot of or all antennas 505 and transceiver 504 can realize than single antenna 505 and transceiver 504 better signal to noise ratio (snr) to receive each transmission (that is, using the multiple output of the single input of prior art (" SIMO ") treatment technology in coding, modulation and signal processing subsystem 503).

When the coding of base station 500, modulation and signal processing subsystem 503 receive described 1 × 3 row 513-515 from each customer set up 507-508 time, its by described 1 × 3 row 513-515 stored in the H matrix 516 of 3 × 3.For customer set up, base station can use much different memory technologies to carry out storage matrix 516, and it includes, but are not limited to, nonvolatile mass storage (such as hard disk) and/or volatile memory (such as SDRAM).Fig. 5 shows the stage that base station had received and stored 1 × 3 row 513 from customer set up 509.When 1 × 3 row 514 and 515 transmits from all the other customer set ups time, they can be transmitted and be kept in H matrix 516, until whole H matrix 516 is stored.

With reference to figure 6, will describe from base station 600 to the embodiment that the MIDO of customer set up 606-608 transmits now.Because each customer set up 606-608 is independently device, so each device receives different transfer of data.Like this, the embodiment of base station 600 comprises at wan interface 601 and coding, modulates the router 602 they being carried out to liaison between signal processing subsystem 603, it receives multiple data flow (form is bit stream) from wan interface 601, corresponds respectively to each customer set up 606-608 by described data flow by the data flow u separated ₁-u ₃send.For this purpose, this router 602 can use various known route technology.

As shown in Figure 6, by described three bit streams, u ₁-u ₃route is entered in described coding, modulation and signal processing subsystem 603, they are encoded to the error correction stream (such as, use ReedSolomon, Viterbi or strengthen coding) of statistical iteration, and by the modulator approach (such as DPSK, 64QAM or OFDM) suitable to channel, they are modulated.In addition, the embodiment of Fig. 6 display comprises signal precoding logical block 630, and based on signal characteristic matrix 616, this signal precoding logical block 630 is for carrying out unique encodings to the signal sent from each antenna 605.Especially, in this embodiment, precoding logical block 630 is by the bit stream u of three in Fig. 6 ₁-u ₃the new bit stream u' of generation three is multiplied by mutually with the inverse matrix of H matrix 616 ₁-u' ₃, instead of each coding and modulated bit stream are routed to antenna separately (as in Fig. 1 do).Then, described three precoding bit circulations are changed to analog signal by D/A converter (not shown), and transceiver 604 and antenna 605 it can be used as RF signal to send.

Before how explanation customer set up 606-608 receives described special stream, by the operation that description precoding module 630 performs.Be similar to the example of MIMO in Fig. 1 above, in three original bit stream, the coding of each bit stream and modulated signal will be expressed as u _n.In the embodiment shown in fig. 6, each u _icomprise router 602 routes three bit streams come data, each such bit stream will become three user's set 606-608 one of them.

But, do not resemble the MIMO example in Fig. 1, there, each x _ithere is each antenna 104 to send, in the embodiments of the invention shown in Fig. 6, receive each u at each customer set up antenna 609 _i(adding noise N any in upper signal channel).For realizing such result, (we are expressed as v to the output of each in three antennas 605 _i) be u _iwith the function of the H matrix of each customer set up of characterization.In an embodiment, the precoding logical block 630 in coding, modulation and signal processing subsystem calculates each v by performing following equalities _i:

v ₁＝u ₁H ^-1 ₁₁+u ₂H ^-1 ₁₂+u ₃H ^-1 ₁₃

v ₂＝u ₁H ^-1 ₂₁+u ₂H ^-1 ₂₂+u ₃H ^-1 ₂₃

v ₃＝u ₁H ^-1 ₃₁+u ₂H ^-1 ₃₂+u ₃H ^-1 ₃₃

Therefore, unlike MIMO, there, channel calculates each x at receiver after being converted by signal _i, and embodiments of the invention described herein solved each v at transmitter before signal converts by channel _i.Each antenna 609 receives from other u for other antenna 609 _n-1the u separated in bit stream _i.Each transceiver 610 converts the signal respectively received to baseband signal, and A/D converter (not shown) carries out digitlization to it here, and each coding, modulation and signal processing subsystem 611 are to its x _ibit stream carries out demodulation code, and its bit stream is delivered to customer set up use data-interface 612(such as, the application program on customer set up).

Embodiments of the invention described herein can use multiple different coding and modulator approach to realize.Such as, in OFDM realizes, its intermediate frequency spectrum is divided into multiple split-band, and technology described herein can be used for each independent split-band of characterization.But as mentioned above, cardinal principle of the present invention is not limited to any specific modulator approach.

If customer set up 606-608 is portable data processing device, such as PDA, notebook computer and/or wireless telephonic words, so because customer set up may move to another one from a position, then channel characteristics can frequently change.Like this, in one embodiment of the invention, the channel characteristics matrix 616 of base station is constantly upgraded.In one embodiment, base station 600 periodically (every 250 milliseconds) sends new training signal to each customer set up, its channel eigenvectors is constantly sent it back base station 600 to guarantee that channel characteristics keeps accurately (such as, if environment change or customer set up move thus have influence on channel) by each customer set up.In one embodiment, in the actual data signal being sent to each customer set up, training signal is interweaved.Typically, the throughput of described training signal is far below the throughput of described data-signal, and therefore this throughput total on system will almost not affect.Correspondingly, in this embodiment, channel characteristics matrix 616 can constantly be upgraded when initiatively communicating with each customer set up in base station, thus when customer set up moves to next position from a position, or keep channel characteristics accurately time environment changes thus has influence on channel.

One embodiment of the present of invention shown in Fig. 7 use MIMO technology to improve uplink communication channel (that is, from customer set up 706-708 to the channel of base station 700).In this embodiment, the uplink channel characteristics logical block 741 in base station is constantly analyzed and characterization the channel come from each customer set up.Especially, each customer set up 706-708 sends training signal to base station 700, and there channel characteristics logical block 741 analyzes the channel characteristics matrix 741 producing N × M, and N is the quantity of customer set up here, and M is the quantity of the antenna that base station uses.Embodiment shown in Fig. 7 uses three antennas 705 and three customer set up 706-708 in base station, which results in 3 × 3 channel characteristics matrixes 741 depositing in base station 700.MIMO uplink shown in Fig. 7 can be used for data being sent it back base station 700 and channel eigenvectors being sent back base station 700 by customer set up, as shown in Figure 5.But and the embodiment shown in Fig. 5 unlike, in Figure 5, the channel eigenvectors of each customer set up transmitted with the time of separating, and the method shown in Fig. 7 allows channel eigenvectors to be transmitted go back to base station 700 from multiple customer set up simultaneously, thus greatly reduce channel eigenvectors to the impact of Return Channel throughput.

As mentioned above, the feature of each signal can comprise several factors, and such as, it comprises phase place relative to the reference signal of receiver inside, absolute reference signal, relative reference signal, characteristic noise or other factors and amplitude.Such as, in the signal that quadrature amplitude modulation is modulated, described feature can be phase place and the amplitude excursion vector of several multipath reflections of signal.Another example is, in the signal that OFDM is modulated, described feature can be several in OFDM frequency spectrum or the phase place of all single component signals and amplitude excursion vector.Described training signal can be generated by the coding of each customer set up and mod subsystem 711, and D/A converter (not shown) converts this training signal to analog signal, and then the transmitter 709 of each customer set up converts it to RF signal from baseband signal.In one embodiment, synchronous in order to ensure training signal, customer set up only transmits training signal (such as, when circulating (round robin)) when base station requests.In addition, can interweave to training signal in the actual data signal sent from each customer set up, or training signal can transmit together with described actual data signal.Therefore, even if customer set up 706-708 is mobile, uplink channel characteristics logical block 741 also can be transmitted continuously and analyze this training signal, thus guarantees that channel characteristics matrix 741 keeps upgrading.

Total channel capacity that previous embodiment of the present invention is supported can be defined as min(N, M), here, M is the quantity of customer set up, and N is the quantity of antenna for base station.That is, capacity limited by the antenna amount of base station side or customer side.So, one embodiment of the present of invention use simultaneous techniques guarantee to be no more than min(N, M within preset time) individual antenna sending/receiving.

In the typical case, the quantity of the antenna 705 of base station 700 will be less than the quantity of customer set up 706-708.Fig. 8 shows an exemplary situation, and it allows 5 customer set up 804-808 to communicate with the base station with three antennas 802.In this embodiment, determine the quantity of total customer set up 804-808 and necessary channel characteristic information detected (such as, description above) after, base station 800 selects first group with its three client 810(carrying out communicating because min(N, M)=3, so be three clients in this instance).After the fixed time that communicated with first crowd of client 810, three clients 811 that base station just selects another group to communicate with.In order to uniform distribution communication channel, two customer set ups 807,808 be not included in first group are selected in base station 800.In addition, because extra antenna is available, the extra customer set up 806 be included in first group is just selected in base station 800.In one embodiment, the throughput of each client equal number in time by this way in client masses circulation, thus can be distributed to effectively in base station 800.Such as, in order to uniform distribution throughput, any combination of except customer set up 806 three customer set ups (that is, because customer set up 806 for communicating with base station in two circulations started) then can be selected in base station.

In one embodiment, except the data communication of standard, base station can use aforementioned techniques to receive training signal and signal characteristic data to transmit training signal to each customer set up with from each customer set up.

In one embodiment, some customer set up or customer set up group can be assigned to the throughput of varying level, such as, customer set up can be distinguished order of priority, thus can guarantee that the customer set up of relatively high priority the client of lower priority must fill family and have more communication cycle (that is, more throughput).Based on the variable of some, can select " priority " of client, described variable comprises, and such as, the subscription fee to wireless bandwidth of user (such as, pay more for being willing to mean additional throughput), and/or communication is to/from data type (such as, the real time communication, such as call voice and video of customer set up, obtain the priority higher than non-realtime traffic, such as Email).

In the present load required based on each customer set up, in the embodiment of base station dynamic assignment throughput.Such as, if customer set up 804 live video stream, and other device 805-808 is performing the non real-time function of such as Email, and so base station 800 distributes relatively many throughputs can to this client 804.But it should be noted, cardinal principle of the present invention is not limited to any specific throughput distribution technology.

As shown in Figure 9, two customer set ups 907,908 can closely, make the channel characteristics of described client be the same actually.As a result, base station will receive and store the in fact equal channel eigenvectors of two customer set ups 907,908, therefore this can not produce for each client unique, the signal of spatial distribution.Correspondingly, in one embodiment, base station will guarantee that phase mutual edge distance two or more customer set ups any are closely assigned to different groups.Such as, in fig .9, first base station 900 communicates with first group 910 of customer set up 904,905 and 908, then communicates with second group 911 of customer set up 905,906,907, which ensure that customer set up 907 and 908 is in different groups.

Selectively, in one embodiment, base station 900 communicates with customer set up 907 and 908 simultaneously, but uses known multi-channel Technology to carry out multiplexing to communication channel.Such as, base station can use time division multiplexing (" TDM "), frequency division multiplexing (" FDM ") or code division multiple access (" CDMA ") technology to separately signal that is single, space correlation between customer set up 907 and 908.

Although above-mentioned each customer set up is equipped with single antenna, cardinal principle of the present invention can be realized to improve throughput by using the customer set up with multiple antenna.Such as, when being used on above-mentioned wireless system, the throughput hoisting that the client with 2 antennas will realize 2 times, the throughput hoisting that the client with 3 antennas will realize 3 times, etc. (that is, supposing that space between antenna and angular separation are enough).When the customer set up by having multiple antenna circulates time, same general rule can be applied in base station.Such as, each antenna can be regarded as client separately by it, and gives that " client " by throughput distribution, just as it is other client any, (such as, guarantees the communication cycle that each client provides enough or suitable).

As mentioned above, one embodiment of the present of invention use above-mentioned MIDO and/or MIMO signal transmission technology to improve signal to noise ratio and throughput in the incident sky wave (" NVIS ") of near vertical.With reference to Figure 10, in one embodiment of the invention, be equipped with a NVIS base station 1001 of the matrix of N number of antenna 1002 for communicating with M customer set up 1004.The antenna 1004 of described NVIS antenna 1002 and multiple user's set is about to become the angle within 15 degree signal uplink to be transmitted obtain the NVIS that wants and drop to minimum by surface wave disturbing effect with vertical direction.In one embodiment, antenna 1002 and customer set up 1004 use the above-mentioned assigned frequency of multiple MIDO and MIMO technology in NVIS frequency spectrum (such as in carrier frequency or the frequency lower than 23MHz, but be usually less than in the frequency of 10MHz) support multiple independently data flow 1006, thus the throughput (that is, being directly proportional with the quantity of the data flow with statistical iteration) significantly improved in assigned frequency.

The described NVIS antenna of serving given base station can have far physical distance each other.Suppose the long distance (the round distances of 300 miles) of propagating lower than long wavelength and the signal of 10MHz, hundreds of code, or even the antenna physical interval of several miles can provide benefit in diversity.In such a situa-tion, independent aerial signal can be withdrawn into center, processes it by traditional wired or wireless communication system.Selectively, each antenna can have local device to process its signal, then uses traditional wired or wireless communication system that this transfer of data is gone back to center.In one embodiment of the invention, NVIS base station 1001 has to internet 1010(or other wide area network) wideband link 1015, thus be supplied to customer set up 1003 long-range, at a high speed, wireless network access.

In one embodiment, base station and/or user can utilize polarization/direction figure diversity (patterndiversity) technology, with while providing diversity and promoting throughput, reduce array size and/or user distance.Such as, in the DIMO system with HF transmission, due to polarization/direction figure diversity, user can be positioned at same position and their signal can not be associated.Especially, by using directional diagram diversity, a user can communicate with base station via earthwave, and other users can communicate with base station via NVIS.

Additional embodiments of the present invention

I, utilize I/Q imbalance to carry out DIDO-OFDM precoding

One embodiment of the present invention adopt the system and method being used for compensating inphase quadrature (I/Q) imbalance had in distributed input distributed output (DIDO) system of OFDM (OFDM).In brief, according to the present embodiment, subscriber equipment is estimated channel, and by this information feedback to base station; Base station calculates pre-coding matrix, to eliminate the interference between the uneven carrier wave caused of I/Q and between user; And parallel data stream is sent to multiple subscriber equipment via DIDO precoding; This subscriber equipment forces (ZF), least mean-square error (MMSE) or maximum likelihood (ML) receiver to carry out demodulation to data, to suppress residual interference via zero.

As detailed below, some notable features of this execution mode of the present invention include, but are not limited to:

The inter-carrier interference (ICI) (caused because I/Q does not mate) of (mirror tone) is adjusted in precoding from mirror image in ofdm system for eliminating;

Precoding caused because I/Q does not mate for the inter-user interference eliminated in DIDO-OFDM system and ICI();

For eliminating ICI(caused because I/Q does not mate via the ZF receiver adopted in the DIDO-OFDM system of block diagonalization (BD)) technology;

For via the precoding (at transmitter place) in DIDO-OFDM system and inter-user interference eliminated by ZF or MMSE filter (at receiver place) and ICI(caused because I/Q does not mate) technology;

For eliminating inter-user interference via the precoding (at transmitter place) in DIDO-OFDM system and the nonlinear detector (at receiver place) that is similar to maximum likelihood (ML) detector and ICI(caused because I/Q does not mate) technology;

Use precoding based on channel condition information for eliminating in ofdm system from the inter-carrier interference (ICI) (caused because I/Q does not mate) that mirror image is adjusted;

Use precoding based on channel condition information for eliminating in DIDO-OFDM system from the inter-carrier interference (ICI) (caused because I/Q does not mate) that mirror image is adjusted;

Use I/Q not mate known DIDO precoder (I/Q mismatch aware DIDOprecoder) in base station, and use the known DIDO receiver of I/Q at user terminal place;

Use I/Q not mate known DIDO precoder (I/Q mismatch aware DIDOprecoder) in base station, use the known DIDO receiver of I/Q at user terminal place, and use I/Q known channel estimator;

I/Q is used not mate known DIDO precoder in base station, use the known DIDO receiver of I/Q at user terminal place, and use I/Q known channel estimator and the known DIDO feedback generator of I/Q (channel condition information is sent to website from user terminal by this maker);

I/Q is used not mate known DIDO precoder in base station, and use I/Q known DIDO configurator (this configurator uses I/Q channel information to perform various function, comprises user's selection, adaptive coding and modulation, empty time-frequency and maps or precoder selection);

Use the known DIDO receiver of I/Q (this receiver is eliminated ICI(via the ZF receiver adopted in the DIDO-OFDM system of block diagonalization (BD) precoder and caused because I/Q does not mate));

Use I/Q known DIDO receiver (this receiver eliminates inter-user interference via the precoding (at transmitter place) in DIDO-OFDM system and the nonlinear detector (at receiver place) that is similar to maximum likelihood (ML) detector and ICI(caused because I/Q does not mate)); And

Use the known DIDO receiver of I/Q (this receiver is eliminated ICI(via ZF or the MMSE filter in DIDO-OFDM system and caused because I/Q does not mate)).

A, background

The transmission of exemplary radio communication system and Received signal strength comprise inphase quadrature (I/Q) component.In the system of reality, this inphase quadrature component may the distortion due to the defect in mixing and baseband operations.These distortions (distortion) show as I/Q phase place, gain and delay and do not mate.Unbalance in phase is by the sine (sine) in modulator/demodulator and cosine (cosine) is correctly not orthogonal causes.Gain imbalance is caused by the different amplification between inphase quadrature component.Because the delay between I and the Q track (rail) in analog circuit is different, also may there is additional distortion, this distortion is referred to as to postpone imbalance.

In an orthogonal frequency division multiplexing (ofdm) system, I/Q imbalance can cause coming the intercarrier imbalance (ICI) of spontaneous emission tune.This impact obtains research in some data, and in following information, propose for not mating the method compensated to the I/Q in single-input single-output SISO-OFDM system: M.D.Benedetto and P.Mandarini, " Analysis of the effect of the I/Qbaseband filter mismatch in an OFDM modem; " Wireless personalcommunications, pp.175-186,2000; S.Schuchert and R.Hasholzner, " A novelI/Q imbalance compensation scheme for the reception of OFDM signals, " IEEETransaction on Consumer Electronics, Aug.2001; M.Valkama, M.Renfors and V.Koivunen, " Advanced methods for I/Q imbalance compensation incommunication receivers, " IEEE Trans.Sig.Proc, Oct.2001; R.Rao and B.Daneshrad, " Analysis of I/Q mismatch and a cancellation scheme for OFDMsystems, " IST Mobile Communication Summit, June 2004; A.Tarighat, R.Bagheri and A.H.Sayed, " Compensation schemes and performance analysis of IQimbalances in OFDM receivers, " Signal Processing, IEEE Transactions on [also can see Acoustics, Speech, and Signal Processing, IEEE Transactions on], vol.53, pp.3257-3268, Aug.2005.

