OTFS signal detection method based on hybrid message passing

Abstract

The invention relates to an OTFS signal detection method based on hybrid message passing, and belongs to the technical field of wireless communication. In the method, after a receiving end receives a signal, block LMMSE pre-detection is firstly carried out in a time domain, a pre-detection result is converted to a time delay-Doppler domain through OTFS inverse transformation and is demodulated through QAM to serve as an initial value of an HMP signal detector, and a time delay-Doppler domain pre-detection result is sent to the HMP signal detector after matched filtering to complete signal detection. According to the method, the initial value is provided for the HMP signal detector through the time domain block LMMSE pre-detection, the signal detection success rate is improved, and the complexity of the algorithm is further reduced by reducing the number of iterations of the HMP algorithm in a matched filtering mode.

Description

Technical Field

[0001] The present invention belongs to the technical field of wireless communications, and mainly relates to a signal detection technology for an OTFS communication system receiving end, and in particular to an OTFS signal detection method based on hybrid message transmission. Background Art

[0002] In 2017, Cohere, an American company, proposed the Orthogonal Time Frequency Space (OTFS) modulation and demodulation technology. Unlike OFDM, OTFS models the time-varying physical channel from the perspective of delay and Doppler frequency shift, modulates the data to be transmitted in the Delay-Doppler (DD) domain, and then transforms it to the Time-Frequency (TF) domain through a specific method. After the data symbols in the DD domain are converted to the TF domain, they will occupy the entire TF domain plane, thereby making full use of the system's time diversity and frequency diversity, and achieving better performance than OFDM under the same conditions.

[0003] Due to the inherent characteristics of OTFS in transmitting two-dimensional modulation symbols and performing two-dimensional transformation between delay-Doppler and time-frequency domains, traditional MMSE detection algorithms and ZF detection algorithms face the problems of high complexity and low detection accuracy. Signal detection based on the high sparsity of the channel matrix will inevitably further reduce the computational complexity. The detection algorithm based on message passing, by establishing a factor graph, only transmits iterative information between the message node and the variable node corresponding to the position of the non-zero element, but its convergence speed is too slow and the number of iterations is too large. Summary of the invention

[0004] In order to solve the problems existing in the prior art, the present invention proposes an OTFS signal detection method based on hybrid message passing. The method can be used in the channel detection processing of the receiving end of the OTFS communication system, by performing block LMMSE detection in the time domain to provide an initial value for the subsequent hybrid message passing algorithm, and adding a matched filtering algorithm before the hybrid message passing detection algorithm in the delay-Doppler domain, so as to further reduce the number of iterations of the detection algorithm and ensure that the performance is not lost.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] An OTFS signal detection method based on hybrid message passing comprises the following steps:

[0007] Step 1: At the transmitting end, the information bits are QAM modulated in the delay-Doppler domain and converted to the time domain after OTFS transformation for transmission;

[0008] Step 2, the transmitted signal s reaches the receiving end after passing through the time and frequency dual selective channel G, and the received signal r is obtained;

[0009] Step 3: After receiving the signal, the receiving end performs block LMMSE pre-detection in the time domain;

[0010] Step 4: Pre-detection of the signal Perform OTFS inverse transformation and transform the signal Converting from the time domain to the delay-Doppler domain, we get

[0011] Step 5: In the time-frequency domain Perform QAM demodulation to obtain QAM demodulation results

[0012] Step 6, Perform matched filtering to obtain

[0013] Step 7, QAM demodulation result As the initial value, Perform hybrid message passing detection to obtain the final OTFS signal detection result

[0014] Furthermore, in step 1, the number of symbols of the OTFS transformation is M×N, where M and N respectively represent the number of delay axis and Doppler axis grid points in the delay-Doppler domain, and M and N also represent the number of subcarriers and symbols in the time-frequency domain.

[0015] Furthermore, in step 2, the relationship between the received signal r and the sent signal s is:

[0016] r=G×s+w

[0017] Among them, G is the time domain channel matrix and w is the noise.

