Symbol-level precoding method based on wide linear processing

Abstract

The invention discloses a symbol-level precoding method based on wide linear processing, which comprises the following steps that: a base station obtains channel state information, and obtains a zero-forcing precoding matrix of the wide linear processing according to the channel state information; optimizing the disturbance variable, pre-coding each symbol in the transmission frame according to the optimized disturbance variable, and obtaining an optimal disturbance variable by minimizing the sending power; and under the optimal disturbance variable, a final sending signal is obtained in combination with the zero-forcing precoding matrix subjected to wide linear processing. According to the method, wide linear processing and symbol-level precoding are combined, the degree of freedom of a communication system can be improved, the effect of non-strict phase rotation constructive interference is generated, the sending power of the system is greatly reduced, and the bit error rate performance of the system is improved.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a symbol-level precoding method based on wide linear processing. Background Art

[0002] In typical multi-user transmission, interference exists because the transmission data of different signals are superimposed and sent in the wireless communication channel. In traditional communication systems, interference is usually regarded as a performance limiting factor. Therefore, traditional multi-user multi-antenna precoding schemes are designed to suppress or even eliminate interference. These methods can be obtained through the instantaneous channel state information (CSI) in each block and are collectively referred to as block-level precoding (BLP). Recent studies have shown that symbol-level precoding (SLP) can manipulate the power and direction of the interference signal in the complex plane symbol by symbol, making the interference signal a useful source of the required signal power.

[0003] Standard statistical signal processing usually imposes a restrictive assumption on complex signals, that is, the complex zero-mean random vector is a second-order circular signal. However, in some more complex communication systems, the use of circular Gaussian signals while ignoring the non-circular properties of complex random variables cannot achieve the optimality of the communication system. In communication systems, non-circular signals are a common signal, such as Binary Phase Shift Keying (BPSK) and Offset Quadrature Amplitude Modulation (OQAM). In addition, the orthogonal phase imbalance phenomenon in the communication link, that is, the power imbalance between the in-phase component and the orthogonal component of the signal, can also lead to the generation of non-circular signals.

[0004] To date, most precoding technologies have focused on signal processing based on a linear processing framework. At the same time, existing SLPs are too complex to be applied in actual communication scenarios, especially in small-scale MIMO situations, where the performance gain of SLPs is less than that of traditional BLPs. Summary of the invention

[0005] Purpose of the invention: In view of the above problems, the purpose of the present invention is to provide a symbol-level precoding method based on wide linear processing.

[0006] Technical solution: A symbol-level precoding method based on wide linear processing of the present invention is applied to a wireless transmission system, wherein the wireless transmission system includes a base station and K users, the base station is equipped with N transmitting antennas, each user is equipped with a single antenna, and N is greater than K;

[0007] The symbol level precoding method comprises:

[0008] The base station obtains channel state information, and obtains a zero-forcing precoding matrix for wide linear processing according to the channel state information;

[0009] Optimize the disturbance variable, precode each symbol in the transmission frame according to the optimized disturbance variable, and obtain the optimal disturbance variable by minimizing the transmission power;

[0010] Under the optimal disturbance variable, the zero-forcing precoding matrix of wide linear processing is combined to obtain the final transmission signal.

[0011] Further, optimizing the disturbance variable, and precoding each symbol in the transmission frame according to the optimized disturbance variable includes:

[0012] Defining the disturbance variable And precode each symbol in the transmission frame to obtain:

[0013] x SLP =G SLP (s+u)

[0014] In the formula, x SLP represents the transmitted symbol vector after symbol-level precoding, is the precoding matrix, s represents the symbol sequence;

[0015] Introducing the equivalent symbol matrix And the diagonal matrix B, the elements in the diagonal matrix B correspond to each element of the symbol sequence s, then the symbol sequence s of the data is equivalent to:

[0016] Introduce a real-valued perturbation vector Optimize the disturbance variables. Every element are all non-negative real numbers, satisfying the restriction that the sign amplitude increases and the phase remains unchanged, then the perturbation vector is equivalent to:

[0017] Then the antenna's transmit symbol vector is re-expressed as:

[0018]