The expansion of this work to multiple-input and multiple-output MIMO-OFDM system is shown: R.Rao and B.Daneshrad in following information, " I/Q mismatch cancellation for MIMO OFDM systems; " in Personal, Indoor and Mobile Radio Communications, 2004; PIMRC 2004.15th IEEE International Symposium on, vol.4,2004, pp.2710-2714.For spatial reuse (SM), refer to R.M.Rao, W.Zhu, S.Lang, C.Oberli, D.Browne, J.Bhatia, J.F.Frigon, J.Wang, P; Gupta, H.Lee, D.N.Liu, S.G.Wong, M.Fitz, B. Daneshrad, and O.Takeshita, " Multiantenna testbeds for research and education inwireless communications; " IEEE Communications Magazine, vol.42, no.12, pp.72-81, Dec.2004; S.Lang, M.R.Rao and B.Daneshrad, " Design anddevelopment of a 5.25GHz software defined wireless OFDM communicationplatform; " IEEE Communications Magazine, vol.42, no.6, pp.6-12, June 2004; For orthogonal space time packet (OSTBC), refer to A.Tarighat and A.H.Sayed, " MIMOOFDM receivers for systems with IQ imbalances, " IEEE Trans.Sig.Proc, vol.53, pp.3583-3596, Sep.2005.

Unfortunately, there is not the data how introduction corrects the I/Q gain in distributed input distributed output (DIDO) communication system and unbalance in phase error at present.The embodiment of the present invention of the following stated provides a kind of scheme addressed these problems.

DIDO system comprises the base station that has spaced antenna, (namely this base station is same as the Radio Resource of traditional SIO system in utilization, identical time-slot duration and frequency band), send parallel data stream (through precoding) to multiple user, strengthening downlink throughput.The application No.10/902 being entitled as " System and Method for DistributedInput-Distributed Output Wireless Communications " that S.G.Perlman and T.Cotter submitted to July 30 in 2004,978(" earlier application ") give the detailed description of DIDO system, this application is transferred the assignee giving the application, and this application is incorporated into this as a reference.

There is various ways and realize DIDO precoder.A kind of scheme is the block diagonalization (BD) described in following information: Q.H.Spencer, A.L.Swindlehurst and M.Haardt, " Zero forcingmethods for downlink spatial multiplexing in multiuser MIMO channels; " IEEETrans.Sig.Proc, vol.52, pp.461-471, Feb.2004; K.K.Wong, R.D.Murch, and K.B.Letaief, " A joint channel diagonalization for multiuser MIMO antennasystems, " IEEE Trans.Wireless Comm., vol.2, pp.773-786, JuI 2003; L.U.Choi and R.D.Murch, " A transmit preprocessing technique for multiuser MIMOsystems using a decomposition approach, " IEEE Trans.Wireless Comm., vol.3, pp.20-24, Jan 2004; Z.Shen, J.G.Andrews, R.W.Heath and B.L.Evans, " Lowcomplexity user selection algorithms for multiuser MIMO systems with blockdiagonalization; " be accepted and be published in IEEE Trans.Sig.Proc, Sep.2005; Z.Shen, R.Chen, J.G.Andrews, R.W.Heath and B.L.Evans, " Sum capacity of multiuserMIMO broadcast channels with block diagonalization, " is submitted to IEEE Trans.Wireless Comm., Oct.2005; R.Chen, R.W.Heath, and J.G.Andrews, " Transmitselection diversity for unitary precoded multiuser spatial multiplexing systemswith linear receivers; " be accepted to IEEE Trans, on Signal Processing, 2005.The method for I/Q compensation given in these materials contemplates BD precoder, and this precoder can be expanded any type to DIDO precoding.

In DIDO-OFDM system, I/Q does not mate can cause two kinds of impacts: ICI and inter-user interference.With similar in SISO-OFDM system, the former is that the interference owing to adjusting from mirror image causes.The latter is due to the fact that and causes, and namely I/Q does not mate and can destroy the orthogonal of DIDO precoder, thus produces interference between users.By method described herein, eliminate the interference of this two class at transmitter and receiver place.There is described herein three kinds of methods compensated for the I/Q in DIDO-OFDM system, and do not mate for having and not there is I/Q, compare their performance.Based on the emulation utilized performed by DIDO-OFDM prototype and actual measurement, illustrate result.

Present embodiment is the expansion of earlier application.Especially, these execution modes are relevant with the following characteristics of earlier application:

System described in earlier application, wherein I/Q track can be subject to the impact of gain and unbalance in phase;

At transmitter place, use the training signal adopted for channel estimating to calculate to have the DIDO precoder that I/Q compensates; And

Signal characteristic data take into account the distortion because I/Q imbalance causes, and at transmitter place, according to the method that this material proposes, use these signal characteristic data to calculate DIDO precoder.

B, embodiments of the present invention

First, Mathematical Modeling of the present invention and framework will be described.

Before displaying this programme, explain that core mathematics concept is very useful.We are made an explanation to it by hypothesis I/Q gain and unbalance in phase (do not comprise phase delay in this description, but this phase delay being processed automatically in the algorithm of DIDO-OFDM form).For explaining basic thought, suppose that we want two plural s=s _i+ js _qand h=h _i+ jh _qbe multiplied, and make x=h*s.We use subscript to represent inphase quadrature component.Call following equation:

x _I＝s _Ih _I-s _Qh _Q

And

x _Q＝s _Ih _Q+s _Qh _I

Its matrix form can be rewritten as:

$\left[\begin{array}{c}{x}_I\\ x_Q\end{array}\right]=\left[\begin{array}{cc}{h}_I& -h_Q\\ h_Q& h_I\end{array}\right]\left[\begin{array}{c}{s}_I\\ s_Q\end{array}\right]$

Normalization conversion is marked by channel matrix (H).Now suppose that s is sent symbol, and h is channel.Modeling is carried out by creating the existence of following non-normalized conversion to I/Q gain and unbalance in phase:

$\left[\begin{array}{c}{x}_I\\ x_Q\end{array}\right]=\left[\begin{array}{cc}{h}_{11}& h_{12}\\ h_{21}& h_{22}\end{array}\right]\left[\begin{array}{c}{s}_I\\ s_Q\end{array}\right]---\left(A\right)$

The effect of this skill confirms to be written as:

$\left[\begin{array}{cc}{h}_{11}& -h_{12}\\ h_{21}& h_{22}\end{array}\right]=\frac{1}{2}\left[\begin{array}{cc}{h}_{11}+h_{22}& h_{12}-h_{21}\\ -(h_{12}-h_{21})& h_{11}+h_{22}\end{array}\right]+\frac{1}{2}\left[\begin{array}{cc}{h}_{11}-h_{22}& h_{12}+h_{21}\\ h_{12}+h_{21}& h_{22}-h_{11}\end{array}\right]$

$=\frac{1}{2}\left[\begin{array}{cc}{h}_{11}+h_{22}& h_{12}-h_{21}\\ -(h_{12}-h_{21})& h_{11}+h_{22}\end{array}\right]+\frac{1}{2}\left[\begin{array}{cc}{h}_{11}-h_{22}& -(h_{12}+h_{21})\\ h_{12}+h_{21}& h_{11}-h_{22}\end{array}\right]\left[\begin{array}{cc}1& 0\\ 0& -1\end{array}\right]$

Now (A) is rewritten:

$\left[\begin{array}{c}{x}_I\\ x_Q\end{array}\right]=\frac{1}{2}\left[\begin{array}{cc}{h}_{11}+h_{22}& h_{12}-h_{21}\\ -(h_{12}-h_{21})& h_{11}+h_{22}\end{array}\right]\left[\begin{array}{c}{s}_I\\ s_Q\end{array}\right]+\frac{1}{2}\left[\begin{array}{cc}{h}_{11}-h_{22}& -(h_{12}+h_{21})\\ h_{12}+h_{21}& h_{11}-h_{22}\end{array}\right]\left[\begin{array}{cc}1& 0\\ 0& -1\end{array}\right]\left[\begin{array}{c}{s}_I\\ s_Q\end{array}\right]$

$=\frac{1}{2}\left[\begin{array}{cc}{h}_{11}+h_{22}& h_{12}-h_{21}\\ -(h_{12}-h_{21})& h_{11}+h_{22}\end{array}\right]\left[\begin{array}{c}{s}_I\\ s_Q\end{array}\right]+\frac{1}{2}\left[\begin{array}{cc}{h}_{11}-h_{22}& -(h_{12}-h_{21})\\ h_{12}+h_{21}& h_{11}-h_{22}\end{array}\right]\left[\begin{array}{c}{s}_I\\ -s_Q\end{array}\right]$

We carry out giving a definition:

$H_e=\frac{1}{2}\left[\begin{array}{cc}{h}_{11}+h_{22}& h_{12}-h_{21}\\ -(h_{12}-h_{21})& h_{11}+h_{22}\end{array}\right]$

And

$H_c=\frac{1}{2}\left[\begin{array}{cc}{h}_{11}-h_{22}& -(h_{12}+h_{21})\\ h_{12}+h_{21}& h_{11}-h_{22}\end{array}\right]$

These two matrixes have normalization structure, therefore can be represented as plural form:

h _e=h ₁₁+h ₂₂+j(h ₂₁-h ₁₂)

And

h _c=h ₁₁-h ₂₂+j(h ₂₁+h ₁₂)

By using all these knowledge, effective equation can be derived back and be had two channel (equivalent channels h by we _ewith conjugate channel h _c) scalar form.Therefore, the efficient transformation in (5) becomes:

x=h _es+h _cs ^*

First channel is called equivalent channels by us, and second channel is called conjugate channel.If there is no I/Q gain and unbalance in phase, then this equivalent channels is the channel that we will observe.

By using similar argument, the Input output Relationship with the discrete time MIMON × M system of I/Q gain and unbalance in phase can show for (the matrix corresponding form by using the scalar equivalent form of value to set up them):

$x\left[t\right]=\underset{l=0}{\overset{L}{\mathrm{\Σ}}}{h}_e\left[l\right]s[t-l]+h_c\left[l\right]s^{*}[t-l]$

Wherein, t is discrete time index, h _e, h _c∈ C ^{m × N}, s=[s ₁..., s _n], x=[x ₁..., x _m] and L is channel tap (channel tap) number.

In DIDO-OFDM system, illustrate the signal received in frequency domain.If meet following equation, then from signal and systems re invocation:

FFT _k{ s [t] }=S [k] then FFT _k{ s ^*[t] }=S ^*[(-k)]=S ^*[K-k] for k=0,1 ..., K-1

Utilize OFDM, the Input output Relationship of equal value for subcarrier k, MIMO-OFDM system is:

$\stackrel{\‾}{x}\left[k\right]=H_e\left[k\right]\stackrel{\‾}{s}\left[k\right]+H_c\left[k\right]{\stackrel{\‾}{s}}^{*}[K-k]---\left(1\right)$

Wherein, k=0,1 ..., K-1 is OFDM subcarrier index, H _eand H _crepresent of equal value and conjugate channel matrix respectively, be defined as follows:

$H_e\left[k\right]=\underset{l=0}{\overset{L}{\mathrm{\Σ}}}{h}_e\left[l\right]e^{-j\frac{2\mathrm{\Πk}}{K}l}$

And

$H_c\left[k\right]=\underset{l=0}{\overset{L}{\mathrm{\Σ}}}{h}_c\left[l\right]e^{-j\frac{2\mathrm{\Πk}}{K}l}$

(1) the second base value in is the interference of adjusting from mirror image.By building following stacked (stacked) matrix system (please carefully noting conjugate), it is processed:

$\left[\begin{array}{c}\stackrel{\‾}{x}\left[k\right]\\ {\stackrel{\‾}{x}}^{*}[K-k]\end{array}\right]=\left[\begin{array}{cc}{H}_e\left[k\right]& H_c\left[k\right]\\ H_c^{*}[K-k]& H_e^{*}[K-k]\end{array}\right]\left[\begin{array}{c}\stackrel{\‾}{s}\left[k\right]\\ {\stackrel{\‾}{s}}^{*}[K-k]\end{array}\right]$

Wherein with be respectively transmission and receiving symbol vector in a frequency domain.

By using the method, can active matrix be set up, operate for DIDO.Such as, utilize DIDO 2 × 2 Input output Relationship (supposing that each user has single receive antenna), first user equipment can consider following equation (when there is not noise):

$\left[\begin{array}{c}{\stackrel{\‾}{x}}_1\left[k\right]\\ {\stackrel{\‾}{x}}_1^{*}[K-k]\end{array}\right]=\left[\begin{array}{cc}{H}_e^{\left(1\right)}\left[k\right]& H_c^{\left(1\right)}\left[k\right]\\ H_c^{\left(1\right)*}[K-k]& H_e^{\left(1\right)*}[K-k]\end{array}\right]W\left[\begin{array}{c}{\stackrel{\‾}{s}}_1\left[k\right]\\ {\stackrel{\‾}{s}}_1^{*}[K-k]\\ {\stackrel{\‾}{s}}_2\left[k\right]\\ {\stackrel{\‾}{s}}_2^{*}[K-k]\end{array}\right]---\left(2\right)$

And the second user notes following equation:

$\left[\begin{array}{c}{\stackrel{\‾}{x}}_2\left[k\right]\\ {\stackrel{\‾}{x}}_2^{*}[K-k]\end{array}\right]=\left[\begin{array}{cc}{H}_e^{\left(2\right)}\left[k\right]& H_c^{\left(2\right)}\left[k\right]\\ H_c^{\left(2\right)*}[K-k]& H_e^{\left(2\right)*}[K-k]\end{array}\right]W\left[\begin{array}{c}{\stackrel{\‾}{s}}_1\left[k\right]\\ {\stackrel{\‾}{s}}_1^{*}[K-k]\\ {\stackrel{\‾}{s}}_2\left[k\right]\\ {\stackrel{\‾}{s}}_2^{*}[K-k]\end{array}\right]---\left(3\right)$

Wherein, represent matrix H respectively _eand H _cm capable, and W ∈ C ^4x4for DIDO pre-coding matrix.According to (2) and (3), the symbol that user m receives can be noticed by I/Q imbalance cause two interference sources (that is, from mirror image adjust inter-carrier interference (that is, ) and inter-user interference (that is, and p ≠ m)) impact.(3) the DIDO pre-coding matrix W in is designed to eliminate this two distracters.

There is multiple different execution mode in the DIDO precoder that can be used for herein, this depends on the joint-detection that receiver place applies.In one embodiment, can adopt according to composite channel (but not ) block diagonalization (BD) that calculates (refers to such as, Q.H.Spencer, A.L.Swindlehurst, and M.Haardt, " Zeroforcing methods for downlink spatialmultiplexing in multiuser MIMO channels, " IEEE Trans.Sig.Proc, vol.52, pp.461-471, Feb.2004.K.K; Wong, R.D.Murch, and K.B.Letaief, " A jointchannel diagonalization for multiuser MIMO antenna systems, " IEEE Trans.Wireless Comm., vol.2, pp.773-786, JuI 2003; L.U.Choi and R.D.Murch, " Atransmit preprocessing technique for multiuser MIMO systems using adecomposition approach, " IEEE Trans.Wireless Comm., vol.3, pp.20-24, Jan 2004; Z.Shen, J.G.Andrews, R.W.Heath, with B.L Evans, " Low complexityuser selection algorithms for multiuser MIMO systems with blockdiagonalization; " be accepted and be published in IEEE Trans.Sig.Proc, Sep.2005; Z.Shen, R.Chen, J.G.Andrews, R.W.Heath, and B.L Evans, " Sum capacity of multiuserMIMO broadcast channels with block diagonalization; " be submitted to IEEE Trans.Wireless Comm., Oct.2005).Therefore, current DIDO Systematic selection precoder, to make:

$H_w\stackrel{\mathrm{\Δ}}{=}\left[\begin{array}{cc}{H}_e^{\left(1\right)}\left[k\right]& H_c^{\left(1\right)}\left[k\right]\\ H_c^{\left(1\right)*}[K-k]& H_e^{\left(1\right)*}[K-k]\\ H_e^{\left(2\right)}\left[k\right]& H_c^{\left(2\right)}\left[k\right]\\ H_c^{\left(2\right)*}[K-k]& H_e^{\left(2\right)*}[K-k]\end{array}\right]W=\left[\begin{array}{cccc}{\mathrm{\α}}_{\mathrm{1,1}}& 0& 0& 0\\ 0& {\mathrm{\α}}_{\mathrm{1,2}}& 0& 0\\ 0& 0& {\mathrm{\α}}_{\mathrm{2,1}}& 0\\ 0& 0& 0& {\mathrm{\α}}_{\mathrm{2,2}}\end{array}\right]\stackrel{\mathrm{\Δ}}{=}\left[\begin{array}{cc}{H}_w^{\left(\mathrm{1,1}\right)}& H_w^{\left(\mathrm{1,2}\right)}\\ H_w^{\left(\mathrm{2,1}\right)}& H_w^{\left(\mathrm{2,2}\right)}\end{array}\right]---\left(4\right)$

Wherein, α _{i, j}for constant, and the method is highly profitable because by use this precoder, due to the impact completely eliminating I/Q gain and unbalance in phase at transmitter place can make DIDO precoder other in keep intact.

