WO2008113216A1 - Méthode d'évaluation d'un canal - Google Patents

Méthode d'évaluation d'un canal Download PDF

Info

Publication number
WO2008113216A1
WO2008113216A1 PCT/CN2007/000916 CN2007000916W WO2008113216A1 WO 2008113216 A1 WO2008113216 A1 WO 2008113216A1 CN 2007000916 W CN2007000916 W CN 2007000916W WO 2008113216 A1 WO2008113216 A1 WO 2008113216A1
Authority
WO
WIPO (PCT)
Prior art keywords
sequence
pseudo
channel
training
random sequence
Prior art date
Application number
PCT/CN2007/000916
Other languages
English (en)
French (fr)
Inventor
Guoping Xu
Yu Xin
Jiao Wu
Original Assignee
Zte Corporation
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zte Corporation filed Critical Zte Corporation
Priority to PCT/CN2007/000916 priority Critical patent/WO2008113216A1/zh
Priority to CN200780046701.7A priority patent/CN101578829B/zh
Publication of WO2008113216A1 publication Critical patent/WO2008113216A1/zh

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L25/00Baseband systems
    • H04L25/02Details ; arrangements for supplying electrical power along data transmission lines
    • H04L25/0202Channel estimation
    • H04L25/0212Channel estimation of impulse response
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L25/00Baseband systems
    • H04L25/02Details ; arrangements for supplying electrical power along data transmission lines
    • H04L25/0202Channel estimation
    • H04L25/024Channel estimation channel estimation algorithms
    • H04L25/0242Channel estimation channel estimation algorithms using matrix methods
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L27/00Modulated-carrier systems
    • H04L27/26Systems using multi-frequency codes
    • H04L27/2601Multicarrier modulation systems