[0019] Furthermore, the optimal disturbance variables obtained by minimizing the transmission power include:

[0020] An optimization function is constructed to minimize the transmit power of the antenna, where the optimization function is expressed as:

[0021] min||x SLP ||2

[0022]

[0023] In the formula, || ||2 represents the bi-norm of the matrix;

[0024] Introduce a temporary matrix A=G SLP B and vector The optimization function is expressed as:

[0025]

[0026] Using the Euclidean norm duality between complex vectors And the augmented vector v = [v1,v2,...,v i ,...,v M ] performs real number simplification on the optimization function, where any complex number is expressed as v i =[a i ,b i ], i = 1, ..., M represents the index value of the vector, M represents the dimension of the complex vector, a i is the real part of the vector, b i is the imaginary part of the vector, j represents the imaginary unit; then the optimization function is expressed as:

[0027]

[0028] In the formula, represents the real part of the matrix, represents the imaginary part of the matrix,

[0029] Set the off-diagonal elements of the matrix product is 0, that is:

[0030]

[0031] Then the closed form of the optimal disturbance variable is:

[0032]

[0033] Furthermore, the zero-forcing precoding matrix of wide linear processing is:

[0034]

[0035] Where F represents the channel state information, and the superscript H represents the conjugate transpose.

[0036] Furthermore, let G SLP =W WL-ZF , the final transmission signal is expressed as:

[0037]

[0038] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages:

[0039] The present invention adopts wide linear processing precoding combined with symbol-level precoding algorithm to improve the degree of freedom of the communication system. Compared with symbol-level precoding based on linear processing, it can produce the effect of non-strict phase rotation constructive interference, greatly reduce the transmission power of the system, and improve the bit error rate performance of the system; the symbol-level precoding design is converted into a non-negative least squares problem, which greatly reduces the computational complexity and is conducive to circuit implementation in FPGA. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 is a flow chart of a symbol level precoding method based on wide linear processing;

[0041] Figure 2 A schematic diagram of the distribution of constellation points for receiving signals in a wireless transmission system without noise;

[0042] Figure 3 This is a comparison chart of the average total power consumption of the transmitting antenna under different numbers of users;

[0043] Figure 4 The figure is a comparison of the bit error rate performance of the precoding algorithms considered under different numbers of users;

[0044] Figure 5 Comparison of bit error rate performance based on various precoding methods after using LDPC coding for a multi-user MISO system. DETAILED DESCRIPTION

[0045] In order to make the objectives, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0046] A symbol-level precoding method based on wide linear processing described in this embodiment is applied to a wireless transmission system, wherein the wireless transmission system includes a base station and K users, the base station is equipped with N transmitting antennas, each user is equipped with a single antenna, and N is greater than K;

[0047] like Figure 1 As shown, the symbol level precoding method comprises the following steps:

[0048] Step 1: The base station obtains channel state information, and obtains a zero-forcing precoding matrix for wide linear processing according to the channel state information;

[0049] Step 2: Optimize the disturbance variable, precode each symbol in the transmission frame according to the optimized disturbance variable, and obtain the optimal disturbance variable by minimizing the transmission power;

[0050] Step 3: Under the optimal disturbance variable, the zero-forcing precoding matrix of wide linear processing is combined to obtain the final transmission signal.

[0051] In step 1, the channel state information obtained by the base station is recorded as F, and the zero-forcing precoding matrix based on wide linear processing is:

[0052]

[0053] Where, represents the channel state information, and the superscript H represents the conjugate transpose.

[0054] Further, optimizing the disturbance variable, and precoding each symbol in the transmission frame according to the optimized disturbance variable includes:

[0055] Defining the disturbance variable And precode each symbol in the transmission frame to obtain:

[0056] x SLP =G SLP (s+u)

[0057] In the formula, x SLP represents the transmitted symbol vector after symbol-level precoding, is the precoding matrix, s represents the symbol sequence;

[0058] Introducing the equivalent symbol matrix And the diagonal matrix B, the elements in the diagonal matrix B correspond to each element of the symbol sequence s, then the symbol sequence s of the data is equivalent to: The superscript T indicates transposition;