Also DIDO precoder can be designed to eliminate inter-user interference in advance, and not eliminate the ICI because IQ imbalance causes in advance.Utilize the method, receiver (but not transmitter) compensates IQ imbalance by adopting one of receiving filter of the following stated.Therefore, the Precoding Design standard in (4) can be modified to:

$H_w\stackrel{\mathrm{\Δ}}{=}\left[\begin{array}{cc}{H}_e^{\left(1\right)}\left[k\right]& H_c^{\left(1\right)}\left[k\right]\\ H_c^{\left(1\right)*}[K-k]& H_e^{\left(1\right)*}[K-k]\\ H_e^{\left(2\right)}\left[k\right]& H_c^{\left(2\right)}\left[k\right]\\ H_c^{\left(2\right)*}[K-k]& H_e^{\left(2\right)*}[K-k]\end{array}\right]W=\left[\begin{array}{cccc}{\mathrm{\α}}_{\mathrm{1,1}}& {\mathrm{\α}}_{\mathrm{1,2}}& 0& 0\\ {\mathrm{\α}}_{\mathrm{2,1}}& {\mathrm{\α}}_{\mathrm{2,2}}& 0& 0\\ 0& 0& {\mathrm{\α}}_{\mathrm{3,3}}& {\mathrm{\α}}_{\mathrm{3,4}}\\ 0& 0& {\mathrm{\α}}_{\mathrm{4,3}}& {\mathrm{\α}}_{\mathrm{4,4}}\end{array}\right]\stackrel{\mathrm{\Δ}}{=}\left[\begin{array}{cc}{H}_w^{\left(\mathrm{1,1}\right)}& H_w^{\left(\mathrm{1,2}\right)}\\ H_w^{\left(\mathrm{2,1}\right)}& H_w^{\left(\mathrm{2,2}\right)}\end{array}\right]---\left(5\right)$

${\stackrel{\‾}{x}}_1\left[k\right]=\left[\begin{array}{cc}{H}_w^{\left(\mathrm{1,1}\right)}& H_w^{\left(\mathrm{1,2}\right)}\end{array}\right]\left[\frac{{\stackrel{\‾}{s}}_1}{{\stackrel{\‾}{s}}_2}\frac{\left[k\right]}{\left[k\right]}\right]---\left(6\right)$

And

${\stackrel{\‾}{x}}_2\left[k\right]=\left[\begin{array}{cc}{H}_w^{\left(\mathrm{2,1}\right)}& H_w^{\left(\mathrm{2,2}\right)}\end{array}\right]\left[\frac{{\stackrel{\‾}{s}}_1}{{\stackrel{\‾}{s}}_2}\frac{\left[k\right]}{\left[k\right]}\right]---\left(7\right)$

Wherein for m transmission signal, and for the symbolic vector that user m receives.At receiver side, in order to transmission symbolic vector estimate, user m adopts ZF filter, and estimated symbolic vector is given as:

Although the most easy to understand of ZF filter, receiver also can apply other filters known in those skilled in the art of any amount.A kind of masses are chosen as MMSE filter, wherein:

And ρ is signal to noise ratio.Alternatively, user can perform maximum likelihood symbol detection (or Sphere decoder device, iteration change).Such as, first user can use ML receiver, and solves following optimization:

${\hat{s}}_m^{\left(\mathrm{ML}\right)}\left[k\right]=\arg \underset{s_1,s_2\∈S}{\min }||\begin{array}{c}{\stackrel{\‾}{y}}_1\left[k\right]-\left[\begin{array}{cc}{H}_w^{\left(\mathrm{1,1}\right)}& H_w^{\left(\mathrm{1,2}\right)}\end{array}\right]\left[\begin{array}{c}{s}_1\left[k\right]\\ s_2\left[k\right]\end{array}\right]\end{array}||---\left(10\right)$

Wherein, S is the set of all possible vectorial s, and depends on constellation sizes.This ML receiver provides can performance preferably, but requires higher complexity at receiver place.One group of similar equation can be applicable to the second user.

Note, in (6) and (7) with be assumed to be and there is zero.This hypothesis is only effective when transmitting precoder can eliminate the inter-user interference for the standard in (4) completely.Similar, with be only diagonal matrix when transmitting precoder can eliminate inter-carrier interference (that is, adjusting from mirror image) completely.

Figure 13 shows a kind of execution mode of the framework with the DIDO-OFDM system that I/Q compensates, and described DIDO-OFDM system comprises the IQ-DIDO precoder 1302, transmitting channel 1304, the channel estimation logic 1306 being positioned at subscriber equipment and ZF, MMSE or ML receiver 1308 that are positioned at base station (BS).Described channel estimation logic 1306 via training signal to channel with estimate, and these are estimated the precoder that feeds back in AP.BS calculates DIDO precoder weight (matrix W), to eliminate interference because I/Q gain and unbalance in phase cause and user's interference in advance, and data is sent to user by wireless channel 1304.Subscriber equipment m adopts ZF, MMSE or ML receiver 1308, and the channel estimating provided by range site 1304 eliminates residual interference, and carries out demodulation to data.

Following three execution modes can be adopted to realize this I/Q backoff algorithm.

Method 1-TX compensates: in this embodiment, transmitter calculates pre-coding matrix according to the standard in (4).At receiver place, subscriber equipment adopts " simplification " ZF receiver, wherein with be assumed to be diagonal matrix.Therefore, formula (8) is reduced to:

${\hat{s}}_m\left[k\right]=\left[\begin{array}{cc}1/{\mathrm{\α}}_{m,1}& 0\\ 0& 1/{\mathrm{\α}}_{m,2}\end{array}\right]{\stackrel{\‾}{x}}_m\left[k\right]---\left(10\right)$

Method 2-RX compensates: in this embodiment, transmitter is based on R.Chen, R.W.Heath, andJ.G.Andrews, " Transmit selection diversity for unitary precoded multiuserspatial multiplexing systems with linear receivers; " accepted to IEEE Trans, onSignal Processing, the traditional B D method described in 2005, calculate pre-coding matrix, and do not eliminate intercarrier and inter-user interference for the standard in (4).Utilize the method, the pre-coding matrix in (2) and (3) is reduced to:

$W=\left[\begin{array}{cccc}{w}_{\mathrm{1,1}}\left[k\right]& 0& w_{\mathrm{1,2}}\left[k\right]& 0\\ 0& w_{\mathrm{1,1}}^{*}[K-k]& 0& w_{\mathrm{1,2}}^{*}[K-k]\\ w_{\mathrm{2,1}}\left[k\right]& 0& w_{\mathrm{2,2}}\left[k\right]& 0\\ 0& w_{\mathrm{2,1}}^{*}[K-k]& 0& w_{\mathrm{2,2}}^{*}[K-k]\end{array}\right]---\left(12\right)$

At receiver place, subscriber equipment adopts ZF filter as in (8).Note, the method is such not as said method 1, and at transmitter, interference is eliminated in advance by place.Therefore, it eliminates inter-carrier interference at receiver place, but can not eliminate inter-user interference.In addition, feedback is required compared to method 1 with in method 2, user only needs to feed back the vector for transmitter to calculate DIDO precoder.Therefore, method 2 is particularly suitable for the DIDO system with low-rate feedback channel.On the other hand, method 2 needs subscriber equipment place to have computation complexity higher a little, in (8) (but not (11)), calculate ZF receiver.

Method 3-TX-RX compensates: in one embodiment, above-mentioned two methods is merged.Transmitter calculates pre-coding matrix as (4), and receiver is estimated transmission symbol according to (8).

I/Q imbalance (no matter being unbalance in phase, gain imbalance, or be postpone imbalance) can cause harmful degradation to the signal quality in wireless communication system.For this reason, circuit in the past is all designed to have lower imbalance.But, as mentioned above, by using the Digital Signal Processing and/or specific receiver of launching precoded form, revise this problem.One embodiment of the present invention comprise the system with multiple New function unit, and it is all very important that each unit corrects for the I/Q realized in ofdm communication system or DIDO-OFDM communication system.

One embodiment of the present invention use the precoding based on channel condition information, to eliminate in ofdm system from the inter-carrier interference (ICI) (causing because I/Q does not mate) that mirror image is adjusted.As shown in figure 11, DIDO transmitter according to the present embodiment comprises subscriber selector unit 1102, multiple coded modulation unit 1104, corresponding multiple map unit 1106, the known precoding unit of DIDO IQ 1108, multiple RF transmitter unit 1114, user feedback unit 1112 and DIDO configurator unit 1110.

The feedback information that described subscriber selector unit 1102 obtains based on feedback unit 1112, selects and multiple user U ₁-U _mthe data be associated, and this information is supplied to each coded modulation unit 1104 in multiple coded modulation unit 1104.The information bit of each coded modulation unit 1104 to each user is encoded and demodulation, and they are sent to map unit 1106.Input bit is mapped to complex symbol by this map unit 1106, and result is sent to the known precoding unit 1108 of DIDO IQ.The channel condition information that the known precoding unit 1108 of this DIDO IQ utilizes feedback unit 1112 to obtain from user, calculates the known precoding weight of DIDO IQ, and carries out precoding to the incoming symbol obtained from map unit 1106.Each pre-code data stream is sent to OFDM unit 1115 by the known precoding unit 1108 of DIDO IQ, and this OFDM unit 1115 calculates IFFT, and adds Cyclic Prefix.This information is sent to D/A unit 1116, and this D/A unit 1116 carries out digital-to-analogue conversion, and sends it to RF unit 1114.Baseband signal raising frequency to intermediate frequency/radio frequency, and is sent it to transmitting antenna by this RF unit 1114.

Described precoder is adjusted routine mediation mirror image and is operated together, to compensate I/Q imbalance.The precoder design standard of any amount can be used, comprise ZF, MMSE or weighting MMSE designs.In a preferred embodiment, precoder can remove completely because I/Q does not mate caused ICI, thus makes receiver not need to perform any ancillary relief.

In one embodiment, described precoder uses block diagonalization standard, with when not exclusively eliminating I/Q impact (this needs accessory receiver process) of each user, eliminates inter-user interference completely.In another embodiment, described precoder uses zero pressure standard to eliminate completely because of the I/Q inter-user interference that causes of imbalance and ICI interference.This execution mode can use traditional DIDO-OFDM processor at receiver place.

One embodiment of the present invention use the precoding based on channel condition information, and to eliminate from the inter-carrier interference (ICI) (caused because I/Q does not mate) that mirror image is adjusted in DIDO-OFDM system, and each user adopts the known DIDO receiver of IQ.As shown in figure 12, in one embodiment of the invention, system (comprising receiver 1202) comprises multiple RF unit 1208, correspondingly multiple A/D unit 1210, IQ known channel estimator 1204 and DIDO feedback generator unit 1206.

Described RF unit 1208 receives the signal sent from DIDO transmitter unit 1114, by this signal down to base band, and the signal after this frequency reducing is supplied to A/D unit 1210.Afterwards, this A/D unit 1210 carries out analog-to-digital conversion to this signal, and sends it to OFDM unit 1213.This OFDM unit 1213 removes Cyclic Prefix, and carries out FFT, to report this signal to frequency domain.During cycle of training, output is sent to IQ known channel estimation unit 1204 by OFDM unit 1213, and this IQ known channel estimation unit 1204 calculates channel estimating in a frequency domain.Alternatively, described channel estimating can be calculated in the time domain.In period in data cycle (data period), output is sent to the known receiver unit 1202 of IQ by OFDM unit 1213.The known receiver unit of this IQ calculates IQ receiver, and carries out demodulate/decode to described signal, to obtain data 1214.Described IQ known channel estimation unit 1204 sends described channel estimating to DIDO feedback generator unit 1206, and this feedback generator unit 1204 can quantize described channel estimating, and is beamed back transmitter via feedback control channel 1112.

Receiver 1202 shown in Figure 12 can work under the standard known in those skilled in the art of any amount (comprising ZF, MMSE, maximum likelihood or MAP receiver).In a preferred embodiment, receiver uses MMSE filter to eliminate because of the uneven ICI caused of the IQ on mirror image tune.In another preferred embodiment, the receiver symbol that uses the nonlinear detector being similar to maximum likelihood searching to carry out joint-detection mirror image to adjust.The method has good performance, but has higher complexity.

In one embodiment, IQ known channel estimator 1204 is used to determine receiver coefficient, to remove ICI.Therefore, we require that the rights and interests of DIDO-OFDM system (using the precoding based on channel condition information to eliminate the inter-carrier interference (ICI) (caused because I/Q does not mate) adjusted from mirror image), the known DIDO receiver of IQ and IQ known channel estimator.Described channel estimator can use traditional training signal, maybe can be used in the training signal of the special structure that inphase quadrature signal sends.The algorithm for estimating of any amount can be implemented, comprise least square method, MMSE or maximum likelihood.Described IQ known channel estimator provides input for the known receiver of IQ.

Channel condition information is provided to website by channel reciprocity or by feedback channel.One embodiment of the present invention comprises DIDO-OFDM system, and this system has the known precoder of I/Q, and for future user terminal channel condition information transfer to the known feedback channel of I/Q of website.This feedback channel can be physics or logical control channel.It can by special or shared in RACH.By using the DIDO feedback generator at user terminal (we also require that the rights and interests of this user terminal) place to generate feedback information.Described DIDO feedback generator using the output of described I/Q known channel estimator as input.It can quantized channel coefficient, maybe can use any amount Limited Feedback algorithm known in the field.

The distribution of user, modulation and encoding rate, can to change according to the result of described DIDO feedback generator to the mapping of space-time frequency coding time slot.Therefore, one execution mode comprises the known DIDO configurator of IQ, this configurator uses the IQ known channel from one or more user to estimate to configure the known precoder of DIDO IQ, the subset of the user that selection modulation rate, encoding rate, permission send and their mapping to space-time frequency coding time slot.

In order to evaluate the performance of proposed compensation method, three DIDO 2 × 2 systems will be compared:

1, there is I/Q not mate: sent by all tune (adjusting except DC is in harmonious proportion edge), and I/Q is not mated and compensate;

2, there is I/Q compensate: sent by all row of transferring in, and compensate by using above-mentioned " method 1 " not mate I/Q;

3, desirable: to transfer in row by means of only odd number and send, to avoid inter-user interference and not mate caused intercarrier (that is, adjusting from mirror image) interference because of I/Q.

After this, illustrate in true propagation situation and utilize DIDO-OFDM prototype to carry out measuring obtained result.Figure 14 depicts the 64-QAM planisphere obtained from above-mentioned three systems.These planispheres obtain when same customer location and fixing average signal-to-noise ratio (~ 45dB).First planisphere 1401 is very noisy (because the uneven institute of I/Q causes the interference from mirror image tune).Second planisphere 1402 shows some and improves (because I/Q compensates).Note, the second planisphere 1402 does not have the ideal situation shown in planisphere 1403 pure like that (owing to there is the phase noise that may produce inter-carrier interference (ICI)).

Figure 15 shows under having and not having the unmatched situation of I/Q, the average SER(symbol error rate of DIDO 2 × 2 system of 64-QAM and 3/4 encoding rate) 1501 and every user's goodput (goodput) 1502.OFDM bandwidth is 250KHZ, has 64 and adjusts and circulating prefix-length L _cp=4.Due in the ideal case, we send data by means of only the subset adjusted, and therefore evaluate SER and goodput performance, to ensure the fair comparison between different situations according to the transmitting power (but not total transmitting power) on average often adjusted.In addition, in following result, we use the normalized value of transmitting power (indicating with decibel), because herein the target of ours compare relative (but not absolute) performance of different schemes.Figure 15 shows to be existed in the unbalanced situation of I/Q, saturated and miss the mark SER(~ 10 of SER ^-2), this and A.Tarighat and A.H.Sayed, " MIMO OFDM receiversfor systems with IQ imbalances, " IEEE Trans.Sig.Proc, the result reported in vol.53, pp.3583-3596, Sep.2005 is consistent.This saturation effect is due to the fact that and causes, and namely signal power and interference power (adjusting from mirror image) increase along with the increase of TX power.But, pass through proposed I/Q compensation method, can interference be eliminated, and obtain SER performance preferably.Note, because 64-QAM modulation needs larger transmitting power, therefore, SER can be caused to have trickle increase at high SNR place because of the amplitude saturation effect in DAC.

In addition, can be observed, when there is I/Q and compensating, SER performance closely ideal situation.Between these two kinds of situations, the 2dB gap of TX power is caused due to phase noise (this phase noise may produce additional interference between adjacent OFDM is adjusted).Finally, goodput curve 1502 shows when applying I/Q method, and it can send the data of twice compared to ideal situation, to adjust but not only odd number adjusts (for ideal situation) because we used all data.

Figure 16 illustrates when having I/Q and compensating or do not have I/Q compensation, the SER performance of different Q AM planisphere.We can be observed, and in this embodiment, the method proposed is particularly advantageous for 64-QAM planisphere.For 4-QAM and 16-QAM, I/Q compensation method can produce than having the worse performance of the unmatched situation of I/Q, and this may be because the method proposed requires that larger power is to carry out data transmission and to eliminate from the interference that mirror image is adjusted.In addition, due to the larger minimum range between constellation point, 4-QAM and 16-QAM is also not as 64-QAM is subject to the unmatched impact of I/Q like that.See A.Tarighat, R.Bagheri, and A.H.Sayed, " Compensation schemes and performance analysis of IQ imbalances in OFDMreceivers; " Signal Processing, IEEE Transactions on [also can see Acoustics, Speech, and Signal Processing, IEEE Transactions on], vol.53, pp.3257-3268, Aug.2005.Also observable Figure 16 also compares with the ideal situation for 4-QAM and 16-QAM by not mated by I/Q and draws this conclusion.Therefore, for the situation of 4-QAM and 16-QAM, the slight interests that the secondary power required for DIDO precoder with interference elimination (adjusting from mirror image) can not compensate for I/Q be gone bail for.Noting, solving this problem by adopting above-mentioned I/Q compensation method 2 and 3.

Finally, under different propagation conditions, measure the relative SER performance of above-mentioned three methods.Also describe in the SER performance that there is the unmatched situation of I/Q, for reference.Figure 17 depicts and is 450.5MHZ for carrier frequency and bandwidth is 64-QAM DIDO 2 × 2 system of 250KHz, at the SER measured by two different customer locations.In position 1, user be in not chummery and be in NLOS(ignore distance) BS of state is at a distance of ~ 6 λ.In position 2, user with there is LOS(sighting distance) BS at a distance of ~ λ.

Figure 17 shows all three kinds of compensation methodes and all has outstanding performance than situation about not compensating.But it should be noted, under any channel conditions, method 3 all surpasses other two kinds of compensation methodes.The relative performance of method 1 and 2 depends on propagation condition.By actual measurement activity, can show that method 1 surpasses method 2 substantially, because it eliminates the inter-user interference that (at transmitter place) I/O imbalance causes in advance.When this inter-user interference is very little, as shown in the curve chart 1702 of Figure 17, method 2 can surpass method 1, because it can not suffer the power loss because I/Q compensation precoder causes.

Up to the present, by only considering limited group of propagation situation (as shown in figure 17), distinct methods is compared.After this, at the independent and tool of desirable i.i.d.(with distribution) measure the relative performance of these methods in channel.The I/Q phase place of transmitting and receiving side and gain imbalance is utilized to emulate DIDO-OFDM system.That is, Figure 18 shows (when only launching pusher side and having gain balance the I rail of the first transmitting chain has gain 0.8, other rails has gain 1), the performance of the method proposed.Can find out, method 3 has surpassed every other method.In addition, compared with obtaining result with position 2 place in the curve chart 1702 of Figure 17, in i.i.d. channel, method 1 comparable method 2 performs better.

Therefore, give three kinds of novel methods uneven to the I/Q compensated in above-mentioned DIDO-OFDM system, method 3 surpasses other proposed compensation methodes.In the system with low-rate feedback channel, using method 2 can reduce needed for DIDO precoding feedback quantity, but poor SER performance can be caused.