Definitions

  • the present invention relates to the field of mobile communications, and in particular to a channel estimation method for a mobile communication system.
  • the method can be applied to a plurality of mobile communication systems, including an OFDM (Orthogonal Frequency Division Multiplexing) system and MIMO- OFDM (Multi-Input Multi-Output Orthogonal Frequency Division Multiplexing) system, and the method can also be applied to CDMA (Code Division Multiple Access) if appropriate frame structure is adopted. Multiple access) Channel estimation of the system.
  • OFDM Orthogonal Frequency Division Multiplexing
  • MIMO- OFDM Multi-Input Multi-Output Orthogonal Frequency Division Multiplexing
  • Orthogonal Frequency Division Multiplexing (OFDM) technology converts high-speed data streams into a set of low-speed parallel data streams, and overlaps the frequency bands of sub-channels. It has strong anti-multipath interference capability and high bandwidth utilization. .
  • Multi-antenna Transceiver (MIMO) systems can achieve greater channel capacity than single-input single-output (SISO) systems under fully scatter channel conditions.
  • SISO single-input single-output
  • a RAKE receiver, a time domain equalization receiver, and a frequency domain equalization receiver for a CDMA system A RAKE receiver, a time domain equalization receiver, and a frequency domain equalization receiver for a CDMA system.
  • Channel estimation techniques can be broadly divided into non-blind estimates and blind estimates, and semi-blind estimates generated on this basis.
  • non-blind estimation is used to obtain better estimation results, and the computational complexity is low, so that the variation of the wireless channel can be better tracked and the receiver performance can be improved.
  • Traditional channel estimation methods often require a large number of matrix operations, which are highly complex and subject to noise.
  • the technical problem to be solved by the present invention is to propose a channel estimation method that can be applied to OFDM (MIMO-OFDM) systems, CDMA systems, and other communication systems.
  • the method reduces the computational burden of channel estimation under the premise of ensuring better channel estimation accuracy.
  • the present invention provides a channel estimation method, which is applied to a mobile communication system, and includes the following steps:
  • the receiving end receives the data of the training sequence, and cross-correlates the received data after the cyclic prefix part is removed from the shifted pseudo-random sequence to obtain a cross-correlation matrix C, which is used as the current training sequence.
  • the pseudo-random sequence is obtained by cyclically shifting 0, 1, ..., L c -1 times, respectively, and the first line of the C matrix is obtained by receiving the data and the pseudo-random sequence as the current training sequence is rotated right by 0 times.
  • the cross-correlation value of the pseudo-random sequence is successively continued.
  • the last row of the C matrix is the cross-correlation value of the pseudo-random sequence obtained by the received data and the pseudo-random sequence which is the current training sequence and rotated rightward by L c -1 times;
  • the method further includes the following steps:
  • the receiving end receives the data corresponding to the training sequence again, and calculates a new cross-correlation matrix C according to the method in step (b), and obtains the current method according to the method in step (c).
  • the estimation of the impulse response of the channel is performed by using iead and tail to obtain an impulse response of the channel corresponding to the signal data segment between the two training sequences.
  • the method further includes the following steps:
  • the receiving end receives the data corresponding to the training sequence again, calculates a new cross-correlation matrix according to the method in step (b), and obtains the current channel according to the method in step (c).
  • the impulse response is estimated to7 , by ii head and the frequency of the corresponding channel Respond to ⁇ and ⁇ . ,.,, Interpolation using ⁇ , ⁇ and ⁇ to7 to obtain the channel frequency response corresponding to the signal data segment between the two training sequences.
  • step (a) the length of a training sequence inserted, which is also the same cyclic prefix length, said step (c) is C p is a fixed matrix, the inverse matrix C P Cp 1 is stored directly in the receiving end.
  • the pseudo-random sequence is a maximum length shift register sequence, that is, an m-sequence.
  • C p has an element on the diagonal of 1 and a non-diagonal element is -1/.
  • the system is an Orthogonal Frequency Division Multiplexing (OFDM) system or a Code Division Multiple Access (CDMA) system.
  • a signal data segment of a certain length separated by two training sequences is a plurality of OFDM symbol data or a plurality of CDMA systems.
  • the system is an OFDM system, wherein a signal data segment of a certain length in the step (a) is a plurality of OFDM symbol data, and in the step (d), ⁇ and i3 ⁇ 4 to 7 are frequencies on each subcarrier of the OFDM at the channel estimation time.
  • the interpolation method in the step is a linear interpolation method, and 0 is an estimated channel frequency response of the jth OFDM symbol data between the two training sequences, and the calculation formula is
  • Num is the number of OFDM symbol data spaced between the two training sequences.
  • the system is a CDMA system, wherein the signal data segment of a certain length in the step (a) is a piece of CDMA data chip data, and in the step (d), ⁇ ⁇ and 0 to , 7 are channel estimation time multipath channels.
  • the frequency response, ⁇ ⁇ ⁇ and ⁇ , ⁇ 7 interpolated to obtain the frequency response of the channel corresponding to the CDMA chip signal between the two training sequences.
  • the system is a CDMA system, wherein a signal data segment of a certain length in the step (a) is a piece of CDMA data chip data, and the step (d) and the step ( 7 ) are impulse responses of the multipath channel at the channel estimation time. , ⁇ d and ii to7 are interpolated to obtain a CDMA chip letter between the two training sequences. The impulse response of the channel corresponding to the number.
  • a multi-antenna transceiver system channel estimation method includes the following steps:
  • each receiving antenna at the receiving end receives the data of the training sequence sent by the current transmitting antenna, and the received data after the cyclic prefix portion is removed is cross-correlated with the shifted pseudo-random sequences to obtain a cross-correlation matrix C, the shifting The bit pseudo-random sequence is obtained by cyclically shifting the pseudo-random sequence as the current training sequence by 0, 1, ..., L c - ⁇ times, respectively, and the first behavior of the C matrix receives data and pseudo-random as the current training sequence.
  • the method further includes the following steps:
  • the receiving antenna receives the data of the same transmitting antenna corresponding to the training sequence again, and calculates a new cross-correlation matrix C according to the method in step (b), according to the method in step (c) get the current impulse response of the channel estimation and using ii, ai, interpolated signal segments between two training sequences corresponding to the channel impulse response, or the use of ⁇ ⁇ ⁇ and ⁇ / channel frequency response corresponding ⁇ ⁇ ⁇ / ⁇ ⁇ ; , interpolating to obtain the channel frequency response corresponding to the signal data segment between the two training sequences.
  • the pseudo-random sequence is a maximum length shift register sequence, that is, an m-sequence.
  • C P has an element on the diagonal of 1 and a non-diagonal element is -1/.
  • the training sequence inserted in the step (a) has the same length and the same cyclic prefix length.
  • C p is a fixed square matrix, and the inverse matrix C- p 1 of the C P is directly stored at the receiving end. .
  • the present invention proposes a new method for channel estimation based on pseudo-random sequences for mobile communication systems, including OFDM (MIMO-OFDM) systems and CDMA systems, and the use of pseudo-random sequence autocorrelation properties reduces the complexity of channel estimation.
  • MIMO-OFDM OFDM
  • CDMA Code Division Multiple Access
  • the channel estimation of the present invention is more accurate than the prior art.
  • the channel estimation method can flexibly adjust the overhead of the training sequence according to the needs of the system transmission rate, so as to obtain a compromise between estimation accuracy and overhead.
  • Figure 1 is a tapped delay line model of a discrete multipath channel
  • FIG. 2 is a flowchart of a channel estimation method based on an OFDM system according to the present invention
  • Figure 3 is a plot of the spectral condition number distribution of the m sequence and the randomly generated sequence
  • FIG. 4 is a graph showing a mean square error curve of a channel estimation at a vehicle speed of 30 km/h according to an example of the present invention
  • FIG. 5 is a graph showing a bit error rate of a channel estimation of a OFDM system at a vehicle speed of 30 km/h according to an example of the present invention
  • FIG. 6 is a block diagram of a MIMO-OFDM system based on time domain channel estimation according to the present invention
  • FIG. 8 is a graph showing a mean square error curve of a channel estimation of a MO-OFDM system at a vehicle speed of 30 km/h;
  • FIG. 9 is a diagram showing the channel estimation of a MIMO-OFDM system at a vehicle speed of 30 km/h according to an example of the present invention.
  • the average bit error rate graph is based on the MMSE - Ordered IC (Minimum Mean Square Error-Ordered Interference Cancellation) detection for the MIMO system.
  • MMSE - Ordered IC Minimum Mean Square Error-Ordered Interference Cancellation
  • the discrete multipath channel can be characterized by a tapped delay line model. It is assumed that the channel coefficients of each path remain unchanged for one OFDM symbol time (if the channel estimation method is applied to other communication systems, that is, the channel remains unchanged for a certain amount of transmission data), the channel length L is unknown, for OFDM The symbol and the CP taken by the training sequence (Cyclic Prefix: if applied to other communication systems, the same CP is required before the training sequence, this CP has no relationship with the CP of the OFDM symbol) and the length is (L C >L ).
  • the length of the CP is greater than -1.
  • the training sequence and the OFDM time domain symbol samples use the same sampling rate.
  • the frequency response on each subcarrier of the channel corresponding OFDM can be obtained.
  • H [H(1), H(2), - -, H(N-1)] , where N is the number of subcarriers of OFDM.
  • the OFDM signal can be equalized using the ZF (Zero-Forcing) or MMSE (Minimum Mean-Square Error) criteria. For example, based on the zero-forcing criterion, the estimated value of the original signal is
  • the m sequence also known as the maximum length shift register sequence, is the best binary sequence of autocorrelation properties. If the length of the m sequence is , it satisfies the normalized autocorrelation function in one cycle.
  • mod(.,.) represents the remainder of the previous number after the modulo calculation of the following numbers.
  • the following theoretical derivation first estimates the impulse response of the channel based on arbitrary pseudo-random sequences, and then proves that the pseudo-random sequence based on good autocorrelation can obtain more accurate channel estimation. Finally, the m-sequence is introduced into the channel estimation method. And provide simulation of performance.
  • a pseudo-random sequence (such as an m-sequence) is inserted as a training signal for time-domain channel estimation.
  • - 1) is pseudo Individual elements of a random sequence (if it is an m-sequence, then these elements are bipolar bits; if they are other pseudo-random sequences, these elements are real numbers).
  • the added cyclic prefix can on the one hand absorb the multipath components of the previous data arriving delayed, and at the same time play a crucial role in simplifying the calculation of channel estimates.
  • the received signal is
  • Equation (11) J hjC P ss +l/L p n r (k) sk) ( 11 )
  • C is a normalized cross-correlation of a cyclic right-shifted pseudo-random sequence and s'', so that equation (11) ) written in the form of a matrix as follows
  • C C P -h + 0(n') ( 12)
  • C [C(0), C(1 .., C(J C -I)f is the cross-correlation calculated from equation (11)
  • the column vector obtained by the value alignment 0 ("') represents the vector calculated by the normalized cross-correlation of the noise calculated by equation (11) and the cyclic right-shift pseudo-random sequence.
  • C p j-th row, i th column is rotated right correlation value and the j-th pseudo-random sequence of i times
  • Cp The square matrix is known and fixed.
  • A- l ⁇ l ⁇ 1. Then the perturbed linear equations (A + A)(z + ) b + Sb also have a unique solution. And
  • the number of linguistic conditions corresponding to the m sequence is 1.8824; and the randomly generated sequence The corresponding spectral conditions are relatively large. If the latter is used for channel estimation, a large error will result.
  • any pseudo-random sequence can be applied to the algorithm, but the random sequence with poor correlation has a large error in channel estimation; only the random sequence with good autocorrelation In order to obtain a channel estimation result with higher accuracy.
  • the m sequence is used in the embodiment of the present invention, and if there are sequences with good or better autocorrelation, the same can be applied to the algorithm.
  • FIG. 2 is a flow chart of a channel estimation method for an OFDM system of the present invention.
  • Step 310 In the 206 part of the transmitting end, the pseudo-random sequence (using the m-sequence in the present embodiment) as the time-domain training sequence is added with the same length of the cyclic prefix as the OFDM symbol, and the cyclic prefix length of the training sequence may also be the OFDM symbol. Not the same, but must be greater than -1 is the maximum delay of the discrete channel, that is, the maximum delay of the channel is converted into the number of samples at the system sampling rate), and the training sequence with the cyclic prefix is inserted into the OFDM symbol at intervals. Between, sent to the receiving end.
  • the pseudo-random sequence as the training sequence inserted twice before and after may be different (for the m sequence, the m-sequences of different lengths or the m-sequences that are cyclically shifted may be regarded as different m-sequences).
  • the training sequence inserted before and after is preferably the same pseudo-random sequence, and the length of the CP also takes the same value.
  • C p will change.
  • the number of OFDM data symbols spaced between the two training sequences is determined by the rate of change of the time varying channel.
  • the maximum Doppler shift of the channel can be calculated from the carrier frequency and the moving speed of the mobile station, and the correlation time of the channel can be calculated from the maximum Doppler shift. In the relevant time, you can The channel is considered to be slowly changing.
  • the correlation time is converted into the number of OFDM data symbols.
  • the number of spaced OFDM data symbols can refer to this converted value. Of course, it is better to leave a certain margin, because the channel still changes after this time.
  • the specific method is to calculate the maximum Doppler shift ⁇ according to the speed of the mobile station.
  • V is the mobile station speed
  • c is the speed of light
  • / is the carrier frequency
  • the channel estimation method inserts a training sequence every several OFDM symbols, then the channel of the intermediate process is obtained by interpolation. Therefore, the number of OFDM symbols in this interval is preferably calculated by the following method:
  • the Num plane is the number of OFDM symbols that can be continuously transmitted, that is, the number of OFDM symbols spaced between two training sequences, r.