[0059] Introduce a real-valued perturbation vector Optimize the disturbance variables. Every element are all non-negative real numbers, satisfying the restriction that the sign amplitude increases and the phase remains unchanged, then the perturbation vector is equivalent to:

[0060] Then the antenna's transmit symbol vector is re-expressed as:

[0061]

[0062] Furthermore, the optimal disturbance variables obtained by minimizing the transmission power include:

[0063] An optimization function is constructed to minimize the transmit power of the antenna, where the optimization function is expressed as:

[0064] min||x SLP ||2

[0065]

[0066] In the formula, || ||2 represents the bi-norm of the matrix;

[0067] In order to transform the above optimization problem into a non-negative least squares problem, a temporary matrix A = G is introduced. SLP B and vector The optimization function is expressed as:

[0068]

[0069] Using the Euclidean norm duality between complex vectors And the augmented vector v = [v1,v2,...,v i ,...,v M ]Simplify the optimization function by real numbers, and any complex number can be expressed as v i =[a i ,b i ], i = 1, ..., M represents the index value of the vector, M represents the dimension of the complex vector, a i is the real part of the vector, b i is the imaginary part of the vector, j represents the imaginary unit; then the optimization function is expressed as:

[0070]

[0071] In the formula, represents the real part of the matrix, represents the imaginary part of the matrix, The above formula is a standard NNLS problem, and the optimal disturbance variable can be solved using the fast least squares algorithm. When all elements in are 0, this SLP algorithm is equivalent to the BLP algorithm, that is,

[0072] Although the optimal disturbance variable can be efficiently calculated using the fast NNLS algorithm However, since the algorithm still requires a large number of iterations, its implementation is somewhat challenging. In order to better meet the real-time and high-efficiency requirements in engineering practice, the closed-form solution of the NNLS problem is used to solve it. Although the performance of the closed-form solution is slightly reduced, it meets the real-time requirements in the communication system well. Set the off-diagonal elements of the matrix product is 0, that is:

[0073]

[0074] Then the closed form of the optimal disturbance variable is:

[0075]

[0076] Furthermore, let G SLP =W WL-ZF , the final transmission signal is expressed as:

[0077]

[0078] In order to further verify the effect of the symbol-level precoding method based on wide linear processing of the present invention, the following examples are used for illustration. For a more practical communication system, Monte Carlo simulation is used in this embodiment. First, the simulation comparison of the transmission power is performed for the number of users K = 2 to 8, and then the bit error rate of the multi-user multiple-input single-output (MISO) downlink for two-user MISO and four-user multi-user multiple-input single-output (MISO) is simulated and analyzed. BPSK signals are sent. For the convenience of comparison, the traditional zero-forcing precoding (denoted as ZF) and the common ZF-based SLP precoding (denoted as ZF-SLP) and the wide linear ZF SLP precoding proposed in the present invention (denoted as WL-ZF-SLP) are used.

[0079] Figure 2 The figure is a schematic diagram of the constellation point distribution of the wireless transmission system receiving the signal in the noise-free state, showing that WL-ZF-SLP produces a constructive interference effect different from that of ZF-SLP. ZF-SLP produces a strict phase constructive interference, while WL-ZF-SLP can produce a non-strict phase constructive interference effect, thus indicating that the present invention can improve the detection performance of the system.

[0080] Figure 3 is the average total power consumption of the transmitting antenna under different numbers of users. It can be seen from the figure that as the number of users increases, the average power consumption of the antenna shows an upward trend. In addition, due to the non-strict phase rotation effect introduced by WLP, the WL-ZF-SLP precoding method based on the present invention has the lowest power consumption. In addition, the power consumption gap between the WL-ZF-SLP precoding method and the other two precoding methods widens as the number of users increases. For situations where the number of users is relatively large, such as when K=8, the WL-ZF-SLP precoding method reduces the antenna transmission power by 19.8dB and 12.73dB respectively compared with the ZF and ZF-SLP precoders.