II, self adaptation DIDO delivery plan

Use description to another execution mode of the system and method for the performance strengthening distributed input distributed output (DIDO) system.Allocation of radio resources, by following the tracks of the channel status of change, is dynamically given different subscriber equipmenies by the method, to increase throughput while meeting some target error rate.Described subscriber equipment is estimated channel quality, and is fed back to base station (BS); This base station processes the channel quality being obtained from subscriber equipment, to select optimal user cluster tool, DIDO scheme, modulation/coding scheme (MCS) and the array configurations for sending next time; Parallel data is sent to multiple subscriber equipment via precoding by described base station, and signal is demodulated at receiver place.

The system that one is the effective Resources allocation of DIDO wireless link is also described.This system comprises the DIDO base station with DIDO configurator, and this base station processes the feedback receiving personal family, to select optimal user set for sending next time, DIDO scheme, modulation/coding scheme (MCS) and array configurations; Receiver in DIDO system, this receiver is measured channel and other relevant parameters, to generate DIDO feedback signal; And DIDO feedback control channel, for by the transmission of feedback information from user to base station.

As detailed in the following, some notable features of this execution mode of the present invention can include, but are not limited to:

For based on channel quality information, select number of users, DIDO delivery plan adaptively (namely, it line options or multiplexing), modulation/coding scheme (MCS) and array configurations, to minimize SER, or maximize the spectrum efficiency of every user or the technology of downlink tone spectrum efficiency;

For defining many group DIDO sending modes using the technology of the combination as DIDO scheme and MCS;

For giving different time slots, the OFDM technology being in harmonious proportion DIDO subflow according to channel status by different DIDO mode assignments;

For the channel quality based on different user, different DIDO pattern is dynamically assigned to the technology of different user;

For switching the standard activated to self adaptation DIDO based on the link quality metric calculated in time domain, frequency domain and spatial domain;

For switching the standard activated to self adaptation DIDO based on look-up table.

The DIDO system in base station with DIDO configurator as shown in figure 19, this system can based on channel quality information, select number of users, DIDO delivery plan adaptively (namely, it line options or multiplexing), modulation/coding scheme (MCS) and array configurations, to minimize SER, or maximize spectrum efficiency or the downlink tone spectrum efficiency of every user;

In base station, there is DIDO configurator and there is at each subscriber equipment place the DIDO system of DIDO feedback generator as shown in figure 20, channel conditions estimated by this system uses and/or other parameters (being similar to estimated SNR) at receiver place, to generate the feedback message inputing to DIDO configurator.

DIDO system, this system has DIDO configurator (in base station), DIDO feedback generator and DIDO feedback control channel (this DIDO feedback channel is used for DIDO specific configuration information to transfer to base station from user).

A, background

In multiple-input and multiple-output (MIMO) system, it is conceivable that diversity scheme (such as, orthogonal space time packet (OSTBC) is (see V.Tarokh, H.Jafarkhani, and A.R.Calderbank, " Spacetime block codes from orthogonal designs, " IEEE Trans.Info.Th., vol.45, pp.1456-467, JuI.1999) or sky line options (see R.W.Heath Jr., S.Sandhu, andA.J.Paulraj, " Antenna selection for spatial multiplexing systems with linear receivers, " IEEE Trans.Comm., vol.5, pp.142-144, Apr.2001), to prevent fading channel, improve link reliability (this reliability can be exchanged into better coverage rate).On the other hand, spatial reuse (SM) can send using multiple parallel data and strengthen throughput of system as means.See G.J.Foschini, G.D.Golden, R.A.Valenzuela, and P.W.Wolniansky, " Simplifiedprocessing for high spectral effciency wireless communication employingmultielement arrays, " IEEE Jour.Select.Areas in Comm., vol.17, no.1 1, pp.1841-1852, Nov.1999.According to deriving from L.Zheng and D.N.C.Tse, " Diversity andmultiplexing:a fundamental tradeoff in multiple antenna channels; " IEEE Trans.Info.Th., vol.49, no.5, theoretical diversity/multiplexing the compromise of pp.1073-1096, May 2003, these benefits can realize in mimo systems simultaneously.One actual form of implementation is the channel status by following the tracks of change, between diversity and multiplexing delivery plan, carry out self adaptation switching.

Now propose a large amount of adaptive MIMO transmission technology.R.W.Heath and A.J.Paulraj, " Switching between diversity and multiplexing in MIMO systems; " IEEETrans.Comm., vol.53, no.6, diversity in pp.962-968, Jun.2005/multiplexing changing method is designed to based on momentary channel quality information, improves the BER(bit error rate sent for fixed rate).Alternatively, can as S.Catreux, V.Erceg, D.Gesbert, and R.W.Heath.Jr., " Adaptive modulation and MIMO coding for broadband wireless datanetworks, " IEEE Comm.Mag., vol.2, pp.108-1 15, in June 2002 (" Catreux "), adopt statistic channel information to activate self adaptation, thus reduce the quantity of feedback overhead and control message.Self adaptation transmission algorithm in Catreux be designed to based on channel time/frequency selects designator, for the predeterminated target error rate in OFDM (OFDM) system, strengthens spectrum efficiency.Also for narrowband systems, propose similar low feedback adaptive method, the method utilizes channel space selectivity to switch between diversity scheme and spatial reuse.See such as A.Forenza, M.R.McKay, A.Pandharipande, R.W.Heath.Jr., and I.B.Collings, " Adaptive MIMOtransmission for exploiting the capacity of spatially correlated channels, " accepted to the IEEE Trans, on Veh.Tech., Mar.2007; M.R.McKay, I.B.Collings, A.Forenza, and R.W.Heath.Jr., " Multiplexing/beamformingswitching for coded MIMO in spatially correlated Rayleigh channels; " be accepted to IEEE Trans, on Veh.Tech., Dec.2007; A.Forenza, M.R.McKay, R.W.Heath.Jr., and I.B.Collings, " Switching between OSTBC and spatial multiplexing withlinear receivers in spatially correlated MIMO channels, " Proc.IEEE Veh.Technol.Conf., vol.3, pp.1387-1391, May 2006; M.R.McKay, I.B.Collings, A.Forenza, with R.W.Heath Jr., " A throughput-based adaptive MIMO BICMapproach for spatially correlated channels, " appears at Proc.IEEE ICC, June 2006.

In this data, we by various previously open in the working range that represents extend to DIDO-OFDM system.See such as R.W.Heath and A.J.Paulraj, " Switching betweendiversity and multiplexing in MIMO systems, " IEEE Trans.Comm., vol.53, no.6, pp.962-968, Jun.2005; S.Catreux, V.Erceg, D.Gesbert, with R.W.Heath Jr., " Adaptive modulation and MIMO coding for broadband wireless datanetworks, " IEEE Comm.Mag., vol.2, pp.108-1 15, June 2002; A.Forenza, M.R.McKay, A.Pandharipande, R.W.Heath Jr., and I.B.Collings, " AdaptiveMIMO transmission for exploiting the capacity of spatially correlated channels, " IEEE Trans, on Veh.Tech., vol.56, n.2, pp.619-630, Mar.2007; M.R.McKay, I.B.Collings, A.Forenza, with R.W.Heath Jr., " Multiplexing/beamformingswitching for coded MIMO in spatially correlated Rayleigh channels; " be accepted to IEEE Trans, on Veh.Tech., Dec.2007; A.Forenza, M.R.McKay, R.W.HeathJr., and I.B.Collings, " Switching between OSTBC and spatial multiplexing withlinear receivers in spatially correlated MIMO channels, " Proc.IEEE Veh.Technol.Conf., vol.3, pp.1387-1391, May 2006; M.R.McKay, I.B.Collings, A.Forenza, with R.W.Heath Jr., " A throughput-based adaptive MIMO BICM approach for spatially correlated channels, " appears at Proc.IEEE ICC, June 2006.

There is described herein NEW ADAPTIVE DIDO sending strategy, this strategy carrys out improved system performance to carry out switching between the user of varying number, the transmitting antenna of varying number and delivery plan based on channel quality information as a kind of means.Note, M.Sharif and B.Hassibi, " On the capacity ofMIMO broadcast channel with partial side information; " IEEE Trans.Info.Th., vol.51, p.506522, Feb.2005 and W.Choi, A.Forenza, J.G.Andrews, and R.W.Heath Jr., " Opportunistic space division multiple access with beam selection; " appear at IEEE Trans, on Communications, propose the scheme of adaptively selected user in multi-user MIMO system.But, opportunistic (opportunistic) space division multiplexing access (OSDMA) scheme during these are open is designed to by utilizing multi-user diversity to maximize total capacity, and they only can realize the part of theory capacity of dirty paper (dirty paper) code, because do not eliminate interference completely in advance at transmitter place.In DIDO transmission algorithm described herein, block diagonalization is adopted to eliminate inter-user interference in advance.But, propose self adaptation sending strategy and can be applied to any DIDO system, without the need to considering the type of precoding technique.

This patent application describes the expansion of the execution mode of the invention described above and earlier application, include but not limited to following supplementary features:

1, can be adopted in earlier application by wireless client device and for the training symbol of channel estimating, the link quality metric in self adaptation DIDO scheme is evaluated.

2, as described in earlier application, base station receives the signal characteristic data from client device.In the present embodiment, signal characteristic data are defined for the link quality metric activated self adaptation.

3, prior application describe one for selecting the mechanism of antenna and number of users, and define throughput distribution.In addition, as earlier application, the throughput of different stage dynamically can be assigned to different clients.Current embodiment of the present invention defines the novel standard relevant to this selection and throughput distribution.

B, embodiments of the present invention

The target of the self adaptation DIDO technology proposed is spectrum efficiency or the downlink tone spectrum efficiency that different user by being dynamically assigned in system by the Radio Resource in time, frequency and space strengthens every user.This overall self adaptation standard is used for while meeting target error rate, improves throughput.According to spread state, this adaptive algorithm also can be used via diversity scheme to improve the link-quality (or coverage rate) of user.The flow chart of Figure 21 display describes the step of self adaptation DIDO scheme.

2102, base station (BS) collects the channel condition information from all users.2104, according to received CSI, base station calculates link quality metric in time domain/frequency domain/spatial domain.2106, use these link quality metric to select by serviced user in next transmission, and for the sending mode of each user.Note, sending mode comprises the various combination of modulation/coding and DIDO scheme.Finally, user is sent data at 2108, BS via DIDO precoding.

2102, the base station selected channel condition information from all subscriber equipmenies (CSI).2104, base station uses this CSI to determine the instantaneous of all subscriber equipmenies or statistical channel quality.In DIDO-OFDM system, can estimate channel quality (or link quality metric) in time domain, frequency domain and spatial domain.Afterwards, 2106, base station uses link quality metric to determine optimal user subset and the sending mode for current propagation state.The set of DIDO sending mode is combined into the combination of DIDO scheme (that is, sky line options or multiplexing), modulation/coding scheme (MCS) and array configurations.2108, by using selected user quantity and sending mode, send data to subscriber equipment.

Model selection is carried out by look-up table (LUT) (this look-up table is pre-calculated based on the bit error rate performance in the different communication environments of DIDO system).Channel quality information is mapped to bit error rate performance by these LUT.In order to build LUT, DIDO system can be evaluated according to SNR and propagating bit error rate performance in situation in difference.Can find out from ber curve, the minimum SNR realized needed for a certain predeterminated target error rate can be calculated.This SNR requirement definition is SNR threshold value by we.Afterwards, evaluate SNR threshold value in different propagation situations and for different DIDO sending modes, and be stored in LUT.Such as, SER result can be used in Figure 24 and Figure 26 to build LUT.Afterwards, according to this LUT, the sending mode for active user can be selected in base station, and this pattern can improve throughput while meeting the predeterminated target error rate.Finally, base station sends data to selected user via DIDO precoding.Note, can adjust and DIDO subflow, to make to carry out self adaptation in time domain, frequency domain and spatial domain to different time slots, OFDM different DIDO mode assignments.

Figure 19-Figure 20 shows a kind of execution mode adopting the adaptive system of DIDO.Introduce some new functional units and implement proposed DIDO adaptive algorithm.Specifically, in one embodiment, the channel quality information 1912 that DIDO configurator 1910 can provide based on subscriber equipment, performs several functions, comprise and select number of users, DIDO delivery plan (that is, sky line options and multiplexing), modulation/coding scheme (MCS) and array configurations.

Subscriber selector unit 1902, based on the feedback information obtained by DIDO configurator 1910, is selected and multiple user U ₁-U _mthe data be associated, and this information is provided each coded modulation unit in every multiple coded modulation unit 1904.The information bit of each coded modulation unit 1904 to each user is encoded and modulates, and they are sent to map unit 1906.Input bit is mapped to complex symbol by this map unit 1906, and sends it to precoding unit 1908.Coded modulation unit 1904 and map unit 1906 all utilize the information being obtained from DIDO configurator unit 1910, are chosen as the modulation/coding scheme type that each user adopts.Described information can be calculated by the channel quality information of configurator unit 1910 by each user utilizing feedback unit 1912 and provide.DIDO precoding unit 1908 utilizes the information obtained by DIDO configurator unit 1910 to calculate DIDO precoding weight, and carries out precoding to the incoming symbol being obtained from map unit 1906.By DIDO precoding unit 1906 by the data flow after each precoding to OFDM unit 1915, this OFDM unit 1915 calculates IFFT and adds Cyclic Prefix.This information is sent to D/A unit 1916, and this D/A unit 1916 carries out digital-to-analogue conversion, and final analog signal is sent to RF unit 1914.Baseband signal raising frequency to intermediate frequency/radio frequency, and is sent it to transmitting antenna by this RF unit 1914.

The RF unit 2008 of each client device receives the signal sent from DIDO transmitter unit 1914, by this signal down to base band, and the signal after frequency reducing is supplied to A/D unit 2010.Afterwards, this signal from analog is converted to numeral by this A/D unit 2010, and sends it to OFDM unit 2013.This OFDM unit 2013 removes Cyclic Prefix, and performs FFT, to report signal to frequency domain.In cycle of training, output is sent to channel estimating unit 2004 by OFDM unit 2013, and this channel estimating unit 2004 calculates channel estimating in a frequency domain.Alternatively, can in time-domain calculation channel estimating.During the data cycle, output is sent to receiver unit 2002 by OFDM unit 2013, and this receiver unit 2002 pairs of signals carry out demodulate/decode, to obtain data 2014.Channel estimating is sent to DIDO feedback generator unit 2006 by described channel estimating unit 2004, and this DIDO feedback generator unit 2006 can quantize channel estimating, and is beamed back transmitter via feedback control channel 1912.

Described DIDO configurator 1910 can be used in the information that base station obtains, or in a preferred embodiment, and the extra DIDO feedback generator 2006(working in each subscriber equipment place that uses is see Figure 20) output.This DIDO feedback generator 2006 uses other parameters being similar to estimated SNR at estimated channel conditions 2004 and/or receiver place to generate feedback message by being input to DIDO configurator 1910.Described DIDO feedback generator 2006 can compress, quantize and/or use at receiver place Limited Feedback strategies more known in the field to information.

Described DIDO configurator 1910 can use the information recovered from DIDO feedback control channel 1912.DIDO feedback control channel is logic OR physical control channel, and this channel can be used for the output of DIDO feedback generator 2006 to be sent to base station from user.Control channel 1912 can be implemented in the mode known in the field of any amount, and can be logic OR physical control channel.As physical channel, it can comprise the dedicated time slot/frequency gap being assigned to user.It also can for the RACH shared by all users.Described control channel can be pre-assigned, or can be created by the bit (stealing bits) of occupying of predetermined way in existing control channel.

In the following discussion, DIDO-OFDM prototype is utilized to carry out measuring obtained result by being described through in true propagation environment.These results show that the realizability of potential gain in self adaptation DIDO system.First represent the performance of different stage DIDO system, show to increase antenna/user quantity, to realize larger downlink throughput.Afterwards, describe the DIDO performance relevant with the position of subscriber equipment, show that the channel status of change followed the tracks of by needs.Finally, the performance of the DIDO system adopting diversity technique is described.

The performance of I, different stage DIDO system

Increasing transmitting antenna (N=M, wherein M is number of users) is utilized to evaluate the performance of different DIDO system.The performance of following system is compared: SISO, DIDO 2 × 2, DIDO 4 × 4, DIDO 6 × 6 and DIDO 8 × 8.DIDO N × M refers to the DIDO at BS place with N number of transmitting antenna and M user.

Figure 22 shows transmit/receive antenna layout.Arrange transmitting antenna 2201 with square array configuration, and user is positioned at around emission array." transmitting " antenna is referred to, U refers to " subscriber equipment " 2202 at Figure 22, T.

Different antennae subset in 8 yuan of emission arrays is in active state, and this depends on for the N value selected by different measuring.For each DIDO rank (N), select can to the fixed size of 8 element array retrain for the antenna subset that covers of maximum true area.This standard is supposed to strengthen the space diversity of given N value.

Figure 23 shows the array configurations for the different DIDO ranks being applicable to available true area (that is, dotted line).Square empty frame has 24 " × 24 " size, corresponding to 450MHz carrier frequency ~ λ × λ.

Based on the commentary relevant to Figure 23 and with reference to Figure 22, existing will definition the performance of each system in following system:

There is the SISO(2301 of T1 and U1)

There is T1,2 and U1, the DIDO 2 × 2(2302 of 2)

There is T1,2,3,4 and U1, the DIDO 4 × 4(2303 of 2,3,4)

There is T1,2,3,4,5,6 and U1,2,3,4,5, the DIDO 6 × 6(2304 of 6)

There is T1,2,3,4,5,6,7,8 and U1,2,3,4,5, the DIDO 8 × 8(2305 of 6,7,8)

Figure 24 shows in 4-QAM and 1/2FEC(forward error correction) in rate situation, SER, BER, SE(spectrum efficiency in above-mentioned DIDO system) and the functional relation of goodput performance and transmitting (TX) power.Observation draws, SER and BER performance can decline because N value increases.This impact is caused by following two phenomenons: for fixing TX power, the input power of DIDO array can be divided between increasing user (or data flow); Space diversity can reduce along with the increase of the number of users in actual DIDO channel.

As shown in figure 24, in order to compare the relative performance of different stage DIDO system, target BER is fixed as 10 ^-4(this value can change according to system), this value roughly corresponds to SER=10 ^-2.The TX performance number corresponding to this target is referred to as TX power threshold (TPT) by us.For any N, if TX power is lower than TPT, we suppose to send under DIDO level n, and we need to switch to other DIDO of even lower level.In addition, at Figure 24, observable draws, when TX power exceedes the TPT for any N value, SE and goodput performance can reach capacity.According to these results, self adaptation sending strategy can be designed to switch between different stage DIDO, to strengthen SE for the fixing predeterminated target error rate or goodput.