  • FDM is the period of one OFDM symbol.
  • Step 320 After receiving the data, the receiving end 209 sends the training sequence of the removed CP to the time domain channel estimation 210, and removes the OFDM symbol input 212 of the CP and the subsequent module, and prepares to perform frequency domain equalization processing based on the result of the channel estimation. .
  • a cross-correlation matrix C of received data and respective cyclically shifted pseudo-random sequences (such as m-sequences) is obtained using the received training sequence.
  • the first row of the C matrix is a cross-correlation value of a pseudo-random sequence (m-sequence) that receives data and loops right-shifted 0 times.
  • the last behavior of the C-matrix is to receive data and to circulate right-shifted L c -1 times.
  • the cross-correlation value of the random sequence (m-sequence); the pseudo-random sequence (m-sequence) is the pseudo-random sequence inserted as the current training sequence in step 310.
  • the estimation result of the channel impulse response is sent to the module 211 to obtain an estimated ⁇ (/ , corresponding to the frequency response on each subcarrier of the OFDM, and the impulse response of the corresponding time domain is denoted as ii head .
  • Step 340 After the predetermined number of OFDM data symbols are separated, the signal corresponding to the training sequence is received again, and in the same steps as before, the current frequencies corresponding to the OFDM subcarriers are calculated in the modules 210 and 211.
  • the estimate of the response ⁇ 7 , its corresponding time domain impulse response i has been ii to ,.
  • the frequency response on each subcarrier of the OFDM symbol between the two training sequences can be obtained by interpolation using ⁇ ⁇ and fi to7 .
  • the expression using the linear interpolation method is as follows:
  • N is the number of OFDM data symbols separated by two training sequences.
  • the previous training sequence is estimated to be approximated, so that the interpolation part can be omitted, and the accuracy of the estimation is affected to some extent.
  • Step 350 The estimation result of the channel frequency response is sent to the module 215 for frequency domain equalization together with the received data of the OFDM, and the frequency domain equalization may be based on the ZF criterion or the MMSE criterion.
  • the time domain equalization can also be performed using the channel impulse response.
  • a new cyclic estimation and frequency domain equalization cycle begins.
  • a new training sequence is received, a specified number of OFDM symbols have actually been transmitted.
  • the two ends of the continuous OFDM symbol are training sequences.
  • the channel interpolation is performed by using the training sequences of the two ends to obtain the channel frequency domain response corresponding to the OFDM symbol transmission part, and then the OFDM symbols are equalized by the interpolation result, which completes a loop.
  • This number of OFDM symbols is then spaced and a training sequence is sent. Then, the training sequence following the OFDM symbols that were sent last time is the training sequence of the first OFDM symbols transmitted this time.
  • the newly transmitted training sequence is the training sequence at the back of these OFDM symbols, which can be interpolated by the same method as before. channel. That is to say, a training sequence is actually used twice: first it is used as the end of the previous OFDM symbol segment for interpolation; later it is also used as the latter OFDM symbol The beginning of the segment is used again for interpolation.
  • the above channel estimation method can be applied to other communication systems, in which case the OFDM symbol data is signal data in the communication system.
  • the channel estimation method of the OFDM system will be described below by way of an example.
  • the RF carrier frequency of the uncoded system is 2 GHz
  • the channel adopts the TU (Typical Urban) simulation model.
  • the simulation is performed at a vehicle speed of 30 km/h, and the number of subcarriers of OFDM is 512.
  • the subcarrier spacing is 15 kHz
  • the modulation scheme on each subcarrier is 16QAM
  • a training sequence is inserted every 10 OFDM symbols.
  • the channel estimation is followed by frequency domain equalization recovery data based on the ZF criterion.
  • the mean squared error function of the channel estimate is defined here.
  • []* indicates the total ⁇ , and N indicates the number of OFDM subcarriers.
  • Figures 4 and 5 are the results of simulations performed on the above simulation platform, where the MSE (Mean Square Error) curve is the mean square error of the estimated channel and real channel parameters, BER (Bit Error Ratio: Error) Bit rate) The curve is the system's bit error rate, and SNR (Signal-to-Noise Ratio) is the signal-to-noise ratio at the time of simulation. Indicates the length of the m-sequence used for the simulation.
  • the LS estimation curve represents the MSE performance curve of the channel estimation based on the LS algorithm.
  • the ideal CSI (Channel State Information) curve indicates that the channel is known to the receiving end. BER performance curve.
  • the LMMSE (Linear Minimum Mean-Square Error) method greatly reduces the computational complexity of OFDM channel estimation under the premise of ensuring performance close to the MMSE (Minimum Mean-Square Error) method.
  • the LMMSE method is based on the correction of the estimation result of the LS method, and the operation of the channel estimation requires a total of N 2 + N multiplication (division) operations.
  • the equation (11) requires c x Z p multiplication
  • the equation (13) requires: c 2 multiplication operations, which can obtain a total of ⁇ x (J L c + ) multiplication operations.
  • the present invention should employ sequences with good autocorrelation, and m sequences are used in the examples.
  • the length of the m sequence is always an integer power of 2 minus 1 because the length of the cyclic prefix CP is greater than -1 (which is the maximum delay of the channel), and since the CP is taken from the back part of the m sequence, the m sequence The length is greater than -1. If the delay of the channel is large, then the length of the m sequence as the training sequence is correspondingly increased, and the short m sequence is no longer usable. In the above embodiment, the m sequence length is 127 or 255, which is only for the channel used in the simulation. If the multipath delay is very large, the longer length m sequence is also used for channel estimation. The algorithm of this channel estimation is performed in the time domain. The longer the length of the m sequence, the more obvious the suppression of noise, the higher the accuracy of channel estimation, but the more system redundancy is added.
  • the channel estimation method of the present invention is applied to a CDMA system, the method of which is similar to the application in an OFDM system, the CDMA chip data corresponds to OFDM symbol data, and the h head and lail are channel estimates.
  • the impulse response of the time multipath channel, iead and interpolated is the impulse response of the channel corresponding to the CDMA chip signal between the two training sequences.
  • the length of the CDMA chip data spaced between the two training sequences is similar to the correlation calculation method in the OFDM system.
  • the maximum Doppler shift can be calculated. Where ⁇ is the mobile station speed, c is the speed of light, / is the carrier frequency.
  • the correlation time is equal to the reciprocal of the maximum Doppler shift:
  • the channel is highly correlated. Since it is appropriate to estimate how many chips are inserted into a training sequence, the interval is also determined by the correlation time:
  • is the number of CDMA chips that can be continuously transmitted, that is, the number of CDMA chips that are separated by two training sequences, and ⁇ is the period of one chip.
  • the frequency response of the channel corresponding to the intermediately transmitted CDMA chip signal can be obtained by interpolation.
  • the impulse response corresponding to 0 ⁇ and 0 ⁇ and to , 7 can perform time domain interpolation to obtain the channel impulse response corresponding to the CDMA chip data between the training sequences, and the interpolation result is applied to the time domain equalization or RAKE receiver.
  • performs linear interpolation of frequency domain channels, or based on linear interpolation of time domain channels
  • the basic idea is to divide the difference between the corresponding parts of the channel into several parts, and regard the middle part as The arithmetic progression is obtained by the arithmetic progression, and then the interpolation result of the channel is obtained according to the position of the estimated part, and the corresponding interpolation value is added. If the interpolation is nonlinear, the computational complexity of the processing method is larger.
  • a MIMO-OFDM system based on time domain channel estimation (taking a 2 transmit/receive system as an example), and a time domain training sequence transmitted by each transmit antenna is also preceded by an OFDM symbol plus a cyclic prefix, where the length of the cyclic prefix Greater than Ll, L is the maximum delay of the discrete channel model between the current transmit antenna and each receive antenna.
  • L is the maximum delay of the discrete channel model between the current transmit antenna and each receive antenna.
  • use the training sequence to remove the CP, and estimate each
  • the channel impulse response between the transmitting and receiving antennas is then subjected to frequency domain equalization and symbol detection of MIMO in the frequency domain.
  • the training sequence on each transmit antenna may be different for the same reason as the previous OFDM system, because the channel estimation algorithm only uses the autocorrelation of the pseudo-random sequence for channel estimation, and is not limited to a specific pseudo-random. sequence. But the same training sequence can make the C p in the receiver the same, so the algorithm is exactly the same. Moreover, the m sequence can obtain better estimation results, and the accuracy of channel estimation by other randomly generated sequences is not as good as the m sequence, so it is preferable that the training sequences on each transmitting antenna are m sequences of the same length. Under the condition that the length is greater than 1 (which is the maximum delay of the discrete channel model), the use of a short training sequence can save system overhead.
  • the training sequence on the two antennas will be transmitted in succession.
  • the OFDM symbol is transmitted after the training sequence, and the number of symbols will be related to the time selectivity of the channel, similar to the determination rule of the number of OFDM symbols spaced between the training sequences in the previous single-issue single-receiving system. Since the training sequences transmitted by the two antennas are orthogonal in time, the corresponding channel estimation method for each pair of transmitting and receiving antennas is the same as the channel estimation method for the previous single-issue OFDM.
  • the transmitting antennas when one of the transmitting antennas transmits the training sequence, the other antennas do not transmit signals, and the transmitting antenna and each of the receiving antennas form a system with a single input and a single output, and the channel estimation is performed.
  • the method is the same as the previous single-issue single-receiving system, so each receiving antenna can directly estimate the channel between itself and this particular transmitting antenna based on the received signal at this time.
  • the channel estimation between each pair of transmitting and receiving antennas can be obtained.
  • the MIMO signal detection in the frequency domain is performed using the result of the channel estimation at the stage of data signal transmission and reception.
  • the training sequence on the two antennas will be the same as the OFDM symbol plus the same length of cyclic prefix, and then sent sequentially, as shown in Figure 7.
  • the training sequence from which the CP is removed is sent to the channel estimator, and the CP is removed.
  • the OFDM data is sent to modules 617 and 618 and subsequent modules ready to perform frequency domain equalization using the results of the channel estimation. Since the training sequences transmitted by the two antennas are orthogonal in time, at block 615, the method for estimating the channel frequency response between each pair of transmitting and receiving antennas is the same as that of the previous one.
  • the channel estimation method of the OFDM system is the same. For details, refer to the channel estimation process of the foregoing OFDM system.
  • the module 623 After obtaining the channel frequency response of the OFDM symbol, the module 623 obtains an estimate of the transmitted data using the MIMO signal detection and interference removal method in the frequency domain.
  • the channel estimation method of the MIMO-OFDM system of the present invention will be described below by way of an example.
  • the receiving end of the MIMO system uses the MMSE criterion-based sorting interference removal detection algorithm (MMSE+Ordered lC) to recover the transmitted data.
  • MMSE+Ordered lC MMSE criterion-based sorting interference removal detection algorithm
  • Figures 8 and 9 are the results of the system simulation.
  • the MSE curve is the mean square error of the estimated channel and the real channel average of each pair of transceiver antennas.
  • the BER curve is the average bit error rate of each pair of transceiver antennas.
  • the SNR is set during simulation. The signal to noise ratio of each pair of antennas. Indicates the length of the m-sequence used for simulation.
  • the LS estimation curve represents the MSE performance curve of the channel estimate based on the frequency domain LS algorithm.
  • the ideal CSI curve represents the BER performance curve obtained when the multipath channel is known to the receiving end.
  • the multi-antenna transceiving channel estimation method is similar to other multi-antenna transceiving communication systems except the MIMO-OFDM system, and the method is similar.
  • the m sequence serves as a training sequence and has two functions:
  • the m-sequence has a binary autocorrelation property, and its autocorrelation value is clear and fixed. From the formula (13) can see the final calculation of the channel estimation to use C P - ', and it is the inverse matrix C P, C P is right-shifted m-sequence of autocorrelation values arranged in a matrix, by the formula (7) Decided. Since the autocorrelation values of the m sequence are known, they are known.
  • any pseudo-random sequence can be applied to the proposed channel estimation method, but the pseudo-random sequence with poor autocorrelation has a large error in channel estimation; only the autocorrelation is better.
  • the pseudo-random sequence can obtain the channel estimation result with higher precision.
  • the m-sequence is employed in the practice of the present invention, and if there are sequences with good correlation or better autocorrelation, the same can be applied to the method.
  • the channel estimation method proposed by the invention adopts a pseudo-random sequence as a training sequence for time domain channel estimation, which reduces the complexity of channel estimation, and the operation complexity is lower than most existing channel estimation methods;
  • the overhead of the training sequence can be flexibly adjusted according to the needs of the system transmission rate to obtain a compromise between estimation accuracy and overhead.
  • the method has a suppression effect on noise and improves the accuracy of estimation.
  • the present invention is a method for channel estimation, which achieves higher channel estimation accuracy with lower computational complexity.
  • Theoretical analysis and simulation experiments also prove the rationality and effectiveness of the present invention.