[0018] Furthermore, in step 3, the transmitted signal s and the received signal r are divided into N blocks s of size M. n and r n , the time domain channel matrix G is a block diagonal matrix, which is decomposed into N parallel linear matrix equations G0, G1, …G N-1 As the corresponding channel matrix G n , the relationship between the received signal and the transmitted signal after block processing is:

[0019] r n =G n ×s n +w n

[0020] Among them, w nFor noise.

[0021] Furthermore, in step 3, the received signal r after block division is n Perform LMMSE pre-detection by:

[0022]

[0023] in,(·) H represents the conjugate transpose operation of the matrix, is the noise variance, I M Represents the unit matrix of dimension size M×M;

[0024] Signal after pre-detection for:

[0025]

[0026] The beneficial effects of the present invention are:

[0027] 1. Reduce the number of iterations of OTFS system signal detection. The present invention adds a matched filter algorithm before the hybrid message passing algorithm, so that the amplitude of the diagonal elements of the channel matrix is ​​greater than the amplitude of the non-diagonal elements, thereby reducing the number of iterations of the detection algorithm.

[0028] 2. Improved detection performance of the OTFS system. The present invention performs block LMMSE detection in the time domain and performs QAM demodulation on the result and uses it as the initial value of the hybrid message passing algorithm for input, thereby accelerating the convergence speed of the hybrid message passing algorithm and improving the detection performance of the OTFS system.

[0029] 3. Reduce the computational complexity of the OTFS detection algorithm. The present invention reduces the complexity while ensuring performance by adopting hybrid message transmission.

[0030] In summary, the present invention improves the message passing algorithm and improves the detection accuracy of the message passing algorithm through methods such as approximate message passing and iterative combination. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is a flow chart of an OTFS signal detection method based on hybrid message passing.

[0032] Figure 2 It is a schematic diagram of the implementation principle of the present invention. DETAILED DESCRIPTION

[0033] An OTFS signal detection method based on hybrid message passing comprises the following steps:

[0034] Step 1: At the transmitting end, the information bits are QAM modulated in the delay-Doppler domain and converted to the time domain after OTFS transformation for transmission.

[0035] Step 2: The transmitted signal s reaches the receiving end after passing through the time and frequency dual selective channel G, and the received signal r is obtained.

[0036] Step 3: After receiving the signal, the receiving end performs block LMMSE pre-detection in the time domain.

[0037] Step 4: Pre-detection of the signal Perform OTFS inverse transformation and transform the signal Converting from the time domain to the delay-Doppler domain, we get

[0038] Step 5: In the time-frequency domain Perform QAM demodulation to obtain QAM demodulation results

[0039] Step 6, Perform matched filtering to obtain

[0040] Step 7, QAM demodulation result As the initial value, the signal Perform hybrid message passing detection to obtain the final OTFS signal detection result

[0041] The following takes 16QAM modulation, 16 symbols and 64 subcarriers as an example to explain the method in more detail. Figure 1 , including the following steps:

[0042] (1) At the transmitting end, the information bits are modulated by 16QAM in the delay-Doppler domain and converted to the time domain for transmission after OTFS transformation. The number of symbols of OTFS transformation is 64×16, that is, the number of grid points of the delay axis and Doppler axis in the delay-Doppler domain is 64 and 16 respectively, and the number of subcarriers and symbols in the time-frequency domain is also 64 and 16.

[0043] (2) The transmitted signal s reaches the receiving end after passing through the time and frequency dual selective channel G, and the received signal r is obtained. The relationship between the received signal r and the transmitted signal s can be expressed as:

[0044] r=G×s+w

[0045] Where G is the time domain channel matrix and w is the noise.

[0046] (3) After receiving the signal, the receiving end performs block LMMSE pre-detection in the time domain. The specific method is:

[0047] Divide the transmitted signal s and the received signal r into 16 blocks of size 64 n and r n , the time domain channel matrix G is a block diagonal matrix, which is decomposed into 16 parallel smaller linear matrix equations G0, G1, …G N-1 As the corresponding channel matrix G n , the received signal and the transmitted signal after block processing can be expressed by the following formula:

[0048] r n =G n ×s n +w n

[0049] The received signal after block division n Perform LMMSE pre-detection, the formula is as follows:

[0050]

[0051] in,(·) H represents the conjugate transpose operation of the matrix, is the noise variance, I M Represents the identity matrix of dimension M×M.