Performance under II, user variable situation

The target of this test is, emulates via in space correlation channel, evaluates the DIDO performance of different user position.DIDO 2 × 2 system is regarded as having 4QAM and 1/2FEC and leads.As shown in figure 25, user 1 is positioned at the side of emission array and penetrates (broadside) direction, and the position of user 2 is penetrated direction and become end-fire (endfire) direction from side.-λ/2, transmitting antenna interval, and to be separated by-2.5 λ with user.

Figure 26 shows the diverse location for subscriber equipment 2, the SE result of SER and every user.Penetrate orientation measurement from the limit of emission array, the angle of arrival (AOA) of subscriber equipment is 0 ° to 90 °.Observation draws, along with the angular distance of subscriber equipment increases, DIDO performance will promote, because DIDO channel memory is in larger diversity.In addition, at target SER=10 ^-2, between AOA2=0 ° and AOA2=90 ° of both of these case, there is the gap of 10dB in place.This result with in Figure 35 for angle spread 10 ° to obtain simulation result consistent.In addition, note, for the situation of AOA1=AOA2=0 °, may there is coupling effect (antenna because of them is adjoining to be caused) between two users, this may make their performance different from the simulation result in Figure 35.

III, preferred situation for DIDO 8 × 8

Figure 24 shows DIDO 8 × 8 and produces the SE larger than even lower level DIDO, but has higher TX power demand.The target of this analysis is to illustrate to there is this situation, and namely DIDO 8 × 8 is not only in peaks spectrum efficiency (SE), but also in TX power demand (or TPT), surpasses DIDO2 × 2, to realize described peak value SE.

Note, desirable at i.i.d.() in channel, TX power exists between DIDO 8 × 8 and the SE of DIDO 2 × 2 ~ gap of 6dB.This gap causes because of this fact, and namely TX power is split by DIDO 8 × 8 between 8 data flow, and DIDO 2 × 2 is only split between two stream.This result is illustrated via the emulation in Figure 32.

But in space correlation channel, TPT is the function of communication environments characteristic (such as, array towards, customer location, angle spread).Such as, Figure 35 to show for two different user devices positions between low angle expansion ~ 15dB gap.Similar result is illustrated in the application Figure 26.

Be similar to mimo system, when user is positioned at the end-on direction of TX array, the performance of DIDO system can decline (caused because lacking diversity).This impact is observed by utilizing current DIDO prototype to carry out measuring and is drawn.Therefore, one illustrates that the mode that DIDO 8 × 8 surpasses DIDO 2 × 2 is end-on direction user be placed in relative to DIDO 2 × 2 array.In this situation, DIDO 8 × 8 has surpassed DIDO 2 × 2, because 8-aerial array provides higher diversity.

In this analysis, following system is considered:

The every time slot of DIDO 8 × 8(of system 1:4-QAM sends 8 parallel data streams);

Every 4 time slots of DIDO 2 × 2(of system 2:64-QAM, once send transmission user X and Y).For this system, we consider four kinds of combinations of TX and RX aerial position: a) T1, T2U1,2(end-on direction); B) T3, T4U3,4(end-on direction); C) T5, T6U5,6(and end-on direction be separated by ~ 30 °); D) T7, T8U7,8(NLOS(ignore distance));

The DIDO 8 × 8 of system 3:64-QAM; And

Every 8 time slots of MISO 8 × 1(of system 4:64-QAM, once send transmission user X).

For all these situations, the FEC of 3/4 is used to lead.

Figure 27 depicts the position of user.

In Figure 28, SER result shows the gap (similar to the simulation result in Figure 35) of a ~ 15dB between system 2a and 2c due to different array directions and customer location.The first subgraph in a second row shows the value (that is, corresponding to BER 1e-4) of the saturated TX power of SE curve.We observe system 1 creates larger each user for lower TX power demand (being less than ~ 5dB) SE than system 2.And, due to the spatial multiplexing gain of DIDO 8 × 8 on DIDO 2 × 2, DIDO8 × 8 compared to the benefit of DIDO 2 × 2 for DL(down link) more obvious SE and DL goodput.Due to the array gain (that is, having the MRC of MISO 8 × 1) of beam forming, system 4 has lower TX power demand (being less than 8dB) than system 1.But system 4 only creates 1/3 of the SE of each user compared to system 1.System 2 is than the poor performance (that is, creating lower SE for larger TX power demand) of system 1.Finally, system 3 creates much bigger SE(due to the modulation of larger exponent number (larger order) than system 1 for larger TX power demand (~ 15dB)).

According to these results, can infer to draw a conclusion:

A kind of channel conditions is confirmed to be DIDO 8 × 8 to be surpassed DIDO 2 × 2(and namely creates larger SE for lower TX power demand);

In this channel conditions, DIDO 8 × 8 creates SE and the DL SE of larger each user than DIDO 2 × 2 and MISO 8 × 1; And

Can pass through with larger TX power demand (being greater than ~ 15dB) as cost uses high order modulation (i.e. 64-QAM, instead of 4-QAM) to increase the performance of DIDO 8 × 8 further.

Iv. there is the DIDO of day line options

Below, the benefit of our assessments Antenna Selection Algorithem of description in " the Transmit selection diversityfor unitary precoded multiuser spatial multiplexing systems with linear receivers " that delivered by R.Chen, R.W.Heath and J.G.Andrews in 2005 on Signal Processing that received by IEEE journal.We lead the result presented for a specific DIDO system with the FEC of two users, 4-QAM and 1/2.Following system is compared in figure 27:

There is T1,2 and U1, the DIDO 2 × 2 of 2; And

There is T1,2,3 and U1, the DIDO 3 × 2 of the use sky line options of 2.

Position of transmitting antenna is identical with Figure 27 with user device location.

Figure 29 shows the DIDO 3 × 2 with day line options can be provided compared with DIDO 2 × 2 system (not having selection) ~ gain of 5dB.Notice that channel is almost static (namely not having Doppler effect), thus selection algorithm to be applicable to path loss relevant with channel space, instead of rapid decay.We should see different gains having much higher general strangling in the situation of effect.And, in this particular experiment, observe Antenna Selection Algorithem and select antenna 2 and 3 for sending.

Iv. for the SNR threshold value of LUT

In selection [0171], we are stated model selection and are realized by LUT.LUT can be come by precomputation by assessment SNR threshold value to realize in different communication environments for the specific predefined target error rate performance of DIDO sending mode., we provide the performance of the DIDO system and do not have with day line options and transformable number of users below, described performance can be used as the guidance of structure LUT.Although Figure 24, Figure 26, Figure 28, Figure 29 are by obtaining with the actual measurement of DIDO prototype, figure is below obtained by emulation.Following BER result hypothesis does not have FEC.

Figure 30 shows the average BER performance of DIDO pre-coding schemes different in independent same distribution channel.The curve indicating " not having to select " refers to the situation using BD.In same figure, the performance of sky line options (ASel) for varying number additional antenna (user for varying number) and be illustrated.Can find out, along with the quantity of additional antenna increases, ASel provides better diversity gain (the BER slope of a curve in Yi Gao SNR district is feature), creates better covering.Such as, if target BER is fixed to 10 by us ^-2(actual value for uncoded system), then the SNR gain provided by ASel increases along with the quantity of antenna.

Figure 31 shows the SNR gain of the ASel of the function of the quantity as the extra transmitting antenna in independent same distribution channel for different target BER.Can find out, by means of only interpolation 1 or 2 antennas, ASel and BD compares and creates huge SNR gain.With in lower part, we by only for the situation of 1 or 2 additional antenna by target BER being fixed to 10-2(for uncoded system) assess the performance of ASel.

Figure 32 shows the SNR threshold value of the function as number of users (M) for BD and ASel in independent same distribution channel with 1 and 2 additional antenna.We observe the larger reception SNR demand due to the user for larger amt, and SNR threshold value increases along with M.Note, we suppose that for the user of any amount be fixing total transmitting power (transmitting antenna with varying number).In addition, Figure 32 show due to sky line options gain for any amount in independent same distribution channel user be constant.

Below, the performance of the DIDO system in spatial correlation channel is we illustrated.We emulate the channel of each user by the COST-259 spatial Channel Model described in " the Channel models for link and system level simulations " that deliver on IEEE 802.16Broadband Wireless Access Working Group in September, 2004 at X.Zhuang, F.W.Vook, K.L.Baum, T.A.Thomas and M.Cudak.We generate the single group being used for each user.As a kind of case study, we assume that NLOS channel, have uniform linear array (ULA) at transmitter, and element spacing is 0.5 λ.For the situation of 2 custom systems, we carry out emulation group with the average angle of AOA1 and AOA2 arrived respectively for the first and second users.AOA is measured relative to the side surface direction of ULA.When there being the user more than two in systems in which, we generate has at scope [-φ _m, φ _m] in the group of user of evenly spaced average A OA, wherein our definition

${\mathrm{\Φ}}_M=\frac{\mathrm{\Δ\φ}(M-1)}{2}---\left(13\right)$

K is the quantity of user, and △ φ is the angular distance between the average A OA of user.Note angular range [-φ _m, φ _m] center is 0 °, direction is penetrated in the side corresponding to ULA.Below, we study the BER performance of the DIDO system of the function as the angular distance between channel angle distribution (AS) and user with BD and ASel delivery plan and different numbers of users.

Figure 33 shows the BER with the average SNR relative to each user of two users of different AS value for being positioned at same angle direction (namely penetrating direction relative to the side of ULA, AOA1=AOA2=0 °).Can find out, along with AS increases, BER performance improvement and close to independent same distribution situation.In fact, higher AS statistically creates the better performance of less covering between the feature mode of two users and BD precoder.

Figure 34 shows the result similar to Figure 33, but has higher angular distance between users.We consider AOA1=0 °, AOA2=90 ° (i.e. 90 ° of angular distances).Best performance is achieved when low AS.In fact, for the situation of high angular distance, when angular distance is low, between the feature mode of user, there is less crossover.What is interesting is, we observe the identical reason for just having mentioned, and the BER performance in low AS is better than independent same distribution channel.

Next, for 10 in different relevant situation ^-2target BER, we calculate SNR threshold value.Figure 35 depicts the SNR threshold value of the function as AS of the different value of the average A OA for user.For low user's angular distance, the reliable transmission with rational SNR demand (i.e. 18dB) is only possible for the channel being feature with high AS.On the other hand, when user is spatially separated, less SNR is needed to meet identical target BER.

Figure 36 shows the SNR threshold value of the situation for 5 users.The user average A OA with the value of different angular distance △ φ is generated according to the definition in (13).We observe for △ φ=0 ° and AS<15 °, and due to the little angular distance between user, BD poor performance, does not meet target BER.For the AS increased, the SNR demand meeting fixing target BER reduces.On the other hand, for △ φ=30 °, obtain minimum SNR demand at low AS, consistent with the result in Figure 35.Along with AS increases, SNR threshold value is saturated to one in independent same distribution channel.Note, △ φ=30 ° with 5 users correspond to the AOA scope of [-60 °, 60 °], and this is typical for the base station had in the cellular system of 120 ° of sector elements.

Next, we have studied the performance of the ASel delivery plan in spatial correlation channel.Figure 37 compares the SNR threshold value with BD and ASel of 1 and 2 additional antenna for two user situations.We consider two kinds of different situations of the angular distance between user: { AOA1=0 °, AOA2=0 ° } and { AOA1=0 °, AOA2=90 ° }.Curve for BD scheme (namely not having a day line options) is identical with in Figure 35.We observe ASel creates 8dB and 10dB with 1 and 2 additional antenna respectively SNR gain for high AS.Along with AS reduces, the gain due to ASel on BD becomes less due to the quantity minimizing of the degree of freedom in MIMO broadcast channel.What is interesting is, for AS=0 ° (namely close to LOS channel) and situation { AOA1=0 °, AOA2=90 ° }, ASel does not provide any gain due to difference in the spatial domain.Figure 38 shows the result similar to Figure 37, but for the situation of 5 users.

The SNR threshold value that we calculate as the function of number of users (M) in systems in which for BD and ASel delivery plan (supposes 10 ^-2general objectives BER).SNR threshold value corresponds to average SNR, is constant to make total transmitting power for any M.We suppose at azimuth coverage [-φ _m, φ _m]=[-60 °, 60 °] in each customer group average A OA between largest interval.Then, the angular distance between user is △ φ=120 °/(M-1).

Figure 39 shows the SNR threshold value for the BD scheme with different AS value.We to observe due to user between large angular distance, for AS=0.1 ° (the namely low angular spread) of user (i.e. K≤20) with relatively smallest number, obtain minimum SNR demand.But for M>50, because △ φ is very little and BD can not carry out, SNR demand is far longer than 40dB.In addition, almost keep constant for AS>10 °, SNR threshold value for any M, the DIDO system in spatial correlation channel is close to the performance of independent same distribution channel.

In order to reduce the value of SNR threshold value and improve the performance of DIDO system, we apply ASel delivery plan.Figure 40 shows has the SNR threshold value in the spatial correlation channel of AS=0.1 ° for BD and ASel with 1 and 2 additional antenna.In order to reference, we also report the curve for independent same distribution situation shown in Figure 32.Can see, for less user (i.e. M≤10), owing to lacking diversity in DIDO broadcast channel, sky line options does not help to reduce SNR demand.Along with number of users increases, ASel is benefited from multi-user diversity, creates SNR gain (be namely 4dB for M=20).In addition, for M≤20, the performance with the ASel of 1 or 2 additional antenna in high spatial correlation channel is identical.

Then we calculate the SNR threshold value for two kinds of other channel conditions: the AS=10 ° in AS=5 ° in Figure 41 and Figure 42.Figure 41 and Figure 40 compares, and shows due to larger angular spread, and ASel creates also for the SNR gain of the user (i.e. M≤10) of relative small number.As reported in Figure 42, reduce further for AS=10 °, SNR threshold value, the gain due to ASel becomes higher.

Finally, we summarize at present for the result that correlated channels proposes.Figure 43 and Figure 44 show have respectively 1 and 2 additional antenna as the SNR threshold value for the number of users (M) of BD and ASel scheme and the function of angular spread (AS).Note, in fact the situation of AS=30 ° corresponds to independent same distribution channel, and we use this value of AS for diagrammatic representation in the drawings.We observe, although BD by channel space relevant affect, ASel creates the almost identical performance for any AS.In addition, for AS=0.1 °, due to multi-user diversity, ASel is similar for low M and BD performance, and for large M(and M >=20) more than BD.

Figure 49 compares the performance of DIDO schemes different in SNR threshold value.The DIDO scheme considered is: BD, ASel, the BD with feature mode selection (BD-ESel) and maximum-ratio combing (MRC).Notice that MRC does not eliminate the interference (unlike other method) at transmitter place in advance, but provide larger gain when user is spatially separated.In Figure 49, we depict when two users lay respectively to penetrate with the side of emission array direction become-30 ° and 30 ° time, for DIDO N × 2 system for target BER=10 ^-2sNR threshold value.We observe, and low AS, MRC scheme are provided compared with other scheme to the gain of 3dB, because the space channel of user is separated well, the impact of the interference between user is very little.Note, the gain of the MRC on DIDO N × 2 is due to array gain.Other scheme is exceeded for the AS being greater than 20 °, QR-ASel scheme and create the gain of about 10dB compared with the not selectable BD of tool 2 × 2.QR-ASel with BD-ESel provides approximately identical performance for the arbitrary value of AS.

Above-described is new self adaptation transmission technology for DIDO system.The method dynamic translation between DIDO sending mode strengthens throughput for fixing target error rate to different users.The performance of the DIDO system of different stage is measured under different propagation conditions, observes and can be realized as the DIDO pattern of the function of propagation condition and number of users by Dynamic Selection in the huge gain of throughput.

III. the precompensation of frequency and phase difference

a. background

As described above, wireless communication system uses carrier wave to transmit information.These carrier waves are normally sinusoidal wave, its amplitude and/or phase response modulated in the information be sent out.Sinusoidal wave nominal frequency is known as carrier frequency.In order to create this waveform, transmitter synthesizes one or two sine wave, and uses up-conversion to create the signal after the modulation overlapped on the sine wave with designated carrier frequency.This can realize by directly changing, and wherein, signal is from directly modulated on carrier wave or by multiple up-conversion stage.In order to process this waveform, the necessary RF signal received by demodulation of receiver, and effectively remove modulated carrier.This needs receiver to synthesize one or more sinusoidal signal to be reversed in the modulated process at transmitter place, is known as frequency reducing conversion.Regrettably, the sine wave signal generated at transmitter and receiver obtains from different reference oscillators.Reference oscillator is not had to create the frequency reference of perfection (perfect); In fact, usual and actual frequency has some deviations.

In a wireless communication system, the difference of the output of the reference oscillator at transmitter and receiver place creates the phenomenon being known as carrier frequency shift or simple frequency shift (FS) at receiver place.In essence, after down conversion, in received signal, there are some residue modulation (corresponding to the difference sent and in reception carrier).This create the distortion in received signal, result in higher bit error rate and lower throughput.

There is the different technologies for the treatment of carrier frequency shift.Most methods estimates the carrier frequency shift at receiver place, then applies offset correction of carrier frequency algorithm.Carrier frequency offset estimation algorithm uses following methods to be blindly (blind): offset-QAM (T.Fusco and M.Tanda, " BlindFrequency-offset Estimation for OFDM/OQAM Systems; " Signal Processing, IEEE Transactions on is [also see Acoustics, Speech, and Signal Processing, IEEETransactions on], vol.55, pp.1828-1838,2007); Cyclophysis (E.Serpedin, A.Chevreuil, G.B.Giannakis and P.Loubaton, " Blind channel and carrier frequencyoffset estimation using periodic modulation precoders; " Signal Processing, IEEETransactions on is [also see Acoustics, Speech, and Signal Processing, IEEETransactions on], vol.48, no.8, pp.2389-2405, Aug.2000); Or Cyclic Prefix (the J.J.van de Beek in OFDM (OFDM) structural approach, M.Sandell and P.O.Borjesson, " ML estimation of time and frequency offset in OFDM systems, " Signal Processing, IEEE Transactions on is [also see Acoustics, Speech, and SignalProcessing, IEEE Transactions on], vol.45, no.7, pp.1800-1805, July 1997; U.Tureli, H.Liu and M.D.Zoltowski, " OFDM blind carrier offset estimation:ESPRIT, " IEEE Trans.Commun., vol.48, no.9, pp.1459-1461, Sept.2000; M.Luise, M.Marselli and R.Reggiannini, " Low-complexity blind carrier frequencyrecovery for OFDM signals over frequency-selective radio channels; " IEEE Trans.Commun., vol.50, no.7, pp.1182-1188, July 2002).