Landscapes

  • Engineering & Computer Science (AREA)
  • Power Engineering (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Cable Transmission Systems, Equalization Of Radio And Reduction Of Echo (AREA)

Description

一种信道估计方法 技术领域
本发明涉及移动通信领域,特别是涉及一种移动通信系统的信道估计方 法, 本方法可应用于多种移动通信系统, 包含 OFDM ( Orthogonal Frequency Division Multiplexing: 正交频分复用) 系统和 MIMO-OFDM ( Multi-Input Multi-Output Orthogonal Frequency Division Multiplexing: 多天线收发 -正交 频分复用)系统, 同时, 如果采用适当的帧结构, 本方法也可应用于 CDMA ( Code Division Multiple Access: 码分多址) 系统的信道估计。 背景技术
正交频分复用 (OFDM )技术将高速数据流转换为一组低速并行传输的 数据流, 并使子信道的频带交叠, 具有较强的抗多径干扰能力和较高的带宽 利用率。 在充分散射信道条件下, 多天线收发 ( MIMO ) 系统可以获得比单 天线收发(SISO, single input single output)系统更大的信道容量。 在基于 MIMO和 OFDM 的新一代无线通信系统中, 由于传输速率较高, 需要使用 相干检测 (coherent detection )技术获得较高的性能, 因此信道估计成为 MIMO和 OFDM相关研究的一个重要方向。
精确的信道估计也可以提高 CDMA系统的性能。 信道估计可以应用于
CDMA系统的 RAKE接收机、 时域均衡接收机以及频域均衡接收机中。
信道估计技术从大的方面可以分为非盲估计和盲估计,以及在此基础上 产生的半盲估计。通常使用非盲估计获得较好的估计效果,计算复杂度较低, 这样还可以更好的跟踪无线信道的变化,提高接收机性能。传统的信道估计 方法往往要进行大量的矩阵运算, 复杂度较高, 受噪声的影响较大。
发明内容
鉴于以上情况, 本发明要解决的技术问题就是提出一种可以应用于 OFDM ( MIMO-OFDM )系统、 CDMA系统以及其它通信系统的信道估计方 法, 在保证较好信道估计精度的前提下, 使信道估计的运算负担减轻。
为解决上述问题,本发明提出一种信道估计方法,应用于移动通信系统, 包含以下步骤:
(a)在发送端发送的信号数据段中, 每间隔一定长度的信号数据段插 入带有长度为 Ze的循环前綴的伪随机序列作为训练序列, 所述 .大于 - 1 , 其中 是当前信道的离散信道模型的最大时延, 不带循环前缀时该伪随机序 列长度为 Lp
(b)接收端接收到训练序列的数据, 将去掉循环前缀部分后的接收数 据与 个移位伪随机序列做互相关, 得到互相关矩阵 C, 所述移位伪随机序 列为作为当前训练序列的伪随机序列分别循环右移 0,1,..., Lc-l次后所得, 所述 C矩阵第一行是接收数据和作为当前训练序列的伪随机序列循环右移 0 次得到的伪随机序列的互相关值, 依次下去, C矩阵最后一行是接收数据和 作为当前训练序列的伪随机序列循环右移 Lc-1 次得到的伪随机序列的互相 关值;
(c) 当前时刻的信道冲激响应的估计 fi = CP- 其中 Cp为一 Jc阶方阵, 为所述作为当前训练序列的伪随机序列分别循环右移 0, 1 ,..., Lc-l次得到的 伪随机序列的自相关矩阵, CP第 j行, 第 i列是所述作为当前训练序列的伪 随机序列循环右移 j次和 i次得到的伪随机序列的互相关值, Cp-1为 Cp的逆 矩阵, 当前时刻信道的冲激响应估计记为 lKd
所述方法进一步包含如下步骤:
(d)在间隔一定长度的信号数据段后, 接收端再次接收到对应于训练 序列的数据, 按步骤(b) 中的方法计算新的互相关矩阵 C, 按步骤(c) 中 方法得到当前信道的冲激响应的估计 , 利用 ieadtail进行插值得到两个 训练序列之间的信号数据段对应的信道的冲激响应。
所述方法进一步包含如下步骤:
(d)在间隔一定长度的信号数据段后, 接收端再次接收到对应于训练 序列的数据, 按步骤(b) 中的方法计算新的互相关矩阵, 按步骤(c) 中方 法得到当前信道的冲激响应的估计 to7 , 由 iihead和 得到对应的信道的频率 响应 ^^和^。,.,, 利用 ή,,^和 ±to7进行插值得到两个训练序列之间的信号数 据段对应的信道频率响应。
所述步糠 ( a ) 中插入的训练序列长度相同, 其循环前缀长度也相同, 所述步驟(c ) 中 Cp为一固定方阵, CP的逆矩阵 C-p1直接存储于接收端。
所述伪随机序列为最大长度移位寄存器序列即 m序列, 所述步骤( c ) 中 Cp其对角线上元素为 1 , 非对角线元素为 -1/ 。
所述系统为正交频分复用 OFDM系统或码分多址 CDMA系统, 所述步 骤(a ) 中, 两个训练序列间隔的一定长度的信号数据段为若干个 OFDM符 号数据或若干个 CDMA码片数据, 其个数根据载波频率和移动台移动速度 确定, N讓 = (l/3 ~ l/5)TeOT/r , 其中, N薩是两个训练序列间隔的 OFDM符 号或 CDMA码片的个数, Γ是一个 OFDM符号或一个 CDMA码片的周期, 为信号最大多普勒频移的倒数, 最大多普勒频移/ max = v/7c , !是移动台 速度, c是光速, /是载波频率。
所述系统为 OFDM系统, 所述步骤(a ) 中一定长度的信号数据段为若 干个 OFDM符号数据, 所述步骤(d ) 中 , ^和 i¾to7是信道估计时刻 OFDM 各个子载波上的频率响应的估计, Ηω和 ήω插值得到的是两个训练序列之 间的 OFDM符号数据各个子载波上的频率响应。
步骤 ) 中所述插值方法为线性插值方法 , 0为两个训练序列之间第 j个 OFDM符号数据的信道频率响应估计值, 其计算式为
H , = H + j (H tail - H w ) /(Num + 1), = 1,2,··· , Num
其中 Num是两个训练序列之间所间隔的 OFDM符号数据的个数。
所述系统为 CDMA系统, 所述步骤(a )中一定长度的信号数据段为一 段 CDMA数据码片数据, 所述步骤( d ) 中 ύΙκαά和 0to,7是信道估计时刻多径 信道的频率响应, ήΛ^和 ή,∞7插值得到的是两个训练序列之间的 CDMA码片 信号所对应的信道的频率响应。
所述系统为 CDMA系统, 所述步骤 ( a )中一定长度的信号数据段为一 段 CDMA数据码片数据, 所述步骤( d )中 和^„7是信道估计时刻多径信 道的冲激响应, ∞d和 iito7插值得到的是两个训练序列之间的 CDMA码片信 号所对应的信道的冲激响应。
一种多天线收发系统信道估计方法, 包含下述步骤:
( a )在发送端, 对每个发送天线, 发送一段规定长度的信号数据段之 前,插入带有长度为^的循环前缀的伪随机序列作为训练序列,所述 ^大于 1 -1 , 其中 是该发送天线和备接收天线之间离散信道模型的最大时延, 不 带循环前缀时该伪随机序列长度为 Lp , 各个发送天线上的训练序列先后发 送,保证每个发送天线上的训练序列与系统中其余发送天线的训练序列在时 间上正交;
( b )接收端的每个接收天线接收到当前发送天线发送的训练序列的数 据,去掉循环前缀部分后的接收数据与 ^个移位伪随机序列做互相关,得到 互相关矩阵 C, 所述移位伪随机序列为作为当前训练序列的伪随机序列分别 循环右移 0,1,..., Lc-\次后所得, 所述 C矩阵第一行为接收数据和作为当前 训练序列的伪随机序列循环右移 0次得到的伪随机序列的互相关值,依次下 去, C矩阵最后一行为接收数据和作为当前训练序列的伪随机序列循环右移 Lc-\次的伪随机序列的互相关值;
( c )每个接收天线和当前发送天线的当前时刻的信道冲激响应的估计 h = Cp 1C , 其中 CP为一个 ^阶方阵, 为所述作为当前训练序列的伪随机序列 分别循环右移 0,1,.. ·, Lc-l次得到的伪随机序列的自相关矩阵, Cp第 j行, 第 i列是所述作为当前训练序列的伪随机序列循环右移 j次和 i次得到的伪 随机序列的互相关值, Cp- 1为 Cp的逆矩阵, 将接收天线和当前发送天线之间 信道的当前时刻信道冲激响应估计记为 所有发送天线发送完毕后, 得 到每对接收天线和发送天线对应的信道的冲激响应估计。
所述方法进一步包含如下步骤:
( d )在间隔规定长度的信号数据之后, 接收天线再次接收到同一发送 天线对应于训练序列的数据, 按步驟(b ) 中的方法计算新的互相关矩阵 C , 按步骤(c )中方法得到当前信道的冲激响应的估计 利用 和 ii,ai,插值 得到两个训练序列之间的信号数据段对应的信道冲激响应, 或利用 έΛ ^和 ^ /对应的信道频率响应 ήΛ∞ί/和 ή ;,进行插值得到两个训练序列之间的信号 数据段对应的信道频率响应。 所述伪随机序列为最大长度移位寄存器序列即 m序列, 所述步骤 (c)中 CP其对角线上元素为 1 , 非对角线元素为- 1/ 。
所述步骤(a ) 中插入的训练序列长度相同, 其循环前缀长度也相同, 所述步骤(c ) 中 Cp为一固定方阵, CP的逆矩阵 C- p1直接存储于接收端。
本发明提出了一种针对移动通信系统, 包括 OFDM ( MIMO-OFDM )系 统和 CDMA系统的基于伪随机序列的信道估计新方法 , 巧妙利用伪随机序 列的自相关特性降低了信道估计的复杂度。当伪随机序列使用自相关较好的 m序列时, 本发明信道估计的精确度比现有技术高。 本信道估计方法能够根 据系统传输速率的需要灵活调整训练序列的开销,以取得估计精度和开销的 折中。 附图概述
结合以下附图以及具体实例对发明所做的详细描述将便于理解本发明 的原理、 步骤、 特点和优点, 附图中:
图 1为离散多径信道的抽头延迟线模型;
图 2为本发明基于 OFDM系统信道估计方法流程图;
图 3 是 m序列和随机产生序列的谱条件数分布图;
图 4为本发明一实例 OFDM系统信道估计在车速 30km/h时的均方误差 曲线图;
图 5为本发明一实例 OFDM系统信道估计在车速 30km/h时的误码率曲 线图;
图 6为本发明基于时域信道估计的 MIMO-OFDM系统框图;
图 7为本发明 2发 2收系统时域训练序列的插入方式;
图 8为本发明一实例 MO-OFDM系统信道估计在车速 30km/h时的均 方误差曲线图;
图 9为本发明一实例 MIMO-OFDM系统信道估计在车速 30km/h时的平 均误码率曲线图,采用的是针对 MIMO系统的 MMSE - Ordered IC( Minimum Mean Square Error- Ordered Interference Cancellation, 最小均方误差一 ^Jf干 扰消除)检测。 本发明的较佳实施方式
下面结合附图和实施例对本发明进行详细描述。
1) 多径信道模型
如图 1所示, 离散多径信道可以用一个抽头延迟线模型来表征。假定各 条路径信道系数在一个 OFDM符号时间内保持不变 (如果本信道估计方法 应用于其他通信系统, 就是一定数量的发送数据对应的时间内信道保持不 变),信道长度 L未知,对 OFDM符号和训练序列所取的 CP ( Cyclic Prefix: 循环前缀; 如果应用于其他通信系统, 训练序列前同样需要加 CP, 这个 CP 和 OFDM符号的 CP没有关系)长度为 ( LC>L ) 。 信道的离散冲激响应 可以表示为一个长度是 Ze的列向量 hH -'U , h满足条件: {/¾=0| ≤Ζ·≤ - 1}。 假设时域的发送数据长度是 , 因为本信道估计方法采 用.的伪随机序列的长度将与 OFDM的子载波数目不相等, 所以在信道估计 阶段 ^代表伪随机序列的长度; 而在 OFDM符号传送阶段, Zs代表了一个 OFDM符号时域采样的个数, 在不过采样的情况下就是 OFDM系统子载波 的数目 。 加上 CP 后, 假设发送序列是 X , 接收序列是 r, 其中 x = [x(0), x(l), -,x(Ls+Lc- ΐ)]τ , r = [r(0), r(l), ''、 LS+LC- 那么由抽头延迟线模型可以得到: r{k) = J htx{k - i) + n{k) ), 1, · · ¾ +^-1 ( 1 )
,·=0
将(1) 式写成矩阵的形式如下:
r = xh + η (2) 其中 S是由发送信号 χ排列而成的(Zs+Zc)> e的矩阵, 如(3)式所示:
Figure imgf000009_0001
取 S的第 Α = 0,1,··-, - 1)列写成列向量的形式如下
(4) 其中 ( = 0,1,2, 代表训练序列前面的数据多径延时到达的部分。 那么 S可以重新表示成 ^Ι Α,.·.,^^], 于是(2)可以表示成
Figure imgf000009_0002
其中
Figure imgf000009_0003
表示信道中的加性高斯白噪声, 服从相同 独立的均值为零、 方差为 σ„2的复高斯分布。