[0081] Figure 4The figure shows the bit error rate (BER) performance of the precoding algorithm under different user numbers. It can be seen from the figure that the WL-ZF-SLP of the present invention has a strong performance improvement compared with the other two precoding methods. -3 At this level, the WL-ZF-SLP precoding method can produce an SNR gain of about 15dB compared to the ZF-SLP and ZF precoding methods. As the number of users increases, the multi-user interference of the MISO system will further increase. For traditional precoders based on interference cancellation such as ZF precoding, its bit error rate performance will not be greatly affected. However, for the SLP algorithm, it achieves the purpose of utilizing the interference signal by manipulating the interference signal. As the number of users increases, the power of the interference signal that the SLP algorithm can utilize also increases further, so that the performance of the SLP algorithm can be further enhanced. In the case of four users, the WL-ZF-SLP precoder is superior to the ZF-SLP precoder and ZF precoder at a bit error rate of 10 -3 This level can produce 7dB and 17dB SNR gains.

[0082] Figure 5 This is the system bit error rate result under the same simulation configuration when using low-density parity check (LDPC) coding with a code rate of 1 / 3. It can be seen from the figure that after adopting LDPC channel coding, the performance of all precoders is significantly improved, but the performance gain of the WL-ZF-SLP precoding method based on the present invention relative to the ZF-SLP precoding method and the ZF precoding method is still very obvious.

Claims

1. A symbol-level precoding method based on wide linear processing, characterized in that: Applied to a wireless transmission system, wherein the wireless transmission system includes a base station and K users, the base station is equipped with N transmitting antennas, each user is equipped with a single antenna, and N is greater than K; The symbol level precoding method comprises: The base station obtains channel state information, and obtains a zero-forcing precoding matrix for wide linear processing according to the channel state information; Optimize the disturbance variable, precode each symbol in the transmission frame according to the optimized disturbance variable, and obtain the optimal disturbance variable by minimizing the transmission power; Under the optimal disturbance variable, the zero-forcing precoding matrix of wide linear processing is combined to obtain the final transmission signal.

2. The symbol-level precoding method based on wide linear processing according to claim 1, characterized in that: Optimizing the disturbance variable and precoding each symbol in the transmission frame according to the optimized disturbance variable includes: Defining the disturbance variable And precode each symbol in the transmission frame to obtain: x SLP =G SLP (s+u) In the formula, x SLP represents the transmitted symbol vector after symbol-level precoding, is the precoding matrix, s represents the symbol sequence; Introducing the equivalent symbol matrix And the diagonal matrix B, the elements in the diagonal matrix B correspond to each element of the symbol sequence s, then the symbol sequence s of the data is equivalent to: Introduce a real-valued perturbation vector Optimize the disturbance variables. Every element are all non-negative real numbers, satisfying the restriction that the sign amplitude increases and the phase remains unchanged, then the perturbation vector is equivalent to: Then the antenna's transmit symbol vector is re-expressed as:

3. The symbol-level precoding method based on wide linear processing according to claim 2, characterized in that: The optimal disturbance variables obtained by minimizing the transmission power include: An optimization function is constructed to minimize the transmit power of the antenna, where the optimization function is expressed as: my ||x SLP ||2 In the formula, || ||2 represents the bi-norm of the matrix; Introduce a temporary matrix A=G SLP B and vector The optimization function is expressed as: Using the Euclidean norm duality between complex vectors And the augmented vector v = [v1,v2,...,v i ,...,v M ] performs real number simplification on the optimization function, where any complex number is expressed as v i =[a i ,b i ], i = 1, ..., M represents the index value of the vector, M represents the dimension of the complex vector, a i is the real part of the vector, b i is the imaginary part of the vector, j represents the imaginary unit; then the optimization function is expressed as: In the formula, represents the real part of the matrix, represents the imaginary part of the matrix, Set the off-diagonal elements of the matrix product is 0, that is: Then the closed form of the optimal disturbance variable is:

4. The symbol-level precoding method based on wide linear processing according to claim 3, characterized in that: The zero-forcing precoding matrix for wide linear processing is: Where F represents the channel state information, and the superscript H represents the conjugate transpose.

5. The symbol-level precoding method based on wide linear processing according to claim 4, characterized in that: Let G SLP =W WL-ZF , the final transmission signal is expressed as:

Images