CN1179856A - Apparatus and method for adaptive beamforming in antenna array - Google Patents

Abstract

An apparatus and a method for receiving and transmitting information from an array of adaptive antenna elements, wherein a predictive filtre (68) supplies an estimate of received signal samples (*(n)) likely to be received in a burst immediately preceding a transmission. Combination (70) of this estimate (*(n)) with received signal samples (x) obtained from actual (historically received) signals, received over a predetermined number of frames, yields estimates of optimum beamforming coefficients wopt for application (74) to data (76) for transmission from an adaptive array of antenna elements (41). As such, available processing time for obtaining the beamforming coefficents is increased.

Description

The device and the method that are used for adaptive beamforming in the antenna array

The communication system that the present invention relates generally to, it especially can be used for adopting the communication system of adaptive beamforming technology.

In communication system (frequency division multiplexing (FDM) system particularly, such as pan-European digital cellular global system for mobile communications (GSM) communication and code division multiple access (CDMA) system alternately) in, the use of adaptive antenna (AA) is tempting day by day, reason is that this adaptive antenna has improved systematic function comprehensively, particularly on (business) capacity is handled.Be appreciated that in adaptive antenna system, can obtain high-precision wave beam by phase place and amplitude (amplitude) component that accurately changes transmitted wave.Or rather, the phase place and the amplitude of the one group of ripple that emits from the antenna array of transceiver change along with " weighting " of the individual element in the antenna array, thus (as certain base station) antenna radiation pattern and normal signal and the relevant area of coverage as the interface environment of a sub-district automatically (optimizations) mate.

In the self adaptation transmit beam-forming in duplex communication system, require beam shaping coefficient (i.e. " weighting " factor) to adjust by responding previous receiving channel information, these reception information produce in the up link of system or down link.In fact, when certain GSM base station of specific consideration, the beam shaping coefficient of certain business model must be tried to achieve at four time slot duration (just 4 * 15/26 milliseconds of (ms), i.e. 2.3ms), yet the time of calculating in practice, the beam shaping coefficient in mobile unit may be shorter.Unfortunately, when considering to calculate the desired treating capacity of (estimation) these beam shaping coefficients, the limited time has seriously been limited obtainable precision.In fact, receive according to signal, the information (during the typical case) in the signal of being included in must be sampled, storage and then demodulation (by synchronous and equilibrium treatment).In addition, must generate transmission weight, before this data load and modulation, be applied to the data that this is used to transmit then from received signal.

In addition, because the inherent shortcoming that this beam shaping mechanism causes, the finite time that can be used for handling is further suffered damage, these problems mainly from: (i) beam shaping coefficient (weight) is (up link and down link normally are operated in different frequency, thereby require to carry out frequency inverted and phase error correction) of frequency dependence; And the time correlation that (ii) in channel circumstance, causes fluctuation by mobile unit and fixed base stations relative motion.The back a bit on, the time change the influence that brings can by as ask average method to alleviate to a certain degree to the time slot weight that receives, but to revise be very coarse the time of this form.

In view of the beam shaping coefficient that (can be understood) in the exemplary communications systems is selected, the Weiner solution has proposed a kind of optimized choice (yes be used to revise between up link and the down link frequency difference), for: $w_{\mathrm{upt}}=R_{\mathrm{xx}}^{-1}{r}_{\mathrm{xd}}------\left(1\right)$

Wherein:

I) x=[x ₁, x ₂... x _(n-1), x _(n-2),] ^TIt is the signal phasor that receives at n branch road (n antenna element just);

Ii) w _Opt=[w ₁, w ₂... w _(n-1), w _(n-2)] ^TIt is the optimization weight vector of these n branch road;

Iii) r _Xd=E[x ^*S] be the relevant of signal phasor that receives and the signal phasor s that needs, the signal phasor that needs is sent out during definite training sequence of a pulse train;

Iv) R _XxBe the signal cross-correlation matrix that receives, equal E[x ^*x ^T];

V) R _Xx ^-1The inverse matrix R of representing matrix _Xx

Vi) x ^*It is the complex conjugate of x;

Vii) T is a vector transfer function, and its row is replaced by row, and vice versa; And,

Viii) E[.] expression one desired value.