Alternatively, special training signal can be utilized, comprise the data symbol (P.H.Moose of repetition, " A technique for orthogonal frequency division multiplexing frequency offsetcorrection; " IEEE Trans.Commun., vol.42, no.10, pp.2908-2914, Oct.1994); Two different symbol (T.M.Schmidl and D.C.Cox, " Robust frequency and timingsynchronization for OFDM, " IEEE Trans.Commun., vol.45, no.12, pp.1613-1621, Dec.1997); Or known symbol sebolic addressing (M.Luise and R.Reggiannini, " the Carrier frequency acquisition and tracking for OFDM systems periodically inserted, " IEEE Trans.Commun., vol.44, no.11, pp.1590-1598, Nov.1996).Correction can occur in analog or digital mode.Receiver Carrier frequency offset estimation can also be used signal that precorrection sends is to eliminate skew.Because multicarrier and ofdm system are to the sensitivity of frequency shift (FS), offset correction of carrier frequency is extensively studied (J.J.van de Beek for multicarrier and ofdm system, M.Sandell and P.O.Borjesson, " ML estimation of time and frequency offset inOFDM systems; " Signal Processing, IEEE Transactions on is [also see Acoustics, Speech, and Signal Processing, IEEE Transactions on], vol.45, no.7, pp.1800-1805, July 1997; U.Tureli, H.Liu and M.D.Zoltowski, " OFDM blindcarrier offset estimation:ESPRIT, " IEEE Trans.Commun., vol.48, no.9, pp.1459-1461, Sept.2000; T.M.Schmidl and D.C.Cox, " Robust frequency andtiming synchronization for OFDM, " IEEE Trans.Commun., vol.45, no.12, pp.1613-1621, Dec.1997; M.Luise, M.Marselli and R.Reggiannini, " Low-complexity blind carrier frequency recovery for OFDM signals overfrequency-selective radio channels; " IEEE Trans.Commun., vol.50, no.7, pp.1182-1 188, July 2002).

Frequency offset estimation and correct for multiple antenna communication or more generally MIMO(multiple-input and multiple-output) system is important problem.In mimo systems, transmitting antenna is locked into a frequency reference, and receiver is locked into another frequency reference, has single skew between the transmitter and receiver.Propose several algorithm to use this problem of training signal process (K.Lee and J.Chun, " Frequency-offset estimation for MIMO and OFDM systems using orthogonaltraining sequences; " IEEE Trans.Veh.Technol., vol.56, no.1, pp.146-156, Jan.2007; M.Ghogho and A.Swami, " Training design for multipath channel andfrequency offset estimation in MIMO systems; " Signal Processing, IEEETransactions on is [also see Acoustics, Speech, and Signal Processing, IEEETransactions on], vol.54, no.10, pp.3957-3965, Oct.2006; And adaptivetracking C.Oberli and B.Daneshrad, " Maximum likelihood tracking algorithms forMIMOOFDM; " in Communications, 2004IEEE International Conference on, vol.4, June 20-24,2004, pp.2468-2472).Encounter prior problem in mimo systems, wherein, transmitting antenna is not locked into same frequency reference, but reception antenna is locked into together.In fact this occur in the up link of space division multiple access access (SDMA) system, and SDMA system is regarded as mimo system, and wherein different user corresponds to different transmitting antennas.In this case, the compensation of frequency shift (FS) is more complicated.Particularly, frequency shift (FS) creates the interference in the different MIMO stream be sent out.Complicated Combined estimator and equalization algorithm can be used to carry out correcting (A.Kannan, T.P.Krauss and M.D.Zoltowski, " Separation of cochannel signals under imperfecttiming and carrier synchronization, " IEEE Trans.Veh.Technol., vol.50, no.1, pp.79-96, Jan.2001), and equilibrium (T.Tang and R.W.Heath after Frequency offset estimation, " Joint frequency offset estimation and interference cancellation forMIMO-OFDM systems [mobile radio], " 2004.VTC2004-Fall.2004IEEE 60 ^thvehicular Technology Conference, vol.3, pp.1553-1557, Sept.26-29,2004, X.Dai, " Carrier frequency offset estimation for OFDM/SDMA systems usingconsecutive pilots, " IEEE Proceedings-Communications, vol.152, pp.624-632, Oct.7,2005).The a few thing process relevant issues of excess phase shift and tracking error, wherein excess phase shift is estimated and is compensated after Frequency offset estimation, but this work only considers up link (the L Haring of SDMAOFDMA system, S.Bieder and A.Czylwik, " Residual carrierand sampling frequency synchronization in multiuser OFDM systems; " 2006. VTC 2006-Spring.IEEE 63rd Vehicular Technology Conference, vol.4, pp.1937-1941,2006).When all transmitting and receiving antennas have different frequency references, there is the most serious situation in mimo systems.Asymptotic analysis (O.Besson and P.Stoica of the evaluated error in flat fading channel has only been processed about the only available work of this topic, " On parameterestimation of MIMO flat-fading channels with frequency offsets; " SignalProcessing, IEEE Transactions on is [also see Acoustics, Speech, and SignalProcessing, IEEE Transactions on], vol.51, no.3, pp.602-613, Mar.2003).

When the different transmit antennas of mimo system does not have identical frequency reference, and reception antenna independently processing signals time, occurred by situation about furtheing investigate.This occurs in be known as in distributed input and output (DIDO) communication system (in the literature also referred to as MIMO broadcast channel) and occurs.DIDO system comprises an access point with spaced antenna, described antenna sends data streams in parallel (via precoding) strengthens down link throughput to multiple user, now uses identical Radio Resource (namely identical time-slot duration and frequency band) as conventional SISO system.DIDO system be described in detail in S.G.Perlman and T.Cotter, in July, 2004 submit to be entitled as in the U.S. Patent application 20060023803 of " System and method fordistributed input-distributed output wireless communications " propose.There is the mode of many enforcement DIDO precoders.A solution is block diagonalization (BD); describe in such as with Publication about Document: Q.H.Spencer; A.L.Swindlehurst and M.Haardt; " Zero-forcing methods for downlink spatial multiplexing inmultiuser MIMO channels; " IEEE Trans.Sig.Proc; vol.52, pp.461-471, Feb.2004; K.K.Wong, R.D.Murch and K.B.Letaief, " A joint-channeldiagonalization for multiuser MIMO antenna systems, " IEEE Trans.WirelessComm., vol.2, pp.773-786, JuI 2003; L.U.Choi and R.D.Murch; " A transmitpreprocessing technique for multiuser MIMO systems using a decompositionapproach, " IEEE Trans.Wireless Comm., vol.3; pp.20-24, Jan 2004; Z.Shen; J.G.Andrews, R.W.Heath and B.L Evans, " Low complexity user selectionalgorithms for multiuser MIMO systems with block diagonalization; " be accepted and be published in IEEE Trans.Sig.Proc, Sep.2005; Z.Shen, R.Chen, J.G.Andrews; R.W.Heath and B.L Evans; " Sum capacity of multiuser MIMO broadcast channels withblock diagonalization, " is submitted to IEEE Trans.Wireless Comm., Oct.2005; R.Chen, R.W.Heath and J.G.Andrews, " Transmit selection diversity for unitaryprecoded multiuser spatial multiplexing systems with linear receivers; " be accepted to IEEE Trans, on Signal Processing, 2005.

In DIDO system, send precoding and be used to be separated the data flow for different user.When transmitting antenna rf chain does not share same frequency reference, carrier frequency shift result in the Railway Project relevant to System Implementation.When it happens, each antenna effectively sends with slightly different carrier frequency.This destroys the integrality of DIDO precoder, cause each user to suffer extra interference.What propose below is several solutions to this problem.In an execution mode of solution, DIDO transmitting antenna carrys out a shared frequency reference by wired, optics or wireless network.In another execution mode of solution, one or more user's estimated frequency offset differences (antenna between skew in relative different) and this information is sent it back transmitter.Then transmitter precorrection frequency shift (FS) proceed to estimate phase place for the training of DIDO and precoder.Delay time is there is and has problem in this execution mode at feedback channel.Reason possible have the residual phase error created by trimming process, and this trimming process does not consider channel estimating subsequently.In order to address this problem, an other execution mode uses new frequency shift (FS) and phase estimating device, solves this problem by estimated delay.Based on emulating and providing result by the actual measurement that DIDO-OFDM prototype performs.

The frequency proposed in this document and phase offset compensation method may be sensitiveer to the evaluated error of the noise due to receiver place.Therefore, another execution mode proposes the method estimated for the time and frequency shift, also very strong under low SNR condition.

There is the different method for time of implementation and Frequency offset estimation.Because it is to the sensitivity of synchronous error, the many methods in these methods propose for OFDM waveform specially.

These algorithms typically use the structure of OFDM waveform, are therefore generally enough for single carrier wave and multicarrier waveform.Algorithm described below is using known fiducial mark (such as, training data) with among the class of assisting synchronous technology.Many methods are that the expansion of the frequency offset estimator of Moose is (see P.H.Moose, " A technique for orthogonal frequency divisionmultiplexing frequency offset correction; " IEEE Trans.Commun., vol.42, no.10, pp.2908-2914, Oct.1994).Moose propose use two repeat training signal and the phase difference be used between received signal to obtain frequency shift (FS).The method of Moose only can correct mark (fractional) frequency shift (FS).The expansion of the method for Moose proposes (T.M.Schmidl and D.C.Cox, " Robust frequency and timing synchronization forOFDM by Schmidl and Cox, " IEEE Trans.Commun., vol.45, no.12, pp.1613-1621, Dec.1997).Their main innovation is to use the OFDM symbol of one-period and the training symbol of other differential coding.The integer offset correction that differential coding in second symbol realizes.Coulson considers at T.M.Schmidl and D.C.Cox, " Robust frequency and timing synchronization forOFDM, " IEEE Trans.Commun., vol.45, no.12, pp.1613-1621, the similar setting described in Dec.1997, and at A.J.Coulson, " Maximum likelihood synchronization forOFDM using a pilot symbol:analysis, " IEEE J.Select.Areas Commun., vol.19, no.12, pp.2495-2503, Dec.2001 and A.J.Coulson, " Maximum likelihoodsynchronization for OFDM using a pilot symbol:algorithms, " IEEE J.Select.Areas Commun., vol.19, no.12, pp.2486-2494, the detailed discussion of algorithm and analysis is provided in Dec.2001.Main difference is the correlation properties that maximal-length sequence that Coulson employs repetition provides.He also advises using linear frequency modulation (chirp) signal, because its constant envelope properties in time domain and frequency domain.Coulson considers actual details but does not comprise integer and estimates.The training signal of multiple repetition is by Minn et.al.in H.Minn, V.K.Bhargava and K.B.Letaief, " A robust timing and frequency synchronization for OFDM systems; " IEEETrans.Wireless Commun., vol.2, no.4, pp.822-839, July 2003 considered, but the structure of training is not optimised.Shi and Serpedin proposes some optimalitys (K.Shi and E.Serpedin that training structure has the idea forming frame synchronization, " Coarse frame and carrier synchronization of OFDM systems:a new metric and comparison; " IEEE Trans.Wireless Commun., vol.3, no.4, pp.1271-1284, July 2004).An embodiment of the invention employ the method for Shi and Serpedin to perform frame synchronization and fractional frequency offset estimation.

Many methods in the literature concentrate on frame synchronization and fractional frequency offset corrects.Integer offset correction is used in T.M.Schmidl and D.C.Cox, " Robust frequency and timingsynchronization for OFDM, " IEEE Trans.Commun., vol.45, other training symbol in no.12, pp.1613-1621, Dec.1997 is solved.Such as, Morrelli etc. are at M.Morelli, A.N.D'Andrea and U.Mengali, " Frequency ambiguity resolution inOFDM systems, " IEEE Commun.Lett., vol.4, T.M.Schmidl and D.C.Cox is obtained in no.4, pp.134-136, Apr.2000, " Robust frequency and timingsynchronization for OFDM; " IEEE Trans.Commun., vol.45, no.12, the improvement version of pp.1613-1621, Dec.1997.The interchangeable method of different preamble structures is used to propose (M.Morelli and U.Mengali by Morelli and Mengali, " An improvedfrequency offset estimator for OFDM applications; " IEEE Commun.Lett., vol.3, no.3, pp.75-77, Mar.1999).This method correlation employed between the identical training symbol of M repetition carrys out the scope being increased fractional frequency offset estimator by the M factor.This is best linear unbiased estimator and receives maximum skew (having suitable design), but the timing synchronization do not provided.

system describes

An embodiment of the invention use eliminates frequency in DIDO system and phase deviation based on the precoding of channel condition information.See Figure 11 and for the associated description above the description of this execution mode.

In an embodiment of the invention, each user uses the receiver being equipped with frequency offset estimator/compensator.Go out as shown in Figure 45, in an embodiment of the invention, the system comprising receiver comprises multiple RF unit 4508, corresponding multiple A/D unit 4510, the receiver being equipped with frequency offset estimator/compensator 4512 and DIDO feedback generator unit 4506.

RF unit 4508 receives the signal sent from DIDO transmitter unit, and signal down is transformed into base band, and the signal after frequency reducing being changed is provided to A/D unit 4510.Then signal from analog is converted to numeral by A/D unit 4510, and it is sent to frequency offset estimator/compensator units 4512.Frequency offset estimator/compensator units 4512 estimated frequency skew also compensating frequency deviation, as described here, is then sent to OFDM unit 4513 by the signal after compensation.OFDM unit 4513 removes Cyclic Prefix and runs fast Fourier transform (FFT) and signal is reported to frequency domain.At training period, OFDM unit 4513 is sent to channel estimating unit 4504 calculates channel estimating in a frequency domain by exporting.Alternatively, channel estimating can be calculated in the time domain.During the data cycle, output is sent to DIDO receiver unit 4502 by OFDM unit 4513, and this DIDO receiver unit 4502 pairs of signals carry out demodulate/decode to obtain data.Channel estimating is sent to DIDO feedback generator unit 4506 by channel estimating unit 4504, and this DIDO feedback generator unit 4506 can be estimated and via feedback control channel, they be sent it back transmitter by quantized channel, as shown.

to the description of an execution mode of the algorithm for DIDO 2 × 2 situation

Described below is the execution mode of algorithm for the frequency/phase migration in DIDO system.DIDO system model starts to be described when having and do not have frequency/phase skew.In order to easy, provide the particular implementation of DIDO 2 × 2 system.But general principle of the present invention can also be implemented in high-order DIDO system.

there is/not have the DIDO system model of frequency and phase deviation

The received signal of DIDO 2 × 2 can be write as first user:

r ₁[t]＝h ₁₁(w ₁₁x ₁[t]+w ₂₁x ₂[t])+h ₁₂(w ₁₂x ₁[t]+w ₂₂x ₂[t]) （1）

And second user is write as:

r ₂[t]＝h ₂₁(w ₁₁x ₁[t]+w ₂₁x ₂[t])+h ₂₂(w ₁₂x ₁[t]+w ₂₂x ₂[t]) （2）

Wherein t is discrete time index, h _mnand w _mnthe channel between m user and the n-th transmitting antenna and DIDO precoding weight respectively, x _mit is the transmission signal for user m.Note, h _mnand w _mnnot the function of t, because we suppose that on the cycle of channel between training and data send be constant.

When frequency and phase deviation exist, the signal received is represented as

$r_1\left[t\right]=e^{j({\mathrm{\&omega;}}_{U1}-{\mathrm{\&omega;}}_{T1})T_s(t-t_{11})}{h}_{11}\left(w_{11}{x}_1\right[t]+w_{21}{x}_2[t\left]\right)+e^{j^{({\mathrm{\&omega;}}_{U1}-{\mathrm{\&omega;}}_{T2})T_s\left({t-t}_{12}\right)}}{h}_{12}\left(w_{12}{x}_1\right[t]+w_{22}{x}_2[t\left]\right)---\left(3\right)$

And

$r_2\left[t\right]=e^{j({\mathrm{\&omega;}}_{U2}-{\mathrm{\&omega;}}_{T1})T_s(t-t_{21})}{h}_{21}\left(w_{11}{x}_1\right[t]+w_{21}{x}_2[t\left]\right)+e^{j^{({\mathrm{\&omega;}}_{U2}-{\mathrm{\&omega;}}_{T2})T_s\left({t-t}_{22}\right)}}{h}_{22}\left(w_{12}{x}_1\right[t]+w_{22}{x}_2[t\left]\right)---\left(4\right)$

Wherein, T _sthe is-symbol cycle; For the n-th transmitting antenna, ω _tn=2 ∏ f _tn; For m user, ω _um=2 ∏ f _um; And f _tnand f _umthe practical carrier frequency (by bias effect) for the n-th transmitting antenna and m user respectively.Value t _mnrepresent at channel h _mnon cause the random delay of phase deviation.Figure 46 depicts DIDO 2 × 2 system model.

For the time, we use to give a definition:

Δω _mn＝ω _Um-ω _Tn（5）

To be used for representing the frequency shift (FS) between m user and the n-th transmitting antenna.

the description of an embodiment of the invention

Method according to an embodiment of the invention is illustrated in Figure 47.The method comprises following general step and (comprises sub-step, as shown): for the cycle of training 4701 of Frequency offset estimation; For the cycle of training 4702 of channel estimating; Data via the balanced DIDO precoding of tool send 4703.These steps are described in detail in the following.

A () is for the cycle of training (4701) of Frequency offset estimation

During the first cycle of training, the one or more training signals from each transmitting antenna are sent to one (4701a) in user by base station.As described here, " user " is wireless client device.For the situation of DIDO 2 × 2, the signal received by m user is provided by following:

$r_m\left[t\right]=e^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{m1}{T}_s(t-t_{m1})}{h}_{m1}{p}_1\left[t\right]+e^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{m2}{T}_s(t-t_{m2})}{h}_{m2}{p}_2\left[t\right]---\left(6\right)$

Wherein, p ₁and p ₂the training sequence sent from the first and second antennas respectively.

M user can use the frequency offset estimator of any type (namely by the convolution of training sequence) and estimate shifted by delta ω _mlwith Δ ω _m2.Then, according to these values, user calculates the frequency shift (FS) between two transmitting antennas:

Δω _T＝Δω _m2-Δω _m1＝ω _T1-ω _T2（7）

Finally, the value in (7) is fed back to base station (4701b).

Note, p in (6) ₁and p ₂it is orthogonal for being designed to, thus user can estimate Δ ω _mlwith Δ ω _m2.Alternatively, in one embodiment, identical training sequence is used in two continuous print time slots, and user therefrom estimates skew.In addition, in order to improve the estimation of the skew in (7), above-described identical calculating for DIDO system all users (not only for m user) can be done, last estimation can be (after the weighting) mean value of the value obtained from all users.But this solution needs more computing time and feedback quantity.Finally, the renewal of the Frequency offset estimation only just needs when frequency shift (FS) changes in time.Therefore, according to the stability of the clock at transmitter place, the step 4701 of algorithm can be performed (namely sending for each data) in long-term basis, and above-mentioned feedback is reduced.