和 OFDM系统的原理一样, 如果信道的最大时延能够折合成 OFDM时 域符号 L个采样的时间长度, 那么 CP的长度就要大于 - 1。 这里训练序列 和 OFDM时域符号采样采用的是同样的采样速率。
2) OFDM基于时域信道估计的频域均衡思想
假设经过时域的信道估计, 已经得到了信道估计值 ii = H、 _x f , 那 么 就可以得到信道对应 OFDM 各个子载波上的频率响应 H = [H(1),H(2),--,H(N-1)] , 其中 N是 OFDM 的子载波数。 应用 ZF ( Zero-Forcing:迫零 )或 MMSE ( Minimum Mean-Square Error: 最小均方误 差)准则可以对 OFDM信号进行均衡。 比如基于迫零准则的均衡, 对原始 信号的估计值为
i=R … - 1) (6)
H H(0)'H(1)' 'H(N-l)
其中
Figure imgf000009_0004
ή是对真实 信道离散频域响应 Η的估值。 3)本发明信道估计方法 ;
m序列又称作最大长度移位寄存器序列,是自相关特性最好的二进制序 列。 如果 m序列的长度为 , 它在一个周期内的归一化自相关函数满足
Figure imgf000010_0001
其中 代表循环移位的位数, mod(.,.)代表前面的数字对后面的数字进行模运 算后得到的余数。
下面的理论推导首先基于任意的伪随机序列对信道的冲激响应进行估 计, 然后将证明基于自相关性好的伪随机序列才能得到比较精确的信道估 计, 最后将 m序列引入本信道估计方法, 并且提供性能的仿真。
每隔一定数量的 OFDM符号 (如果本信道估计方法应用于其他通信系 统, 就是每隔一定数量的发送数据) , 就插入一个伪随机序列 (如 m序列) 作为时域信道估计的训练信号 , 这个训练序列也要同 OFDM符号一样, 将 最后的 e (CP的长度)个元素作为循环前缀放在前面。 也就是说, 假设伪 随机序列为 s = [PN(0),PN(1), ···, PN(Z -1)],其中 PN() (=0, 1 ,… - 1)为伪随机序列 的各个元素 (如果是 m序列, 那么这些元素就是双极性的比特位; 如果是 其它的伪随机序列, 这些元素就是一些实数)。 那么加入循环前缀后的训练 序歹 为 X = [x(0),x(l ), ···, x(LP +LC -1)] =[PN(I -Lc),---, PN(IP - 1), PN(0), ···, PN(J - 1)]。 加入的循环前缀一方面可以吸收延迟到达的前面数据的多径分量,同时对于 简化信道估计的计算起着至关重要的作用。
参照 (5) 式, 对应于训练序列, 接收信号为
Figure imgf000010_0002
其中 (j = 0, 1, 2,… , Zc— 1)是信 的冲激响应, n = ["(0), "(1), -,n{LP+Lc- l)f是力口 性高斯白噪声, fpN =[r( )>rO-X-;r(LP+Lc -I)]7"为训练序列的接收向量, ^对应 于 (4)给出的定义, 可以表示成 =|Χ- · · ¾ -1),Λ(0Μ1),· . +/c-j-l)f。 接收端对训练序列的接收数据要先去掉循环前缀部分。 对照 (8) 式, 去掉对应于循环前缀部分的矩阵的行,写出删除循环前缀后的接收向量的等 效表达式为
Lc-ί
PN + η (9) 户 ο
rPN = [rm(0),rPN(i),- -,rm(Lp -l)]r 为 fPN的后 LP个元素, n' = [η'(0),η' (I),-、"'(LP - 1)]7'为 n 的后 ^个元素, 可以看出: 是伪随机序列8循环右移 _次后得到的伪随机 序列。 由 (9) 式, rPN各个元素的计算公式为
Figure imgf000011_0001
其中 s^)是伪随机序列 s循环右移 j次后得到的伪随机序列的第 t个元素。
rPN与各个 s''的互相关表达式为
Lp-\
C =1/ w(„)
Figure imgf000011_0002
= J hjCP s s +l/Lp nr(k)s k) ( 11 ) 式(11 ) 中 C 是循环右移伪随机序列 和 s''的归一化互相关, 于是可以将 式(11 )写成矩阵的形式如下
C = CP-h + 0(n') ( 12) 其中 C = [C(0),C(1 ..,C(JC-I)f是由式(11 )计算得到的各个互相关值排列得 到的列向量。 0("')代表式( 11 )计算出来的噪声和循环右移伪随机序列的归 一化互相关排列而成的向量。 而 Cp=[C ], , = 0,1,..., - 1是一个 Jc阶的方 阵, Cp第 j行, 第 i列是循环右移 j次和 i次的伪随机序列的相关值, 很容 易得到, Cp是一个对称矩阵, 也就是说 CP的第 i行、 第 j列和第 j行、 第 i 列是相等的, 当使用某个确知的伪随机序列进行时域的信道估计时, Cp这 个方阵就是已知并且固定不变的。 当使用 m序列作为训练序列进行信道估 计时, CP对角线上的元素都是 1, 而非对角线元素都是 - 1/ZP。 因此当使用 某个固定长度的 m序列进行时域的信道估计时, CP这个方阵的逆矩阵 Cp-'就 是已知并且固定不变的, 可以直接存储于接收机中,从而降低信道估计的复 杂度。 显而易见, 采用不同长度的 m序列进行信道估计, Cp要进行相应的 调整。 0(«')忽略不计,最后得到一对收发天线之间的信道冲激响应估计值为
ii = CP 1C ( 13) 下面讨论为什么自相关性好的伪随机序列 (比如 m序列)适合于我们 的信道估计算法, 首先介绍定理 1:
定理 1: 设线性方程组 Az = b有唯一解2。 记 和^)分别为 A与 b的扰 动, 且满足 |A- l^l <1。 则扰动后的线性方程组(A + A)(z + ) = b + Sb也有 唯一解。 且
Figure imgf000012_0001
其中, ||·|表示矩阵或者向量的任何一种范数, |卜||2表示谱范数, κ(·)表示 矩阵的谱条件数, 定义如下:
Figure imgf000012_0002
由( 14)看出,谱条件数 A:(A)反映了线性方程组 =1>的解对于 A和 b的 扰动的稳定程度。 式( 14)的右端是 (A)的单调增函数, 所以 AT(A)愈大, 对 于 Α和 b的同样大小的扰动, 方程组解的相对改变量可能愈大。
同理, 根据式 ( 12)求解多径信道 h实际上就是解线性方程组 C = Cp.h, 而 0(«')相当于一个扰动向量。 CP的谱条件数/(CP)决定了方程组的解 h对于 扰动。("')的稳定程度。 K(CP)越小 , 那么由式(13)估计出的信道误差越小。 显然, 用于本信道估计的伪随机序列自相关特性越好, 那么 K(CP)的条件数 越小。
m序列是自相关特性最好的二进制序列。根据上面的理论分析, 可以得 知 m序列比较适合于本信道估计算法。 随机产生 100个长度为 63的实数伪 随机序列, 针对这些伪随机序列和长度为 63 的 m序列, 分别计算它们的 CP=[C ], j,i = 0,l,-,Lc- 这里假定 Zc =30, 然后分别计算这些矩阵的谱 条件数。 图 3显示了这些语条件数的大小关系(当 ^取小于 63的任意值时, 会得到类似的结果)。 m序列对应的语条件数是 1.8824; 而随机产生的序列 对应的谱条件数都比较大。 如果采用后者进行信道估计将造成较大的误差。 综上所述,可以得出这样的结论:任何的伪随机序列都能应用于本算法, 但是自相关性不好的随机序列造成信道估计的误差较大;只有自相关性较好 的随机序列才能取得精度较高的信道估计结果。 本发明实施例中采用了 m 序列, 如果还有自相关性很好或者更好的序列, 一样可以应用于本算法。
下面结合附图和实施例对信道估计的过程给予具体说明。
图 2是本发明 OFDM系统信道估计方法流程图。
步骤 310: 在发送端的 206部分, 作为时域训练序列的伪随机序列 (本 实施例中使用 m序列 )和 OFDM符号一样加上同样长度的循环前缀, 训练 序列的循环前缀长度也可以和 OFDM符号不一样, 但必须大于 -1 是离 散信道的最大时延,也就是把信道的最大时延折算成系统采样速率下的采样 的数目 ) , 并将带有循环前缀的训练序列间隔地插入 OFDM符号之间, 发 送到接收端。
其中, 前后两次插入的作为训练序列的伪随机序列可以不一样(对于 m 序列来说, 不同的长度的 m序列或者经过循环移位的 m序列都可以视为不 同的 m序列) 。 但是为了方便, 这里前后所插入的训练序列最好是一样的 伪随机序列, CP的长度也取相同的值。
如果训练序列不一样了, 那么 Cp就要发生变化。 比如说两次所采用的 分别是不同长度的 m序列, 如果设定的循环前缀的长度 c始终是相等的 , 分析一下 CP =[C '], j, / = 0, 1,… , . - 1的特征就是: 因为 "是 m序列的归一 化自相关, 可以知道 Cp的对角线上是 1 , 但是其他的位置都是 - 1/ZP , ^是 m序列的长度。 Zp不同, -l/i^也就不同了。 这些值是由 m序列的归一化自 相关决定的。 所以说不同的 m序列, 对应的 CP也是不一样的。 为了简化接 收机的结构, 最好采用一样长度的 m序列和 CP长度。
两个训练序列之间间隔 OFDM数据符号的个数, 要才艮据时变信道的变 化速率来决定。根据载波频率和移动台的移动速度可以算出信道的最大多普 勒频移, 由最大多普勒频移可以算出信道的相关时间。 在相关时间内, 可以 认为信道是慢变的。 将相关时间折算成 OFDM数据符号的个数, 间隔的 OFDM数据符号的个数可以参考这个折算值, 当然最好可以留有一定的余 量, 因为信道在这个时间内毕竟还是有变化的。
具体方法是根据移动台的速度, 可以核算出最大多普勒频移
f J max = -v f J
C
其中 V是移动台速度, c是光速, /是载波频率。 而相关时间?;。,等于最大多 普勒频移的倒数: cor
J max
在实际系统中可以取这个相关时间的 l/3 ~ l/5, 这个时间范围内, 可以 认为信道是相关性很强的。 因为本信道估计方法是每隔若干个 OFDM符号 插入一次训练序列, 然后中间过程的信道是通过插值得到的。 所以这个间隔 的 OFDM符号的数目最好是通过以下方法核算出来:
Num riOFDM = (1/3 ~ 1/5)
丄 OFDM
其中, Num麵是核算得到的可以连续传送的 OFDM符号的个数, 即两个训 练序列之间间隔的 OFDM符号的个数, r。FDM是一个 OFDM符号的周期。 步驟 320: 接收端 209接收到数据后, 去除 CP的训练序列送入 210进 行时域信道估计, 而去除了 CP的 OFDM符号送入 212以及后续模块, 准备 基于信道估计的结果进行频域均衡处理。 在模块 210, 首先基于式(11 ) , 利用接收到的训练序列,得到接收数据与各个循环移位伪随机序列 (比如 m 序列)的互相关矩阵 C。 所述 C矩阵第一行是接收数据和循环右移 0次的伪 随机序列 (m序列)的互相关值, 依次下去, C矩阵最后一行为接收数据和 循环右移 Lc -1次的伪随机序列 ( m序列)的互相关值; 所述伪随机序列 ( m 序列) 均为步骤 310中插入的作为当前训练序列的伪随机序列。
步骤 330: 然后利用 (13 ) 式计算当前时刻的信道冲激响应的估计 , ^Ph = Cp-IC , 其中 Cp是循环右移 0,1...^ - 1次的训练序列 (m序列) 的归一 化互相关。将信道冲激响应的估计结果送入模块 211 ,得到对应于 OFDM各 个子载波上的频率响应的估计 ∞(/ , 其对应的时域的冲激响应记为 iihead
步骤 340: 在间隔规定数目的 OFDM数据符号后,再次接收到对应于训 练序列的信号, 利用与前述相同的步骤, 在模块 210和 211中, 计算得到当 前的对应于 OFDM各个子载波上的频率响应的估计 ∞7,其对应的时域的冲 激响应 i己为 iito,,。
对于两个训练序列之间的 OFDM符号各个子载波上的频率响应, 可以 使用 ήΑ∞ίί和 fito7进行插值得到。 例如使用线性插值方法的表达式如下:
H = H + j χ (H - H head ) l(Num + 1), 7 = 1, 2, · · · , Num
其中 N画是两个训练序列之间相隔的 OFDM数据符号的个数。
当然由于两个训练序列之间的信号数据长度是由相关时间核算出来的, 前面的那个训练序列估计近似 ,这样就可以省略了插值部分,估计的精确度 会受到一定的影响。
步骤 350:信道频率响应的估计结果和 OFDM的接收数据一起被送入模 块 215进行频域均衡,频域均衡可以基于 ZF准则或者 MMSE准则。也可以 利用信道冲击响应进行时域的均衡。
当接收到新的训练序列时, 新的信道估计与频域均衡的循环过程开始。 当接收到新的训练序列时, 实际上又已经传送了规定数目的 OFDM符 号。 连续的 OFDM符号两头是训练序列, 利用两头的训练序列进行信道插 值得到 OFDM符号传送部分对应的信道频域响应, 然后利用插值的结果对 OFDM符号进行均衡,这就算完成了一个循环。然后间隔这个数目的 OFDM 符号, 再发送一次训练序列。 那么上一次发送的那些 OFDM符号后面的那 个训练序列就是这一次发送的这些 OFDM符号的前头的训练序列, 最新发 送的训练序列就是这些 OFDM符号的后头的训练序列, 利用前面同样的方 法可以插值得到信道。也就是说一个训练序列实际上被使用两次: 首先它作 为前面的 OFDM符号段的结尾用于插值; 后来它还作为后面的 OFDM符号 段的开头再次用于插值。