[0052] Finally, the output of the time domain block LMMSE detection algorithm is

[0053] (4) Signal after pre-detection Perform OTFS inverse transformation and transform the signal Converting from the time domain to the delay-Doppler domain, we get

[0054] (5) In the time-frequency domain Perform QAM demodulation to obtain QAM demodulation results

[0055] (6) Yes Perform matched filtering to obtain

[0056] (7) QAM demodulation results As the initial value, the signal Perform hybrid message passing detection to obtain the final OTFS signal detection result

[0057] The hybrid message passing algorithm is a combination of the message passing algorithm and the Gaussian approximate message passing algorithm. The algorithm divides the channel path into a main path and an auxiliary path, executes the message passing algorithm (non-approximate algorithm) on the main path, and executes the Gaussian approximate message passing algorithm on the auxiliary path, thereby reducing complexity while ensuring performance.

[0058] Figure 2 It is a schematic diagram of the implementation principle of the present invention, which includes a time domain block LMMSE pre-detector, a delay-Doppler domain matched filter (MF) device and a hybrid message passing (HMP) signal detector. Among them, after receiving the signal, the receiving end first performs block LMMSE pre-detection in the time domain, and the pre-detection result is converted to the delay-Doppler domain through OTFS inverse transformation, and is demodulated by QAM as the initial value of the HMP signal detector. The delay-Doppler domain pre-detection result is sent to the HMP signal detector after matched filtering to complete signal detection.

[0059] In summary, the present invention provides an initial value for the HMP signal detector through time-domain block LMMSE pre-detection, thereby improving the signal detection success rate, and further reduces the complexity of the algorithm by reducing the number of iterations of the HMP algorithm through matched filtering.

Claims

1. An OTFS signal detection method based on hybrid message passing, characterized in that: The following steps are involved: Step 1: At the transmitting end, the information bits are QAM modulated in the delay-Doppler domain and converted to the time domain after OTFS transformation for transmission; Step 2, the transmitted signal s reaches the receiving end after passing through the time and frequency dual selective channel G, and the received signal r is obtained; Step 3: After receiving the signal, the receiving end performs block LMMSE pre-detection in the time domain; Step 4: Pre-detection of the signal Perform OTFS inverse transformation and transform the signal Converting from the time domain to the delay-Doppler domain, we get Step 5: In the time-frequency domain Perform QAM demodulation to obtain QAM demodulation results Step 6, Perform matched filtering to obtain Step 7, QAM demodulation result As the initial value, Perform hybrid message passing detection to obtain the final OTFS signal detection result 2. The OTFS signal detection method based on hybrid message passing according to claim 1, characterized in that: In step 1, the number of symbols of OTFS transformation is M×N, where M and N represent the number of delay axis and Doppler axis grid points in the delay-Doppler domain, respectively, and M and N also represent the number of subcarriers and symbols in the time-frequency domain.

3. The OTFS signal detection method based on hybrid message passing according to claim 2, characterized in that: In step 2, the relationship between the received signal r and the transmitted signal s is: r=G×s+w Among them, G is the time domain channel matrix and w is the noise.

4. The OTFS signal detection method based on hybrid message passing according to claim 3, characterized in that: In step 3, the transmitted signal s and the received signal r are divided into N blocks of size M n and r n , the time domain channel matrix G is a block diagonal matrix, which is decomposed into N parallel linear matrix equations G0, G1, …G N-1 As the corresponding channel matrix G n , the relationship between the received signal and the transmitted signal after block processing is: r n =G n ×s n +w n Among them, w n For noise.

5. The OTFS signal detection method based on hybrid message passing according to claim 4, characterized in that: In step 3, the received signal r after block division is n Perform LMMSE pre-detection by: in,(·) H represents the conjugate transpose operation of the matrix, is the noise variance, I M Represents the unit matrix of dimension size M×M; Signal after pre-detection for:

Images