Essential beam shaping coefficient must be estimated from received signal to calculate a continuous information frame, because correlation matrix R _XxAnd r _XdCan not directly obtain (because of these correlation matrixes of there is no telling before receiving the signal relevant) with these matrixes.In this, estimation (representing with a horizontal line) utilizes following formula, is suitable for calculating the approximated weights of a subsequent frame (n+1): $\stackrel{-}{R_{\mathrm{xx}}}(n+1)=\frac{1}{B}\underset{k=n-B+1}{\overset{n}{\mathrm{\Σ}}}{x}^{*}\left(k\right)x^T\left(k\right)-----\left(2\right)$

Wherein B be the sampling section (as pulse train) considered in each estimation number (in some cases, every frame comprises the more than pulse train of possibility), its explanation can consult J.H.Winters's " Signal Acquisition and Tracking with Adaptive Arrays in theDigital mobile Radio System IS-54 with Flat-Fading ", be published on the IEEE Transactions on Vehicular Technology 377-384 page or leaf in November 42 (4) in 1993.Equally, the estimation of correlation matrix is based on the actual reception signal.

Equally, generally need reliable improved mechanism (mainly raising the efficiency) to come the compute beam form factor.

Be used to receive and send the device of information from an adaptive antenna unit battle array, this device comprises storage device, be used to preserve reception information, it is characterized in that a predictive filter, it is by response reception information, be used for estimating and at least once transmitting the information that may receive in the future by this device to this device, and merge previous reception information and information of forecasting in order to produce the device of beam shaping coefficient, the beam shaping coefficient is used for the information that weighting is sent by this adaptive antenna unit paroxysm subsequently, so just allow at least once transmission in the future of this device, before this device receives information, the compute beam form factor.

Second of the present invention is that it provides a kind of and receives in the device that comprises an adaptive antenna unit battle array and the method for the information of transmission, this method comprises the step of storage reception information, it is characterized in that following steps:, to the transmission in future at least once of this device, estimating the information that may be received by this device by response reception information; And be used to merge previous reception information and information of forecasting to produce the beam shaping coefficient, be used for the information that weighting is sent by this adaptive antenna unit paroxysm subsequently, allow whereby at least once transmitting in the future to this device, before this device receives information, the compute beam form factor.

To be described illustrated embodiment of the present invention in conjunction with the accompanying drawings now.

What Fig. 1 represented is the duplex communication channel that adopts prior art.

Fig. 2 represents is the present invention when realizing in conjunction with the duplex communication channel among Fig. 1, the regularly corresponding improvement that obtains.

Fig. 3 represents to be used for the mechanism of adaptive beamforming and the function diagram of device (corresponding to preferred implementation of the present invention).

Referring to Fig. 1, represented is adopts the duplex communication channel 10 of prior art, it comprise a plurality of frame 12-18 (in this special case, for for purpose of brevity only illustration four frames).Each frame is divided into 8 and separates time slot t ₀-t ₇(yet the duration of system's time slot corresponding number and each time slot may be different).Be appreciated that this duplex communication channel 10 can be a Traffic Channel (TCH) or Broadcast Control Channel (BCCH), the obvious difference of these multi-form channels is in BCCH, distributes a special time slot (t normally at least ₀) be used for system control purpose.If we regard duplex communication channel 10 as a TCH, time slot t so ₀Usually be used as down link, and time slot t ₃Will be as corresponding up link.Remaining time slot will be assigned with in a similar fashion/match.Therefore, in this example, each frame 12-18 produces the buffering of two time slots between downlink transmission and up link reception; And, as top explanation, produce four time slot (t between up link reception in consecutive frame and the downlink transmission ₄-t ₇) buffering 20.Obviously, under the mobile unit situation, cushion just opposite.