上述信道估计方法可应用于其它通信系统, 此时 OFDM符号数据为通 信系统中的信号数据。
下面通过一个实例说明 OFDM系统的信道估计方法。
首先建立如下仿真平台: 未编码系统的射频载波频率是 2GHz, 信道采 用 TU ( Typical Urban: 典型市区)仿真模型, 在车速为 30km/h的情况下进 行仿真, OFDM的子载波数量是 512, 子载波间隔为 15kHz, 每个子载波上 的调制方式均为 16QAM, 循环前缀的长度 c = 60 , 每隔 10个 OFDM符号 插入一个训练序列。 LS ( Least Square: 最小二乘)估计采用全导频方式; 而对于本发明的估计方法, 为了比对不同长度 m序列的训练序列对信道估 计精确度的影响, 分别采用^ = 63,127,255,511四种长度的 m序列。 信道估计 之后采用基于 ZF准则的频域均衡恢复数据。
为了比较信道估计的精确度, 这里定义信道估计的均方误差函数为
Figure imgf000016_0001
[]*表示取共扼, N表示 OFDM子载波的数目。
图 4和 5是上述仿真平 '台上进行仿真的结果,其中, MSE ( Mean Square Error: 均方误差) 曲线是估计出的信道与真实信道参数的均方误差, BER ( Bit Error Ratio: 误比特率) 曲线是系统的误码率, SNR ( Signal-to-Noise Ratio: 信噪比)是仿真时的信噪比。 表示用于仿真的 m序列的长度, LS estimation 曲线表示基于 LS 算法得到的信道估计的 MSE性能曲线, ideal CSI ( Channel State Information: 信道状态信息) 曲线表示信道对接收端已 知的情况下得到的 BER性能曲线。
从图 4看出,随着训练序列长度的增加,信道估计的均方误差是下降的, 这是由于噪声的影响被压缩成原来的 1/ P。 从图 5看, 从长度等于 63的 m 序列开始,误码率曲线开始逼近理想信道的情况,但是长度大于 255之后误 码率曲线改善减緩。 从图 4和图 5还可以看出, 本发明方法性能优于原有 LS信道估计方法。 从计算量分析: 若 OFDM系统所有子载波上都插入导频, LS信道估计 在所有子载波上共需要 W次除法运算, 其中 N是 OFDM子载波的数目。 虽 然 LS 方法计算量较小, 但是性能较差。 LMMSE ( Linear Minimum Mean-Square Error: 线性最小均方误差) 方法在保证性能接近 MMSE ( Minimum Mean-Square Error: 最小均方误差)方法的前提下, 大大降低了 OFDM信道估计的运算复杂度。 LMMSE方法建立在对 LS方法的估计结果 进行修正的基础上, 信道估计的运算总共需要 N2 +N次乘(除)法运算。 本 发明中 (11 )式需要 cxZp次乘法运算, (13 )式需要 :c 2次乘法运算, 可以 得到本方法总共需要 ^ x(JLc + )次乘法运算。 LC < N , LP < N , 显而易见本 发明的复杂度比 LMMSE算法复杂度要低。 而且,在训练序列长度增加的情 况下, 信道估计的性能已经接近了理想信道的性能。 实际上, 本方法的复杂 度也比大多数性能较好的信道估计方法计算量要低的多。 随着 m序列长度 的增加,信道估计的计算量会有所增加, 从计算复杂度和信道估计的精确度 折中考虑, 取 m序列长度为 127或者 255比较合适。
本发明应该采用自相关性好的序列, 实例中采用了 m序列。 m序列的 长度总是 2的整数次幂减去 1 , 因为循环前缀 CP的长度要大于 - 1 ( 是信 道的最大时延) , 又因为 CP是取自 m序列的后面部分, 所以 m序列的长 度要大于 - 1。如果信道的时延很大, 那么这个时候作为训练序列的 m序列 的长度也要相应的增加, 短的 m序列已经不能够使用了。 上述实施例中 m 序列长度为 127或 255, 仅仅是针对仿真中所采用的信道而言的 , 如果多径 时延非常大, 那么也要采用长度较长的 m序列进行信道估计。 本信道估计 的算法是在时域中进行的, m序列的长度越长, 对噪声的压制作用越明显, 信道估计的精度越高, 但是所加入的系统冗余也越多。
在本发明的另一实施例中, 将本发明信道估计方法应用于 CDMA系统 中, 其方法类似于在 OFDM系统中的应用, CDMA码片数据对应于 OFDM 符号数据, hheadlail是信道估计时刻多径信道的冲激响应, iead和 ,插值 得到的是两个训练序列之间的 CDMA码片信号所对应的信道的冲激响应。 两个训练序列之间间隔的 CDMA码片数据的长度和在 OFDM系统中的相关 计算方法类似, 首先根据移动台的速度, 可以核算出最大多普勒频移 , 其中 ^是移动台速度, c是光速, /是载波频率。 相关时间 等于最大多普 勒频移的倒数:
在实际系统中可以取这个相关时间的丄 ~丄,这个时间范围内,可以认为
3 5
信道是相关性很强的。因为要估算间隔多少个码片插入一次训练序列比较合 适, 那么间隔也是由相关时间决定的:
〜 ^^ 其中, 是核算得到的可以连续传送的 CDMA码片的个数, 即两个训 练序列间隔的 CDMA码片个数, φ是一个码片的周期。
如果是结合 CDMA频域均衡技术应用于 CDMA系统, 步骤 340中, 可 以使用 ^和 进行插值得到中间发送的 CDMA码片信号所对应的信道 的频率响应。 0^和0^对应的冲激响应 和 to,7可以进行时域插值, 得到 训练序列之间的 CDMA码片数据所对应的信道冲激响应, 插值结果应用于 时域均衡或者 RAKE接收机。
不论是基于 和 ή,αη进行频域信道线性插值, 还是基于 和 ^„进行 时域信道线性插值,其基本思想都是将信道前后对应部分的差值等分成若干 份, 将中间部分看成是等差数列, 然后按照要估计部分的位置, 加上相应的 插值份额就得到了信道的插值结果了。如果是非线性的插值,处理方法的计 算复杂度还要大一些。
4 )本发明 MIMO-OFDM信道估计方法
图 6为基于时域信道估计的 MIMO - OFDM系统(以 2发 2收系统为 例) , 每个发送天线所发送时域训练序列前也与 OFDM符号一样加上循环 前缀, 其中循环前缀的长度大于 L-l , L是当前发送天线和各接收天线之间 离散信道模型的最大时延。 在接收端使用去除 CP的训练序列, 估计出每一 对收发天线间的信道冲激响应, 然后在频域中进行 MIMO的频域均衡与符 号检测。
每个发送天线上的训练序列可以是不一样的, 其原因和前面的 OFDM 系统一样,因为信道估计算法中仅仅是利用伪随机序列的自相关性进行信道 估计, 并不局限于特定的伪随机序列。 但是相同的训练序列, 可以使接收机 中的 Cp是相同的, 所以算法就是完全一样的。 而且 m序列可以取得较好的 估计结果, 其他的随机产生的序列进行信道估计的精度也不如 m序列, 所 以最好每个发送天线上的训练序列都是相同长度的 m序列。 在满足长度大 于 1 ( 是离散信道模型的最大时延) 的条件下, 采用长度短的训练序列 可以节省系统开销。
为了保证各个发送天线训练序列在时间上的正交性 ,以 2发 2收系统为 例, 两根天线上的训练序列将先后发送。 训练序列后面发送 OFDM符号, 符号的数目将与信道的时间选择性有关,和前面的单发单收系统中训练序列 之间间隔的 OFDM符号数目的确定规则类似。 由于两根天线发送的训练序 列在时间上是正交的,所以每一对收发天线间信道冲激相应的估计方法与前 面单发单收 OFDM的信道估计的方法相同。
具体来说, 当某一根发射天线发射训练序列的时候, 其它的天线不发送 信号,这一根发射天线就和每一根接收天线之间组成了一个单输入单输出的 系统,其信道估计方法和前面的单发单收系统一样, 所以每一个接收天线可 以根据这时的接收信号,直接估计出本身和这一根特定的发射天线之间的信 道。 同理可以得到每一对收发天线之间的信道估计。 信道估计完成以后, 在 数据信号发送与接收的阶段, 利用信道估计的结果来完成频域中的 MIMO 信号检测。 - 在模块 611和 612部分, 两根天线上的训练序列将和 OFDM符号一样 加上同样长度的循环前缀, 然后被先后发送, 如图 7所示。 在接收端的模块 613和 614, 去掉了 CP的训练序列被送入信道估计器, 去掉了 CP的
OFDM数据被送入模块 617和 618以及后续模块,准备利用信道估计的结果 进行频域均衡。 由于两根天线发送的训练序列在时间上是正交的 , 所以在模 块 615 , 每一对收发天线间的信道频率响应的估计方法与前面单发单收的 OFDM系统的信道估计方法是一样的, 详细情况可参看前述 OFDM系统的 信道估计过程。
在得到 OFDM符号的信道频率响应之后,模块 623在频域中利用 MIMO 的信号检测与干扰删除方法得到发送数据的估计值。
下面通过一个实例说明本发明 MIMO-OFDM系统信道估计方法。 首先 建立的仿真平台如下: 2发 2收的未编码 MIMO-OFDM系统, 为了比对不同 长度 m序列的训练序列对信道估计精确度的影响 ,分别采用 Lp = 63,127, 255三 种长度的 m序列。 MIMO系统的接收端采用基于 MMSE准则的排序干扰删 除检测算法(MMSE+Ordered lC )来恢复发送数据。 其他系统设置与前面所 述 OFDM的仿真系统的参数设置相同。
图 8和 9是系统仿真的结果, 其中, MSE曲线是各对收发天线的估计 信道与真实信道平均的均方误差, BER曲线是各对收发天线的平均误码率, SNR是仿真时设定的各对天线的信噪比。 表示用于仿真的 m序列的长度, LS estimation 曲线表示基于频域 LS 算法得到的信道估计的 MSE性能曲线, ideal CSI 曲线表示多径信道对接收端已知的情况下得到的 BER性能曲线。
从图 8中可以看出, 随着训练序列长度的增加, 信道估计的均方误差是 下降的, 这是由于噪声的影响被压缩成原来的 1/^。 从图 9看, 从长度等于 63的 m序列开始, 误码率曲线开始逼近理想信道的情况, m序列长度越长 性能越好。
在前面对 OFDM 系统中的信道估计算法的复杂度分析中, 需要大概
Zc x( lc + )次乘法, 将算法引入了 MIMO-OFDM系统中以后, 因为每一对 天线进行的是相同的运算, 假设总共存在 M对天线(比如说 2发 2收系统, 存在 M = 4对收发天线) , 那么需要的乘法运算是^ χ( + ) χ Μ , 很多性 能较好的 MMO-OFDM信道估计需要复杂的运算, 本发明降低了运算的复 杂度。
该多天线收发信道估计方法除 MIMO-OFDM系统外, 同样适用于其 他多天线收发通信系统, 方法类似。 在本发明中, m序列作为训练序列, 有两个作用:
1 )对于信道估计来说, 固定长度的 m序列可以让式( 13 ) 中的 C 1是 固定的、 已知的。 已知的矩阵可以降低接收机的复杂度。
由式(7 )可以看到, m序列具有二值自相关特性, 而且它的自相关的 值是明确的、 固定的。从式( 13 )可以看到最后信道估计的计算要用到 CP-', 而它是 CP的逆矩阵, CP是 m序列的右移位自相关值排列成的矩阵, 是通过 式(7 )决定的。 因为 m序列的自相关值是已知的, 所以 就是已知的。
2 )可以降低噪声的影响。
从式( 11 )的第二项可以看到, m序列与噪声是独立的, 在计算接收信 号与 m序列的右移位互相关的时候, m序列和噪声的互相关是^!艮小的。 这 样, 噪声的作用就得到了压制。
从上面的分析可以得到这样的结论,其实本信道估计方法可以使用任 何的伪随机序列, 不同的序列 CP-'是不一样的, 换掉一个用于信道估计的伪 随机序列就需要重新计算 CP- !。 但是根据前面定理 1和相关的证明, 可以得 知: 式(12 )中的 cP的傅条件数; (cp)决定了方程组的解 h对于扰动 0(«')的稳 定程度。 K(Cp)越小, 那么由式(13 )估计出的信道误差越小。 显然, 用于本 信道估计的伪随机序列自相关特性越好,那么 ^^的条件数越小。 m序列是 自相关特性最好的二进制序列, 所以可以得知 m序列比较适合于本信道估 计算法。
综上所述, 可以得出这样的结论: 任何的伪随机序列都能应用于本信道 估计方法,但是自相关性不好的伪随机序列造成信道估计的误差较大; 只有 自相关性较好的伪随机序列才能取得精度较高的信道估计结果。本发明实施 中采用了 m序列, 如果还有自相关.性很好或者更好的序列, 一样可以应用 于本方法。
工业实用性
本发明提出的信道估计方法,采用伪随机序列作为时域信道估计的训练 序列,降低了信道估计的复杂度,运算复杂度低于现有大多数信道估计方法; 可以根据系统传输速率的需要灵活调整训练序列的开销,以取得估计精度和 开销的折中; 在信道估计的相关运算中, 本方法对噪声有抑制作用, 提高了 估计的精确度。
总之, 本发明是一种用于信道估计的方法, 以较低的运算复杂度取得了 较高的信道估计精度,理论分析和仿真实验也证明了本发明的合理性和有效 性。