B () is for the cycle of training (4702) of channel estimating

During the second cycle of training, base station is first from m user or the frequency shift (FS) feedback maybe must from multiple user with the value (7).Value in (7) is used to the frequency shift (FS) of precompensation at transmitting terminal.Then, training data is sent to all users and comes for channel estimating (4702a) by base station.

For DIDO 2 × 2 system, the signal received at first user place is provided by following:

$r_1\left[t\right]=e^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{11}{T}_s(t-{\tilde{t}}_{11})}{h}_{11}{p}_1\left[t\right]+e^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{12}{T}_s(t-{\tilde{t}}_{12})}{h}_{12}{e}^{-\mathrm{j\&Delta;}{\mathrm{\&omega;}}_T{T}_{s}t}{p}_2\left[t\right]---\left(8\right)$

And second user place:

$r_2\left[t\right]=e^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{21}{T}_s(t-{\tilde{t}}_{21})}{h}_{21}{p}_1\left[t\right]+e^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{22}{T}_s(t-{\tilde{t}}_{22})}{h}_{22}{e}^{-\mathrm{j\&Delta;}{\mathrm{\&omega;}}_T{T}_{s}t}{p}_2\left[t\right]---\left(9\right)$

Wherein, with Δ t is the random or known delay between first of base station sends and second sends.In addition, p ₁and p ₂the training sequence sent from the first and second antennas of user's frequency shift (FS) and channel estimating respectively.

Note, precompensation is only applied to the second antenna in this embodiment.

Launch (8), we obtain

$r_1\left[t\right]=e^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{11}{T}_{s}t}{e}^{{\mathrm{j\&theta;}}_{11}}\left[h_{11}{p}_1\right[t]+e^{j({\mathrm{\&theta;}}_{12}-{\mathrm{\&theta;}}_{11})}{h}_{12}{p}_2[t\left]\right]---\left(10\right)$

Similar to the second user:

$r_2\left[t\right]=e^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{21}{T}_{s}t}{e}^{{\mathrm{j\&theta;}}_{21}}\left[h_{21}{p}_1\right[t]+e^{j({\mathrm{\&theta;}}_{22}-{\mathrm{\&theta;}}_{21})}{h}_{22}{p}_2[t\left]\right]---\left(11\right)$

Wherein, ${\mathrm{\&theta;}}_{\mathrm{mn}}=-\mathrm{\&Delta;}{\mathrm{\&omega;}}_{\mathrm{mn}}{T}_s{\tilde{t}}_{\mathrm{mn}}.$

At receiving terminal, user is by using training sequence p ₁and p ₂carry out compensating frequency offset residue.Then, user is undertaken estimating (4702b) by trained vector channel:

$h_1=\left[\begin{array}{c}{h}_{11}\\ e^{j({\mathrm{\&theta;}}_{12}-{\mathrm{\&theta;}}_{11})}{h}_{12}\end{array}\right]$ $h_2=\left[\begin{array}{c}{h}_{21}\\ e^{j({\mathrm{\&theta;}}_{22}-{\mathrm{\&theta;}}_{21})}{h}_{22}\end{array}\right]---\left(12\right)$

These channels in (12) or channel condition information (CSI) are fed back to base station (4702b), and base station calculates DIDO precoder as described in lower part.

C () has the DIDO precoding (4703) of precompensation

Base station is received the channel condition information (CSI) (12) from user and is calculated the weight (4703a) of precoding by block diagonalization (BD), to make

$w_1^T{h}_2=0,$ $w_2^T{h}_1=0---\left(13\right)$

Wherein, vector h ₁be defined in (12), and w _m=[w _m1, w _m2].Note, the present invention proposed in the disclosure can be used in any other DIDO method for precoding except BD.Delay (the Δ t sent between current transmission also by using the estimation in (7) to carry out precompensation frequency shift (FS), and is trained in base station by estimation second ₀) carry out precompensation phase deviation (4703a).Finally, data are sent to user (4703b) via DIDO precoder by base station.

After process of transmitting, the signal received at user 1 place is provided by following:

$r_1\left[t\right]=e^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{11}{T}_s(t-{\tilde{t}}_{11}-{\mathrm{\&Delta;t}}_o)}{h}_{11}\left[w_{11}{x}_1\right[t]+w_{21}{x}_2[t\left]\right]$

$=e^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{12}{T}_s(t-{\tilde{t}}_{12}-\mathrm{\&Delta;}{t}_o)}{h}_{12}{e}^{-\mathrm{j\&Delta;}{\mathrm{\&omega;}}_T{T}_S(t-{\mathrm{\&Delta;t}}_o)}\left[w_{12}{x}_1\right[t]+w_{22}{x}_2[t\left]\right]$

$={\mathrm{\&gamma;}}_1\left[t\right][h_{11}\left(w_{11}{x}_1\right[t]+w_{21}{x}_2[t\left]\right)+e^{j(\mathrm{\&Delta;}{\mathrm{\&omega;}}_{11}{t}_{11}-\mathrm{\&Delta;}{\mathrm{\&omega;}}_{12}{t}_{12})T_s}{h}_{12}\left(w_{12}{x}_1\right[t]+w_{22}{x}_2[t\left]\right)]$

$={\mathrm{\&gamma;}}_1\left[t\right]\left[(h_{11}{w}_{11}+e^{j({\mathrm{\&theta;}}_{12}-{\mathrm{\&theta;}}_{11})}{h}_{12}{w}_{12})x_1\right[t]+(h_{11}{w}_{21}+e^{j({\mathrm{\&theta;}}_{12}-{\mathrm{\&theta;}}_{11})}{h}_{12}{w}_{22})x_2[t\left]\right]---\left(14\right)$

Wherein, use attribute (13), we obtain

$r_1\left[t\right]={\mathrm{\&gamma;}}_1\left[t\right]w_1^T{h}_1{x}_1\left[t\right]---\left(15\right)$

Similarly, for user 2, we obtain:

$r_2\left[t\right]=e^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{21}{T}_s(t-{\tilde{t}}_{21}-{\mathrm{\&Delta;t}}_o)}{h}_{21}\left[w_{11}{x}_1\right[t]+w_{21}{x}_2[t\left]\right]$

$+e^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{22}{T}_s(t-{\tilde{t}}_{22}-\mathrm{\&Delta;}{t}_o)}{h}_{22}{e}^{-\mathrm{j\&Delta;}{\mathrm{\&omega;}}_T{T}_s(t-{\mathrm{\&Delta;t}}_o)}\left[w_{12}{x}_1\right[t]+w_{22}{x}_2[t\left]\right]---\left(16\right)$

Launch (16):

$r_2\left[t\right]={\mathrm{\&gamma;}}_2\left[t\right]w_2^T{h}_2{x}_2\left[t\right]---\left(17\right)$

Wherein, ${\mathrm{\&gamma;}}_2\left[t\right]=e^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{21}{T}_s(t-{\tilde{t}}_{21}-\mathrm{\&Delta;}{t}_o)}.$

Finally, user calculates frequency offset residue and channel estimating carrys out demodulated data stream x ₁[t] and x ₂[t] (4703c).

be generalized to DIDO N × M

In the portion, the technology before described is generalized to the DIDO system with N number of transmitting antenna and M user.

i. the cycle of training of user's Frequency offset estimation

During the first cycle of training, because the signal received by m user of the training sequence sent from N number of antenna is provided by following:

$r_m\left[t\right]=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{e}^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{\mathrm{mn}}{T}_s(t-t_{\mathrm{mn}})}{h}_{\mathrm{mn}}{p}_n\left[t\right]---\left(18\right)$

Wherein, p _nit is the training sequence sent from the n-th antenna.

At estimation shifted by delta ω _mnafterwards, m user calculates the frequency shift (FS) between first and the n-th transmitting antenna:

Δω _T，1n＝Δω _mn-Δω _m1＝ω _T1-ω _Tn（19）

Finally, the value in (19) is fed back to base station.

ii. for the cycle of training of channel estimating

During the second cycle of training, base station is first from m user or the frequency shift (FS) feedback obtaining the value had (19) from multiple user.Value in (19) is used to the frequency shift (FS) of precompensation at transmitting terminal.Then, training data is sent to all users and comes for channel estimating by base station.

For DIDO N × M system, the signal received at m user place is provided by following:

$r_m\left[t\right]=e^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{m1}{T}_S(t-{\tilde{t}}_{m1})}{h}_{m1}{p}_1\left[t\right]+\underset{n=2}{\overset{N}{\mathrm{\&Sigma;}}}{e}^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{\mathrm{mn}}{T}_s(t-{\tilde{t}}_{\mathrm{mn}})}{h}_{\mathrm{mn}}{e}^{-\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{T,1n}{T}_{s}t}{p}_n\left[t\right]$

$=e^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{m1}{T}_s(t-{\tilde{t}}_{m1})}\left[h_{m1}{p}_1\right[t]+\underset{n=2}{\overset{N}{\mathrm{\&Sigma;}}}{e}^{j({\mathrm{\&theta;}}_{\mathrm{mn}}-{\mathrm{\&theta;}}_{m1})}{h}_{\mathrm{mn}}{p}_n[t\left]\right]$

$=e^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{m1}{T}_s(t-{\tilde{t}}_{m1})}\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{e}^{j({\mathrm{\&theta;}}_{\mathrm{mn}}-{\mathrm{\&theta;}}_{m1})}{h}_{\mathrm{mn}}{p}_n\left[t\right]---\left(20\right)$

Wherein, and Δ t is the random or known delay between first and second of base station sends.In addition, P _nit is the training sequence sent from the n-th antenna for frequency shift (FS) and channel estimating.

At receiver side, user is by using training sequence P _ncarry out compensating frequency offset residue.Then, each user m is estimated by trained vector channel:

$h_m=\left[\begin{array}{c}{h}_{m1}\\ e^{j({\mathrm{\&theta;}}_{m2}-{\mathrm{\&theta;}}_{m1})}{h}_{m2}\\ .\\ .\\ .\\ e^{j({\mathrm{\&theta;}}_{\mathrm{mN}}-{\mathrm{\&theta;}}_{m1})}{h}_{\mathrm{mN}}\end{array}\right]---\left(21\right)$

And feeding back to base station, base station is as to calculate DIDO precoder as described in lower part.

iii. there is the DIDO precoding of precompensation

Base station is received the channel condition information (CSI) (12) from user and is calculated the weight of precoding by block diagonalization (BD), to make

$w_m^T{h}_l=0,\&ForAll;m\&NotEqual;l,m=1,...,M---\left(22\right)$

Wherein, vector h _mbe defined in (21), and w _m=[w _m1, w _m2..., w _mN].Delay (the Δ t sent between current transmission also by using the estimation in (19) to carry out precompensation frequency shift (FS), and is trained in base station by estimation second ₀) carry out precompensation phase deviation.Finally, data are sent to user via DIDO precoder by base station.

After process of transmitting, the signal received at user i place is provided by following:

$r_i\left[t\right]=e^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{i1}{T}_s(t-{\tilde{t}}_{i1}-\mathrm{\&Delta;}{t}_o)}{h}_{i1}\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{w}_{m1}{x}_m\left[t\right]+$

$+\underset{n=2}{\overset{N}{\mathrm{\&Sigma;}}}{e}^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{\mathrm{in}}{T}_s(t-{\tilde{t}}_{\mathrm{in}}-\mathrm{\&Delta;}{t}_o)}{h}_{\mathrm{in}}{e}^{-\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{T,1n}{T}_s(t-{\mathrm{\&Delta;}t}_o)}\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{w}_{\mathrm{mn}}{x}_m\left[t\right]$

$=e^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{i1}{T}_s(t-\mathrm{\&Delta;}{t}_o)}{e}^{-\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{i1}{T}_s{\tilde{t}}_{i1}}{h}_{i1}\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{w}_{m1}{x}_m\left[t\right]$

$+\underset{n=2}{\overset{N}{\mathrm{\&Sigma;}}}{e}^{\mathrm{j\&Delta;}{\mathrm{\&omega;}}_{i1}{T}_s(t-\mathrm{\&Delta;}{t}_o)}{e}^{-j{\mathrm{\&Delta;\&omega;}}_{\mathrm{in}}{T}_s{\tilde{t}}_{\mathrm{in}}}{h}_{\mathrm{in}}\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{w}_{\mathrm{mn}}{x}_m\left[t\right]$

$={\mathrm{\&gamma;}}_i\left[t\right]\left[h_{i1}\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{w}_{m1}{x}_m\right[t]+\underset{n=2}{\overset{N}{\mathrm{\&Sigma;}}}{e}^{j({\mathrm{\&theta;}}_{\mathrm{in}}-{\mathrm{\&theta;}}_{i1})}{h}_{\mathrm{in}}\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{w}_{m1}{x}_m[t\left]\right]$

$={\mathrm{\&gamma;}}_i\left[t\right]\left[\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{e}^{j({\mathrm{\&theta;}}_{\mathrm{in}}-{\mathrm{\&theta;}}_{i1})}{h}_{\mathrm{in}}\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{w}_{\mathrm{mn}}{x}_m\right[t\left]\right]$

$={\mathrm{\&gamma;}}_i\left[t\right]\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}\left[\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{e}^{j({\mathrm{\&theta;}}_{\mathrm{in}}-{\mathrm{\&theta;}}_{i1})}{h}_{\mathrm{in}}{w}_{\mathrm{mn}}\right]x_m\left[t\right]$

$={\mathrm{\&gamma;}}_i\left[t\right]\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{w}_m^T{h}_i{x}_m\left[t\right]---\left(23\right)$

Wherein, use attribute (22), we obtain:

$r_i\left[t\right]={\mathrm{\&gamma;}}_i\left[t\right]w_i^T{h}_i{x}_i\left[t\right]---\left(24\right)$

Finally, user calculates frequency offset residue and channel estimating carrys out demodulated data stream x _i[t].

result

Figure 48 shows the SER result of DIDO 2 × 2 system and do not have with frequency shift (FS).Can see, the method proposed completely eliminates frequency/phase skew, creates the SER identical with the system without skew.

Next, we assess the sensitivity of proposed compensation method for the fluctuation of frequency offset error and/or real time offset.Therefore, (14) are rewritten as by we:

$r_1\left[t\right]=e^{j{\mathrm{\&Delta;\&omega;}}_{11}{T}_s(t-{\tilde{t}}_{11}-\mathrm{\&Delta;}{t}_o)}{h}_{11}\left[w_{11}{x}_1\right[t]+w_{21}{x}_2[t\left]\right]$

${+e}^{j{\mathrm{\&Delta;\&omega;}}_{12}{T}_s(t-{\tilde{t}}_{12}-\mathrm{\&Delta;}{t}_o)}{h}_{12}{e}^{-j(\mathrm{\&Delta;}{\mathrm{\&omega;}}_T+2\mathrm{\&Pi;}\&Element;)T_s(t-\mathrm{\&Delta;}{t}_o)}\left[w_{12}{x}_1\right[t]+w_{22}{x}_2[t\left]\right]---\left(25\right)$

Wherein, ε represents evaluated error and/or the change of the frequency shift (FS) between training and data transmission.Note, the effect of ε destroys the orthogonal property in (13), do not eliminated in advance completely at transmitter place to make the distracter in (14) and (16).Because like this, SER performance reduces along with the ε value increased.

Figure 48 shows the SER performance of the frequency offset compensation method for different ∈ values.These results hypothesis T _snamely=0.3ms(has the signal of 3KHz bandwidth).We observe, for ε=0.001Hz(or less), SER performance is similar to the situation not having to offset.

f. for the description of an execution mode of the algorithm of time and frequency shift estimation

Below, we describe the other execution mode (4701b in Figure 47) of time of implementation and Frequency offset estimation.The structure that transmits considered is at H.Minn, V.K.Bhargava and K.B.Letaief, " A robust timing and frequency synchronization for OFDM systems, " IEEETrans.Wireless Commun., vol.2, no.4, pp.822-839, propose in July 2003, at K.Shi and E.Serpedin, " Coarse frame and carrier synchronization of OFDM systems:a new metric and comparison, " IEEE Trans.Wireless Commun., vol.3, no.4, pp.1271-1284, studied in great detail in July 2004.The sequence of the association attributes usually had is used to training.Such as, for our system, Chu sequence is used, Chu sequence as at D.Chu, " Polyphasecodes with good periodic correlation properties (corresp.), " IEEE Trans.Inform.Theory, vol.18, be described in no.4, pp.531-532, July 1972.These sequences have interesting attribute, and namely they have perfect circular correlation.Allow L _cprepresent the length of Cyclic Prefix, N _trepresent the length of component training sequence.Make N _t=M _t, wherein M _tthe length of training sequence.Under these assumptions, the symbol sebolic addressing for the beginning sent can be written to:

S [n]=t [n-N _t] for n=-1 ... ,-L _cp

S [n]=t [n] for n=0 ..., N _t-1

S [n]=t [n-N _t] for n=N _t..., 2N _t-1

S [n]=-t [n-2N _t] for n=2N _t..., 3N _t-1

S [n]=t [n-3N _t] for n=3N _t..., 4N _t-1

Notice that the structure of this training signal can be expanded to other length, but be repeated block structure.Such as, in order to use 16 training signals, we consider a kind of structure, such as:

[CP,B,B,-B,B,B,B,-B,B,-B,-B,B,-B,B,B,-B,B,]。

By using this structure, and make N _t=4M _t, all algorithms that will describe can be used when not revising.Effectively, our repetition training sequence.This is particularly useful in the disabled situation of suitable training signal possibility.

After the filtering that symbol rate is mated and down sample, consider received signal below:

$r\left[n\right]=e^{2\mathrm{\&pi;\&epsiv;n}}\underset{l=0}{\overset{L}{\mathrm{\&Sigma;}}}h\left[l\right]s[n-l-\mathrm{\&Delta;}]+v\left[n\right]$

Wherein ε is unknown discrete-time frequency skew, and Δ is unknown vertical shift, and h [l] is unknown discrete-time channel coefficient, and v [n] is additional noise.In order to explain with the key idea in lower part, ignore the existence of additional noise.