According to equation 2, each burst transmission can (from a training sequence the position cross-correlation, training sequence can be as the middle rank in GSM jogging (mid-amble) sequence) push away the received signal vector x (k) of a frame k, the quantity of the pulse train that each simultaneously estimation requires is according to R _XxThe prediction rate of change do corresponding the adjustment.Yet equation 2 will be used x (n), and therefore available time restriction is between the message pick-up and transmission of the communicator such as base station or mobile unit.

Preferred implementation of the present invention is to utilize linear prediction filtering, the estimation of the received received signal sample value x of possibility (n) in the last transmission burst is provided, and with this estimation and the received signal sample value merging that obtains from actual (before having received) signal, these actual signals are from (being scheduled to) number pulse train or frame reception arbitrarily, as three frames.Be appreciated that linear prediction filtering can represent in order to following equation: $\stackrel{-}{x}\left(n\right)-\underset{m=n-M}{\overset{n-1}{\mathrm{\Σ}}}{a}_m^{T}x\left(m\right)------\left(3\right)$

Wherein:

I) a _mBe that the filter coefficient vector that obtains with technology well known to those of ordinary skill in the art is (referring to reference book " the Adaptive Filter Theory " second edition that Simon Haykin is shown, New Jersey, U.S.A.; Prentice-Hall, the method for optimized choice am among the 1986.ISBN:0-13-01326-5);

Ii) M is a linear prediction filter length;

Iii) m is the integer label;

Iv) n is a present frame.

Therefore, the estimation with the corresponding correlation matrix of the preferred embodiment for the present invention utilizes following formula to realize: $\stackrel{-}{R_{\mathrm{xx}}}(n+1)=\frac{1}{B}(\underset{k=n-B+1}{\overset{n-1}{\mathrm{\Σ}}}{x}^{*}\left(k\right)x^T\left(k\right)+\stackrel{-}{x^{*}}\left(n\right)\stackrel{-}{x^T}\left(n\right))---\left(4\right)$

Therefore mechanism of the present invention allow the beam shaping coefficient before pulse train receives, just calculated (because the effect of signals that had before received is to beam shaping coefficient of back), for example be at time slot t in the base station scenario of Fig. 1 ₃Before.Thereby between the reception of data and transmission, obtained the extra slots that can be used for handling, therefore the buffering 30 that increases is provided.Increase later buffering as shown in Figure 2, in conjunction with the corresponding processing time of duplex communication channel of Fig. 1, the benefit of the timing that embodiments of the present invention are brought among the figure as can be seen.Be appreciated that the buffering 30 after the increase can be an entire frame or more, but it is the time between the pulse train (it may produce) of last received pulse string and linear prediction filter estimation at least in same frame.

Although predictive filtering itself requires to handle in the microprocessor (or similar device) of communicator, both allow to adopt more complex decoding and beam shaping algorithm (latter generally can improve the resolution and the precision of beam shaping in communication system) extra time that offers communicator, also allowed to use the processor of slower (thereby more cheap).And the extra process that requires to carry out in communicator can be optimized by used pulse train B in the restriction estimation.For simplicity, to calculating

Mechanism done detailed description, although should understand for estimation

Preferably adopt identical mathematical method, but will do suitable replacement, be about to x ^TOr

Become s ^T

By to each weighting in the equation 4, basic thought of the present invention can further be developed, and weighted factor is corresponding to estimating R _XxRate of change, thereby it is predictable that the correlation matrix estimation has been become itself.This can be expressed as follows on mathematics: $\stackrel{-}{R_{\mathrm{xx}}}(n+1)=\frac{1}{B}(\underset{k=n-B+1}{\overset{n-1}{\mathrm{\Σ}}}c(k-n+B)x^{*}\left(k\right)x^T\left(k\right)+c\left(B\right)\stackrel{-}{x^{*}}\left(n\right)\stackrel{-}{x^T}\left(n\right))---\left(5\right)$

Wherein numerical value is gathered c=[c (1), c (2) ... c (B)] ^TEstimate in advance,, make estimation error minimum (between a mobile unit and a fixed base stations, measuring) by by virtue of experience measuring the point that receives data from the overlay area.Therefore, correlation matrix R is considered in this prediction weighting _XxActual rate of change.Equally, the introducing of coefficient c provides relative weighting for the progression item in the equation 5, makes R _XxMinimum is out of one's reckoning.