Claims

权 利 要 求 书
1、 一种信道估计方法, 应用于移动通信系统, 包含以下步骤:
(a)在发送端发送的信号数据段中, 每间隔一定长度的信号数据段插 入带有长度为 的循环前缀的伪随机序列作为训练序列, 所述 大于 - 1, 其中 Z是当前信道的离散信道模型的最大时延, 不带循环前缀时该伪随机序 列长度为 LP
(b)接收端接收到训练序列的数据, 将去掉循环前缀部分后的接收数 据与 ^个移位伪随机序列做互相关, 得到互相关矩阵 C, 所述移位伪随机序 列为作为当前训练序列的伪随机序列分别循环右移 0,1,..., Lc-\次后所得, 所述 C矩阵第一行是接收数据和作为当前训练序列的伪随机序列循环右移 0 次得到的伪随机序列的互相关值, 依次下 C矩阵最后一行是接收数据和 作为当前训练序列的伪随机序列循环右移 Lc - 1 次得到的伪随机序列的互相 关值;
(c)当前时刻的信道冲激响应的估计
Figure imgf000023_0001
阶方阵, 为所述作为当前训练序列的伪随机序列分别循环右移 0,1,..., Lc-\次得到的 伪随机序列的自相关矩阵, CP第 j行, 第 i列是所述作为当前训练序列的伪 随机序列循环右移 j次和 i次得到的伪随机序列的互相关值, Cp- 1为 CP的逆 矩阵, 当前时刻信道的冲激响应估计记为 head
2、 如权利要求 1所述的方法, 其特征在于: 所述方法进一步包含如 下步骤:
(d)在间隔一定长度的信号数据段后, 接收端再次接收到对应于训练 序列的数据, 按步骤(b) 中的方法计算新的互相关矩阵 C, 按步骤(c) 中 方法得到当前信道的冲激响应的估计 ^, 利用 £^αίίM,7进行插值得到两个 训练序列之间的信号数据段对应的信道的冲激响应。
3、 如权利要求 1所述的方法, 其特征在于: 所述方法进一步包含如 下步驟:
(d)在间隔一定长度的信号数据段后, 接收端再次接收到对应于训练 序列的数据, 按步骤(b ) 中的方法计算新的互相关矩阵, 按步驟(C ) 中方 法得到当前信道的冲激响应的估计 由 £,,^和^寻到对应的信道的频率 响应 0/tt 和 ήωί/ , 利用 ήΛ∞ί/和 进行插值得到两个训练序列之间的信号数 据段对应的信道频率响应。
4、 如权利要求 1所述的方法, 其特征在于: 所述步骤(a )中插入的 训练序列长度相同, 其循环前缀长度也相同, 所述步骤(c ) 中 CP为一固定 方阵, Cp的逆矩阵 C 1直接存储于接收端。
5、 如权利要求 1至 4之间任一所述的方法, 其特征在于: 所述伪随 机序列为最大长度移位寄存器序列即 m序列, 所述步骤(c )中 CP其对角线 上元素为 1 , 非对角线元素为- 1/ΖΡ
6、 如权利要求 1至 3之间任一所述的方法, 其特征在于: 所述系统 为正交频分复用 OFDM系统或码分多址 CDMA系统, 所述步骤 ( a ) 中, 两个训练序列间隔的一定长度的信号数据段为若干个 OFDM符号数据或若 干个 CDMA码片数据, 其个数根据载波频率和移动台移动速度确定, Num = (1/3 ~ H5)Tcor IT , 其中, Num是两个训练序列间隔的 OFDM符号或 CDMA码片的个数, Γ是一个 OFDM符号或一个 CDMA码片的周期, 7;。,为 信号最大多普勒频移的倒数, 最大多普勒频移/ V是移动台速度, c是光速, /是载波频率。
7、 如权利要求 3所述的方法,其特征在于:所述系统为 OFDM系统, 所述步骤( a ) 中一定长度的信号数据段为若干个 OFDM符号数据, 所述步 骤(d ) 中 ^和^„是信道估计时刻 OFDM各个子载波上的频率响应的估 计, 和 to7插值得到的是两个训练序列之间的 OFDM符号数据各个子载 波上的频率响应。
8、 如权利要求 7所述的方法, 其特征在于: 步骤(d )中所述插值方 法为线性插值方法, 为两个训练序列之间第 j个 OFDM符号数据的信道 频率响应估计值, 其计算式为
H = H +;' x (H ,。,, - H head ) /(Num = 1, 2, · · · , Num
其中 Num是两个训练序列之间所间隔的 OFDM符号数据的个数。
9、 如权利要求 3所述的方法,其特征在于:所述系统为 CDMA系统, 所述步驟(a )中一定长度的信号数据段为一段 CDMA数据码片数据, 所述 步骤(d ) 中 ^和 ή,。,7是信道估计时刻多径信道的频率响应, ήω和 0toi/插 值得到的是两个训练序列之间的 CDMA码片信号所对应的信道的频率响 应。
10、 如权利要求 2所述的方法,其特征在于:所述系统为 CDMA系统, 所述步骤(a )中一定长度的信号数据段为一段 CDMA数据码片数据, 所述 步骤(d ) 中 和 ,7是信道估计时刻多径信道的冲激响应, Λ∞ί/和 ,7插值 得到的是两个训练序列之间的 CDMA码片信号所对应的信道的冲激响应。
11、 一种多天线收发系统信道估计方法, 包含下述步骤:
( a )在发送端, 对每个发送天线, 发送一段规定长度的信号数据段之 前,插入带有长度为 .的循环前缀的伪随机序列作为训练序列,所述 Ze大于 L-1 , 其中 是该发送天线和各接收天线之间离散信道模型的最大时延, 不 带循环前缀时该伪随机序列长度为 Lp , 各个发送天线上的训练序列先后发 送,保证每个发送天线上的训练序列与系统中其余发送天线的训练序列在时 间上正交;
( b )接收端的每个接收天线接收到当前发送天线发送的训练序列的数 据,去掉循环前缀部分后的接收数据与 个移位伪随机序列做互相关,得到 互相关矩阵 C , 所述移位伪随机序列为作为当前训练序列的伪随机序列分别 循环右移 0,1,..., Lc-\次后所得, 所述 C矩阵第一行为接收数据和作为当前 训练序列的伪随机序列循环右移 0次得到的伪随机序列的互相关值,依次下 去, C矩阵最后一行为接收数据和作为当前训练序列的伪随机序列循环右移 Lc-l次的伪随机序列的互相关值;
( c )每个接收天线和当前发送天线的当前时刻的信道冲激响应的估计 h = Cp 1C , 其中 Cp为一个 阶方阵, 为所述作为当前训练序列的伪随机序列 分别循环右移 0,1,..., 次得到的伪随机序列的自相关矩阵, CP第 j行, 第 i列是所述作为当前训练序列的伪随机序列循环右移 j次和 i次得到的伪 随机序列的互相关值, Cp- 1为 Cp的逆矩阵, 将接收天线和当前发送天线之间 信道的当前时刻信道冲激响应估计记为 6,^ , 所有发送天线发送完毕后, 得 到每对接收天线和发送天线对应的信道的冲激响应估计。
12、 如权利要求 11所迷的方法, 其特征在于: 所述方法进一步包含如 下步骤:
•( d )在间隔规定长度的信号数据之后, 接收天线再次接收到同一发送 天线对应于训练序列的数据, 按步骤(b ) 中的方法计算新的互相关矩阵 C , 按步骤(c )中方法得到当前信道的冲激响应的估计 ,; 利用 ^和^。,7插值 得到两个训练序列之间的信号数据段对应的信道冲激响应, 或利用 ίι 和 to,7对应的信道频率响应 ήΛ ^和 ήω,进行插值得到两个训练序列之间的信号 数据段对应的信道频率响应。
13、 如权利要求 11或 12所述的方法, 其特征在于: 所述伪随机序列 为最大长度移位寄存器序列即 m序列, 所述步驟 (c)中 Cp其对角线上元素为 1 , 非对角线元素为- l/iP
14、 如权利要求 11或 12所述的方法, 其特征在于: 所述步骤(a )中 插入的训练序列长度相同, 其循环前缀长度也相同, 所述步骤(c ) 中 CP为 一固定方阵, CP的逆矩阵 直接存储于接收端。
PCT/CN2007/000916 2007-03-21 2007-03-21 Méthode d'évaluation d'un canal WO2008113216A1 (fr)