I. rough frame synchronization

The object of rough frame synchronization solves unknown vertical shift Δ.Let us is made giving a definition:

r ₁[n]：＝[r[n]，r[n+1]，...，r[n+N _t-1]] ^T，

${\stackrel{\&OverBar;}{r}}_1\left[n\right]:={\left[r\right[n+L_{\mathrm{cp}}],r[n+1],...,r[n+N_t-1\left]\right]}^T,$

r ₂[n]：＝[r[n+N _t]，r[n+1+N _t]，...，r[n+2N _t-1]] ^T，

${\stackrel{\&OverBar;}{r}}_2\left[n\right]:={\left[r\right[n+L_{\mathrm{cp}}+N_t],r[n+1+L_{\mathrm{cp}}+N_t],...,r[n+{L_{\mathrm{cp}}+2N}_t-1\left]\right]}^T,$

r ₃[n]：＝[r[n+2N _t]，r[n+1+2N _t]，...，r[n+3N _t-1]] ^T，

${\stackrel{\&OverBar;}{r}}_3\left[n\right]:={\left[r\right[n+L_{\mathrm{cp}}+2N_t],r[n+L_{\mathrm{cp}}+1+{2N}_t],...,r[n+{L_{\mathrm{cp}}+3N}_t-1\left]\right]}^T,$

r ₄[n]：＝[r[n+3N _t]，r[n+1+3N _t]，...，r[n+4N _t-1]] ^T，

${\stackrel{\&OverBar;}{r}}_4\left[n\right]:={\left[r\right[n+L_{\mathrm{cp}}+3N_t],r[n+L_{\mathrm{cp}}+1+{3N}_t],...,r[n+{L_{\mathrm{cp}}+4N}_t-1\left]\right]}^T,$

The rough frame synchronization algorithm proposed is from K.Shi and E.Serpedin, " Coarse frame andcarrier synchronization of OFDM systems:a new metric and comparison; " IEEETrans.Wireless Commun., vol.3, no.4, algorithm in pp.1271-1284, July 2004 takes a hint, and obtains according to maximum-likelihood criterion.

The rough frame synchronization that method 1-improves: rough frame synchronization estimator solves following optimization:

$\widehat{\mathrm{\&Delta;}}=\arg \underset{k\&Element;Z}{\max }\frac{\left|P_1\left(k\right)\right|+\left|P_2\left(k\right)\right|+\left|P_3\left(k\right)\right|}{{\left|\right|r_1\left|\right|}^2+{\left|\right|r_2\left|\right|}^2+{\left|\right|r_3\left|\right|}^2+{\left|\right|r_4\left|\right|}^2+\frac{1}{2}({\left|\right|{\stackrel{\&OverBar;}{r}}_1\left|\right|}^2+{\left|\right|{\stackrel{\&OverBar;}{r}}_2\left|\right|}^2+{\left|\right|{\stackrel{\&OverBar;}{r}}_3\left|\right|}^2+{\left|\right|{\stackrel{\&OverBar;}{r}}_4\left|\right|}^2)}$

Wherein,

$P_1\left[k\right]=r_1^{*}\left[k\right]r_2\left[k\right]-r_3^{*}\left[k\right]r_4\left[k\right]-{\stackrel{\&OverBar;}{r}}_2^{*}\left[k\right]{\stackrel{\&OverBar;}{r}}_3\left[k\right]$

$P_2\left[k\right]=r_2^{*}\left[k\right]r_4\left[k\right]-r_1^{*}\left[k\right]r_3\left[k\right]$

$P_3\left[k\right]={\stackrel{\&OverBar;}{r}}_4^{*}\left[k\right]{\stackrel{\&OverBar;}{r}}_4\left[k\right],$

The signal be corrected is defined as:

Other correction term is used to the little inceptive impulse in compensate for channel and can be conditioned based on application.This extra delay will be included in the channel afterwards.

Ii. fractional frequency offset corrects

Fractional frequency offset corrects after rough frame synchronization block.

The fractional frequency offset that method 2-improves corrects: fractional frequency offset is solution below:

This is known as fractional frequency offset, because algorithm only can correcting offset

$\left|{\hat{e}}_f\right|<\frac{1}{2N_t}$

This problem will be solved in next part.Fine frequency offset correction signal is allowed to be defined as:

$r_f\left[n\right]=e^{-j2\mathrm{\&pi;}{\widehat{\mathrm{\&epsiv;}}}_f}{r}_c\left[n\right]$

Note, method 1 and 2 is for the good K.Shi that works in frequency selective channel, E.Serpedin, " Coarse frame and carrier synchronization of OFDM systems:a newmetric and comparison; " IEEE Trans.Wireless Commun., vol.3, no.4, the improvement of pp.1271-1284, July 2004.Here one especially innovation be the use of r recited above and use improve former estimator, because it have ignored the sampling that is affected due to internal symbol interference.

Iii. integer frequency deviation corrects

In order to correct integer frequency deviation, be necessary to write an equivalent system model for signal received after fine frequency offset correction.The timing error of reservation is absorbed in channel, is not had noisy received signal to have following structure:

$r_f\left[n\right]=e^{j2\mathrm{\&pi;}\frac{\mathrm{nk}}{N_s}}\underset{l=0}{\overset{L_{\mathrm{cp}}}{\mathrm{\&Sigma;}}}g\left[l\right]s[n-l]$

Wherein n=0,1 ..., 4N _t– 1.Integer frequency deviation is k, and the equivalent channels of the unknown is g [l].

The integer frequency deviation that method 3-improves corrects: integer frequency deviation is following solution:

$\hat{k}=\arg \underset{m=\mathrm{0,1},...,N_t-1}{\max }{r*D\left[k\right]S(S*S)}^{-1}S*D\left[k\right]*r$

Wherein:

r＝D[k]Sg

$D\left[k\right]:=\mathrm{diag}\{1,e^{j2\mathrm{\&pi;}\frac{n1}{N_t}},...,e^{j2\mathrm{\&pi;}\frac{n(4N_t-1)}{N_t}}\}$

$S:=\left[\begin{array}{ccccc}s\left[0\right]& s[-1]& ...& ...& s[-L_{\mathrm{cp}}]\\ s\left[1\right]& s\left[0\right]& s[-1]& ...& s[-L_{\mathrm{cp}}+1]\\ s[4N_t-1]& s[4N_t-2]& s[4N_t-3]& ...& s[4N_t-1-L_{\mathrm{cp}}]\end{array}\right]$

$g:=\left[\begin{array}{c}g\left[0\right]\\ g\left[1\right]\\ .\\ .\\ .\\ g\left[L_{\mathrm{cp}}\right]\end{array}\right]$

These give the estimation of total frequency shift (FS):

$\hat{e}=\frac{\hat{k}}{N_t}+{\hat{e}}_f$

In fact, method 3 has very high complexity.In order to reduce complexity, following observation can be made.First, product S (S*S) ^-1s can by precomputation.Regrettably, this still leaves sizable matrix multiplication.Adopt the observation with proposed training sequence alternatively, S*S ≈ I.This generates the method for the support type of following reduction.

The integer frequency deviation of the improvement of method 4-low-complexity corrects:

The integer frequency deviation estimator of low-complexity has solved

$\hat{k}=\arg \underset{m=\mathrm{0,1},...,N_t-1}{\max }(S*D[k]*r)*(S*D[k]*r)$

Iv. result

In the portion, we compare the performance of the different estimators proposed.

First, in Figure 50, we compare the amount of the expense required for often kind of method.Notice that expense is reduced 10 times to 20 times by two kinds of new methods.In order to the performance of more different estimators, MonteCarlo experiment is performed.The setting considered sends waveform from our common NVIS of the linear modulation structure with 3K symbol symbol rate per second, and corresponding to the pass band width of 3KHz, and the cosine impulse risen is shaped.Realize for each Monte Carlo, frequency shift (FS) is from [-f _max, f _max] on be uniformly distributed and generate.

There is f _maxthe little frequency shift (FS) of=2Hz and do not have the emulation of integer offset correction to be illustrated in Figure 51.Can find out that there is N from this Performance comparision _t/ M _tthe performance of=1 is slightly demoted from original estimator, although the expense of substantially reducing.There is N _t/ M _tthe performance of=4 is better, is almost 10dB.Due to the error in integer bias estimation, all curves experienced by complications at low SNR point.Little error in integer skew can create large frequency error and large splicing square error.Integer offset correction can be closed to improve performance in little skew.

Deposit in case at multi-path channel, the performance of frequency offset estimator generally reduces.But, in Figure 52, close integer offset estimator and present extraordinary performance.Therefore, in multi-path channel, the fine correction algorithm of the improvement after the rough correction performing robust is prior.Note, there is N _t/ M _tthe offset behavior of=4 is much better in multipath situation.

The various steps that embodiments of the present invention propose above can comprising.Described step can realize in the mode of machine-executable instruction, and described instruction makes universal or special processor perform particular step.Such as, the various assemblies in base station/AP recited above and customer set up may be implemented as the software performed on universal or special processor.In order to avoid the fuzzy parties concerned of the present invention, such as the various known individual calculus thermomechanical components of computer storage, hard disk, input unit etc. is saved from figure.

Alternatively, in one embodiment, shown here various functional module and correlation step can by comprising the specialized hardware components (such as application-specific integrated circuit (ASIC) (ASIC)) of the hardwired logic performing step or being performed by the combination in any of programmed computer components and custom hardware components.

In one embodiment, the particular module of all codings as described above, modulation and signal processing logic 903 can be implemented on programmable digital signal processor (DSP) (or DSP group), described DSP such as uses the DSP(of the TMS320x framework of Texas Instrument (Texas Instruments) such as, TMS320C6000, TMS320C5000 etc.).DSP in this embodiment can be embedded in the package card of personal computer, such as, and pci card.Certainly, when general principle according to the invention, various different DSP framework can be used.

Various parts of the present invention also may be provided in the machine readable media for storing machine executable instruction.Machine readable media can include but not limited to flash memory, CD, CD-ROM, DVD ROM, RAM, EPROM, EEPROM, magnetic card or optical card, communication media or be suitable for the machine readable media of other type of store electrons instruction.Such as, the present invention can be downloaded as computer program, this computer program can be sent to requesting computer (such as client) via communication link (such as, modulator-demodulator or network connect) from remote computer (such as server) by the mode being included in the data-signal of carrier wave or other communication media.

Throughout aforementioned description, in order to the object explained, many specific detail are suggested to the complete understanding providing native system and method.But, it will be apparent to one skilled in the art that system and method can when do not have in these specific detail some be implemented.Therefore, scope of the present invention and essence should be judged according to claims.

In addition, in the foregoing written description, many documents are cited to provide of the present invention and more fully understand.All these lists of references quoted are by reference to being integrated in the application.

Claims (29)

1. a multi-user multi-aerial system MU-MAS, comprising:

One or more centralized unit or base station, be coupled to multiple distributed transceiver station or antenna via network service;

Described network, by wired or wireless link or the two be bonded, is used as backhaul communication channel;

N number of data stream is become the data flow of M precoding by described centralized unit, and the data flow of each precoding is the combination of some or all of N number of data flow;

The data flow of a described M precoding is sent to described distributed transceiver station on the network;

Described distributed transceiver station sends the data flow of described precoding on wireless links at least one customer set up simultaneously, receives at least one in original N number of data flow to make at least one customer set up.

2. system according to claim 1, comprise a kind of wireless client device used in systems in which for the frequency and phase deviation compensating MU-MAS communication, this wireless client device comprises:

One or more RF unit, for receiving the signal transmitted from one or more MU-MAS transmitter unit, and is transformed into base band by described signal down;

One or more modulus A/D converting unit, for receive frequency reducing conversion after signal and this signal from analog signal is converted to digital signal;

Frequency/phase bias estimation/compensating unit, estimated frequency and/or phase deviation and by this information feed back to transmitter for precompensation;

One or more orthogonal frequency division multiplex OFDM unit, for removing Cyclic Prefix and performing fast fourier transform FFT to report the signal in frequency domain in described digital signal;

Channel estimating unit, receives the signal exported from described one or more OFDM unit and also responsively calculates channel evaluation data during cycle of training; And

Feedback generator unit, for described channel evaluation data or training being sent to base station with carrying out signal using in precoding, carrying out precoding to signal and carrying out before signal is transferred to described wireless client device.

3. system according to claim 2, wherein channel estimating is calculated in the time domain by using the input of described OFDM unit.

4. system according to claim 2, wherein said feedback generator unit also comprises logic, and this logic is used for quantizing frequency shift (FS) and channel estimating before frequency shift (FS) and channel estimating are transferred to described base station.

5. system according to claim 2, this system also comprises: receiver unit, and this receiver unit receives the output from described OFDM unit, and responsively calculates receiver and carry out demodulate/decode to obtain the estimation to transmitted data to signal.

6. system according to claim 5, wherein said receiver unit is that least mean-square error MMSE receiver, zero forces ZF receiver, maximum likelihood ML or MAP receiver.

7. system according to claim 2, wherein said MU-MAS communication comprises distributed input distributed output DIDO and communicates, and wherein said one or more RF unit receives the signal from one or more DIDO transmitter unit transmission and this signal down is transformed into base band.

8. system according to claim 2, wherein said frequency/phase bias estimation/compensating unit performs rough frame synchronization, fractional frequency offset corrects, integer frequency deviation corrects and/or integer frequency deviation corrects.

9. use the unbalanced wireless client device of inphase quadrature I/Q for compensating multi-user multi-aerial system MU-MAS communication in systems in which, this wireless client device comprises:

One or more RF unit, for receiving the signal transmitted from one or more MU-MAS transmitter unit, and is transformed into base band by described signal down;

One or more modulus A/D converting unit, for receive frequency reducing conversion after signal and this signal from analog signal is converted to digital signal;

One or more orthogonal frequency division multiplex OFDM unit, for removing Cyclic Prefix and performing fast fourier transform FFT to report the signal in frequency domain in described digital signal;

I/Q channel perception estimation unit, receives the signal exported from described one or more OFDM unit and also responsively calculates channel evaluation data during cycle of training; And

Feedback generator unit, for described channel evaluation data or training being sent to base station with carrying out signal using in precoding, carrying out precoding to signal and carrying out before signal is transferred to described wireless client device.

10. wireless client device according to claim 9, wherein channel estimating is calculated in the time domain by using the input of described OFDM unit.

11. wireless client device according to claim 9, wherein said feedback generator unit also comprises logic, and this logic is used for quantizing channel estimating before channel estimating is transferred to described base station.

12. wireless client device according to claim 9, this wireless client device also comprises:

I/Q perception receiver unit, receives the output from described OFDM unit, and responsively calculates IQ receiver and carry out demodulate/decode to obtain the estimation to transmitted data to signal.

13. wireless client device according to claim 12, wherein said I/Q perception receiver unit is that least mean-square error MMSE receiver, zero forces ZF receiver, maximum likelihood ML or MAP receiver.

14. wireless client device according to claim 12, wherein said I/Q perception receiver unit comprises MMSE or ZF filter to eliminate the uneven intercarrier interference caused of the I/Q adjusted by mirror image.

15. wireless client device according to claim 12, wherein said I/Q perception receiver unit comprise nonlinear detector (i.e. ML) come joint-detection mirror image adjust on symbol.

16. wireless client device according to claim 12, wherein said I/Q channel perception estimation unit calculate I/Q perception receiver unit can coefficient to remove intercarrier interference.

17. 1 kinds of wireless client device using the communication characteristic for dynamically adapting multi-user multi-aerial system MU-MAS communication system in systems in which, this wireless client device comprises:

One or more RF unit, for receiving the signal transmitted from one or more MU-MAS transmitter unit, and is transformed into base band by described signal down;

One or more modulus A/D converting unit, for receive frequency reducing conversion after signal and this signal from analog signal is converted to digital signal;

One or more orthogonal frequency division multiplex OFDM unit, for removing Cyclic Prefix and performing fast fourier transform FFT to report the signal in frequency domain in described digital signal;

Channel estimator, receives the signal exported from described one or more OFDM unit and also responsively calculates link-quality matrix during cycle of training; And

Feedback generator unit, for described link-quality matrix being sent to base station with carrying out signal using during modulation/coding, precoding and user select, modulation/coding, precoding and user being carried out to signal and selects to carry out before signal is transferred to described wireless client device.

18. wireless client device according to claim 17, wherein channel estimating is calculated in the time domain by using the input of described OFDM unit.

19. wireless client device according to claim 17, wherein said feedback generator unit also comprises logic, and this logic is used for quantizing channel estimating and/or described link-quality matrix before channel estimating and/or described link-quality matrix are transferred to described base station.

20. wireless client device according to claim 17, this wireless client device also comprises:

Receiver unit, receives the output from described OFDM unit, and responsively carries out demodulate/decode to obtain the estimation to transmitted data to signal.

21. wireless client device according to claim 20, wherein said receiver unit is that least mean-square error MMSE receiver, zero forces ZF receiver, maximum likelihood ML or MAP receiver.

22. 1 kinds of methods implemented in multi-user multi-aerial system MU-MAS, the method comprises:

One or more centralized unit or base station are coupled to multiple distributed transceiver station or antenna via network service, and described network, by wired or wireless link or the two be bonded, is used as backhaul communication channel;

At centralized unit, N number of data stream is become the data flow of M precoding, the data flow of each precoding is the combination of some or all of N number of data flow;

On the network by the data flow of a described M precoding extremely described distributed transceiver station;

The data flow of described precoding is sent at least one customer set up from described distributed transceiver station on wireless links simultaneously, receives at least one in original N number of data flow to make at least one customer set up.

23. according to method described in claim 22, and the method also comprises:

Receive at one or more RF unit of wireless client device the signal transmitted from one or more MU-MAS transmitter unit, and described signal down is transformed into base band;

Described signal from analog signal is converted to digital signal;

Estimated frequency and/or phase deviation and by this information feed back to described transmitter for precompensation;

Remove Cyclic Prefix and in described digital signal, perform fast fourier transform FFT to report the signal in frequency domain;

During cycle of training, receive the signal of generation and responsively calculate channel evaluation data; And

Described channel evaluation data or training being sent to base station with carrying out signal using in precoding, precoding being carried out to signal and carried out before signal is transferred to described wireless client device.

24. methods according to claim 23, wherein said channel estimating is calculated in the time domain by using the input of orthogonal frequency division multiplex OFDM unit.

25. methods according to claim 23, wherein said feedback generator unit also comprises logic, and this logic is used for quantizing frequency shift (FS) and channel estimating before frequency shift (FS) and channel estimating are transferred to described base station.

26. methods according to claim 23, the method also comprises:

Receive the signal that produces and correspondingly this signal of demodulation code to obtain the estimation of data transmitted.

27. methods according to claim 26, wherein said RF unit is that least mean-square error MMSE receiver, zero forces ZF receiver, maximum likelihood ML or MAP receiver.

28. methods according to claim 23, wherein said MU-MAS communication comprises distributed input distributed output DIDO and communicates, and wherein said one or more RF unit receives the signal transmitted from one or more DIDO transmitter unit, and this signal down is transformed into base band.

29. methods according to claim 23, the method also comprises:

Perform rough frame synchronization, fractional frequency offset corrects, integer frequency deviation corrects and/or integer frequency deviation corrects.