Referring now to Fig. 3,, it is a kind of mechanism of adaptive beamforming and function diagram (corresponding to preferred implementation of the present invention) of device 40 of being used for.Device 40 is communicators, for example is to comprise base station or the mobile unit (suitable) that receives and send the antenna element battle array 41 of code signal 42.This antenna element battle array 41 is coupled to a duplexer battle array 44, and it is used for selectively receiver array 46 or transmitter array 48 being coupled to antenna element battle array 41.

Receive path one, the information orientation (bearing) signal (being x) that is received and handled by receiver array 46 by antenna element battle array 41 is coupled to buffer 49 by an analog to digital converter 50.This buffer design becomes can preserve at least B pulse train.Be kept at data x in the buffer 49 and be input in the correlation matrix estimation device 52, correlation matrix estimation device 52 also responds the register 54 of the storage replication that comprises training sequence s.This correlation matrix estimation device 52 is given R _XxAnd r _XdProvide numerical value (according to equation 2) with as response to x and s.Power calculator 56 receives R _XxAnd r _XdThe computing of finishing equation 1 is to generate w _OptValue (just should receive the beam shaping coefficient of path), these values act on the sample value of the buffer 49 from beam-shaper 58 respectively.Output from beam-shaper 58 is coupled to demodulator 60, and demodulator 60 provides a decoded output signal 62 to output device 64 conversely again, for example is Voice decoder or visual display unit (VDJ).

At a transmission path, it is relevant with previous frame to be stored in the buffer 49 data, and it is imported into according to equation 3 and calculates in the signal estimation device 68 of x.The data x that is stored in the buffer 49 also is input in the correlation matrix estimation device 70 (it does further response to x, and it also responds the duplicating of training sequence s that is stored in the register 54), finishes equation 4 or equation 5 by it, generates

With

A secondary power calculator 72 (it may be a power calculator 56) receives

With

Finish equation 1 to generate w _OptValue (being used for this transmission path), they are used for the data 76 of coming from such as input units such as modulator-demodulator or keyboards in beam-shaper 74 (it may be a beam-shaper 58).Give modulator 80 arrays from an output coupling of beam-shaper 74, and modulator array 80 provides conversely coding output signal 82 to reflector 48, finally offers antenna unit array 41 by duplexer array 44.

Be appreciated that correlation matrix estimation device 52 and 70, power calculator 56 and 72, beam-shaper 58 and 74 and the typical case of signal estimation device 68 be implemented in the microprocessor 90, and register 54 can be positioned at the inside (as shown in the figure) or the outside of this microprocessor 90.

The information that is received during this pulse train by this communicator for example can be data or encoded voice.In detail, under the data occasion, have a plurality of frames when beginning to communicate by letter and be cushioned to allow to send accurately beam shaping.Yet, under the voice communication situation, being necessary to begin communication with the beam shaping coefficient of estimation gained and the comprehensive pattern of overlay area, optimizing initial weighted factor whereby, may be to introduce mechanism of the present invention after at least one burst transmission is received the time the earliest then.

Although the present invention describes in conjunction with the pan-European digital cellular communication system of GSM, the present invention can be applicable to any bilateral system, comprises that those use Time Division Multiplexing agreement, sound wave and duplex system.And embodiments of the present invention can be in certain mobile unit, or control in the base station of a plurality of mobile units at certain.

Certainly can understand, the present invention has only provided example at this, can do the modification on the details within the scope of the present invention, before back to back burst transmission, may surpass a frame as predictive filtering technology (use of this predictive filtering technology can be in conjunction with the Data Receiving of reality, and need not to be confined to the linear prediction filtering described in the illustrated embodiment of the present invention).Like this, although the processing time can increase, precision can corresponding improve.