Priority Applications (2)

Application Number Priority Date Filing Date Title
PCT/CN2007/000916 WO2008113216A1 (fr) 2007-03-21 2007-03-21 Méthode d'évaluation d'un canal
CN200780046701.7A CN101578829B (zh) 2007-03-21 2007-03-21 一种信道估计方法

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2007/000916 WO2008113216A1 (fr) 2007-03-21 2007-03-21 Méthode d'évaluation d'un canal

Publications (1)

Publication Number Publication Date
WO2008113216A1 true WO2008113216A1 (fr) 2008-09-25

Family

ID=39765357

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2007/000916 WO2008113216A1 (fr) 2007-03-21 2007-03-21 Méthode d'évaluation d'un canal

Country Status (2)

Country Link
CN (1) CN101578829B (zh)
WO (1) WO2008113216A1 (zh)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101547171B (zh) * 2009-05-11 2011-03-30 清华大学 一种适用于无线突发通信的信道估计方法及其系统
CN102474474A (zh) * 2009-07-29 2012-05-23 高通股份有限公司 协作多点通信中的自适应传输
CN103973398A (zh) * 2013-01-31 2014-08-06 中兴通讯股份有限公司 数据发送、数据接收方法和装置
CN112688889A (zh) * 2020-12-11 2021-04-20 北京邮电大学 一种无人机测控系统的信道估计方法和装置
CN113612560A (zh) * 2021-09-16 2021-11-05 西安交通大学 一种面向三维mimo信道仿真的无人机信道模拟方法、装置和系统
CN115549722A (zh) * 2022-09-02 2022-12-30 北京理工大学 变步长归一化lms全双工非线性自干扰消除方法及系统
CN115861141A (zh) * 2022-12-02 2023-03-28 北京领云时代科技有限公司 基于pcnn神经网络的无人机获取图像处理系统及方法

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105162265A (zh) * 2015-09-15 2015-12-16 华南理工大学 一种基于多天线接收的无线电磁波能量采集装置与方法
CN105245036B (zh) * 2015-09-15 2018-12-11 华南理工大学 一种基于多发射单元的无线电磁波能量采集的方法和装置
CN106953682B (zh) * 2017-03-28 2019-09-27 电子科技大学 一种用于无人机组网协同的方法
CN112636798B (zh) * 2020-12-28 2022-04-15 中山大学 一种面向低轨星座通信的信号发送与接收方法

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060062336A1 (en) * 2004-09-17 2006-03-23 Benq Corporation Method of channel estimation
CN1801792A (zh) * 2004-12-31 2006-07-12 电子科技大学 一种mimo-ofdm系统中的信道估计方法
CN1917397A (zh) * 2006-09-19 2007-02-21 北京邮电大学 一种mimo-ofdm系统信道估计的方法

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1277359C (zh) * 2002-08-01 2006-09-27 上海交通大学 正交频分复用通信系统载波频率偏移估计方法

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060062336A1 (en) * 2004-09-17 2006-03-23 Benq Corporation Method of channel estimation
CN1801792A (zh) * 2004-12-31 2006-07-12 电子科技大学 一种mimo-ofdm系统中的信道估计方法
CN1917397A (zh) * 2006-09-19 2007-02-21 北京邮电大学 一种mimo-ofdm系统信道估计的方法

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101547171B (zh) * 2009-05-11 2011-03-30 清华大学 一种适用于无线突发通信的信道估计方法及其系统
CN102474474A (zh) * 2009-07-29 2012-05-23 高通股份有限公司 协作多点通信中的自适应传输
CN102474474B (zh) * 2009-07-29 2015-08-26 高通股份有限公司 协作多点通信中的自适应传输
US9172561B2 (en) 2009-07-29 2015-10-27 Qualcomm Incorporated Adaptive transmissions in coordinated multiple point communications
CN103973398A (zh) * 2013-01-31 2014-08-06 中兴通讯股份有限公司 数据发送、数据接收方法和装置
CN112688889A (zh) * 2020-12-11 2021-04-20 北京邮电大学 一种无人机测控系统的信道估计方法和装置
CN113612560A (zh) * 2021-09-16 2021-11-05 西安交通大学 一种面向三维mimo信道仿真的无人机信道模拟方法、装置和系统
CN115549722A (zh) * 2022-09-02 2022-12-30 北京理工大学 变步长归一化lms全双工非线性自干扰消除方法及系统
CN115861141A (zh) * 2022-12-02 2023-03-28 北京领云时代科技有限公司 基于pcnn神经网络的无人机获取图像处理系统及方法

Also Published As

Publication number Publication date
CN101578829B (zh) 2013-01-30
CN101578829A (zh) 2009-11-11

Similar Documents

Publication Publication Date Title
WO2008113216A1 (fr) Méthode d&#39;évaluation d&#39;un canal
Liu et al. Transmit antennae space-time block coding for generalized OFDM in the presence of unknown multipath
Zeng et al. A semi-blind channel estimation method for multiuser multiantenna OFDM systems
Van der Veen et al. A subspace approach to blind space-time signal processing for wireless communication systems
CN1643867B (zh) 用于估计信道的设备和方法
Tepedelenlioglu et al. Transmitter redundancy for blind estimation and equalization of time-and frequency-selective channels
CN106254284B (zh) 一种基于低轨卫星系统的快变信道估计方法
JP4287777B2 (ja) 送信装置及び受信装置
Lin et al. Linear precoding assisted blind channel estimation for OFDM systems
CN100385824C (zh) 一种mimo-ofdm系统的自适应信道估计方法
CN109688075B (zh) 一种无线通信的信道估计方法及装置
CN107332797A (zh) 一种电力线ofdm通信系统中的信道估计方法
CN103414678B (zh) 基于Vector OFDM的双选择性信道的变换域均衡方法
US7619964B2 (en) High doppler channel estimation for OFD multiple antenna systems
Pan et al. An improved subspace-based algorithm for blind channel identification using few received blocks
Ahmed et al. Parameter estimation and equalization techniques for communication channels with multipath and multiple frequency offsets
CN109302240A (zh) 基于双选择衰落信道的低复杂度osdm串行均衡方法
Barhumi et al. Estimation and direct equalization of doubly selective channels
JP4299302B2 (ja) マルチキャリア信号を事前コード化するための装置及び方法
Niranjane et al. Performance analysis of different channel estimation techniques
WO2012035345A2 (en) Improvements in ofdm communication systems
Özdemir et al. Sparsity-aware joint frame synchronization and channel estimation: Algorithm and USRP implementation
Sahu et al. A comparative analysis of LS and MMSE channel estimation techniques for MIMO-OFDM system
CN109217954A (zh) 基于双选择衰落信道的低复杂度osdm块均衡方法
Wan et al. Semi-blind channel estimation of MIMO-OFDM systems with pulse shaping

Legal Events

Date Code Title Description
WWE Wipo information: entry into national phase

Ref document number: 200780046701.7

Country of ref document: CN

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 07720490

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 07720490

Country of ref document: EP

Kind code of ref document: A1