Claims (10)

1. be used for from the device (40) of certain adaptive antenna unit battle array (41) reception and transmission information (42), this device comprises that storage device (49) to be used to storing reception information (x), is characterized in that:

One predictive filter (68) in response to the information that receives, may be installed the information of forecasting that receives by this in estimation at least once transmission in the future of this device;

Be used to merge the device (70) of previous reception information (x) and this information of forecasting (x), to generate beam shaping coefficient (w _Opt), be used for the information (76) that weighting is sent by adaptive antenna unit battle array (41) subsequently, thereby allow at least once transmission in the future of this device, before this device receives information, the compute beam form factor.

2. according to claim 1, wherein predictive filter (68) is a linear filter, and form is as follows: $\stackrel{-}{x}\left(n\right)=\underset{m=n-M}{\overset{n-1}{\mathrm{\Σ}}}{a}_m^{T}x\left(m\right)$

Wherein:

I) x is that forecasting institute gets information;

Ii) a _mIt is the filter coefficient vector;

Iii) x (m) is a reception information;

Iv) T is the vector transfer function, and its row is replaced with row, and vice versa;

V) M is a linear prediction filter length;

Vi) m is the integer label;

Vii) n is a present frame.

3. according to the device of claim 2, the device (70) that wherein is used to merge comprises correlation matrix estimation device, and it estimates that according to following formula forecasting institute gets the correlation matrix between information (x) and the reception information (x): $\frac{1}{B}(\underset{k=n-B+1}{\overset{n-1}{\mathrm{\Σ}}}{x}^{*}\left(k\right)x^T\left(k\right)+\stackrel{\text{-}}{x^{*}}\left(n\right)\stackrel{-}{x^T}\left(n\right))$

Wherein:

I) x=[x ₁, x ₂... x _(n-1), x _(n-2)] T is the received signal vector in adaptive antenna unit battle array;

Ii) x ^*It is the complex conjugate of x; And

Iii) B is the sample value quantity of being considered in each estimation partly.

4. according to the device of claim 2, the device (70) that wherein is used to merge comprises correlation matrix estimation device, and it estimates that according to following formula forecasting institute gets the correlation matrix between information (x) and the reception information (x): $\frac{1}{B}(\underset{k=n-B+1}{\overset{n-1}{\mathrm{\Σ}}}c(k-n+B)x^{*}\left(k\right)x^T\left(k\right)+c\left(B\right)\stackrel{-}{x^{*}}\left(n\right)\stackrel{-}{x^T}\left(n\right))$ Wherein:

I) x=[x ₁, x ₂... x _(n-1), x _(n-2),] ^TIt is the received signal vector in adaptive antenna unit battle array;

Ii) x ^*It is the complex conjugate of x;

Iii) B is the sample value quantity of being considered in each estimation partly;

Iv) c is constant set [c (1) ... c (B)], the prediction rate of change of correspondence and correlation matrix.

5. according to the device of aforementioned claim, wherein the reception of information and transmission are in pulse train.

6. according to the device of claim 5, wherein pulse train is in a time-division multiplexing (TDM) communication system.

7. according to the device of claim 5 and 6, wherein reception information is that the pulse train from predetermined quantity obtains.

8. according to the device of any claim in front, wherein this device is a base station.

9. according to any one device of claim 1 to 7, wherein this device is a mobile unit.

10. one kind is used for comprising the step of storage (49) reception information (x) in device (40) reception that comprises an adaptive antenna unit battle array (41) and the method for the information of transmission, it is characterized in that following steps:

In response to reception information (x), at least once transmission in the future of this device, estimation (68) may be by the prediction (x) of this device reception; And,

Merge (70) and before received information and this information of forecasting, to produce the beam shaping coefficient, be used for the information that weighting is sent by this adaptive array subsequently, thereby permission information is at least once transmission in the future of this device, before this device receives information, the compute beam form factor.