Method for mMIMO user activity detection and channel estimation based on compressive sensing

Abstract

The mMIMO user activity detection and channel estimation method based on compressive sensing disclosed in the application belongs to the technical field of wireless communication.The application realizes the method as follows: the large-scale MIMO technology is applied to a low-orbit satellite communication system, so that the satellite has the ability to implement flexible beamforming, can fully utilize the spatial degrees of freedom of large-scale MIMO, and significantly improve the spectral efficiency and power efficiency of the LEO satellite communication system.The channel estimation based on compressive sensing can maximize the reduction of the number of pilots by means of the sparsity of the channel in the transform domain, improve the spectral utilization rate, thereby reduce the dimension of the matrix and the algorithm complexity.The SOMP algorithm is used for sparse signal recovery, the SOMP algorithm uses a set of joint sparse signal residuals to select the most matched support set elements in each iteration, reduces the iteration number, reduces the calculation complexity, increases the accuracy of the support set element selection, and improves the recovery signal accuracy.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication technology and relates to a method for joint user activity detection and channel estimation based on compressed sensing for mMIMO LEO satellite IoT. Background Technology

[0002] In recent years, the rapid development of wireless communication transmission technology and the increasing popularity of smart communication devices have made massive multiple-input multiple-output (mMIMO) technology a key technology for satellite Internet of Things (IoT). This technology, by utilizing multiple transmit and receive antennas at both the transmitting and receiving ends, fully leverages space resources, effectively improving the spectrum utilization, information transmission rate and capacity of communication systems, and enhancing communication quality. Low Earth Orbit (LEO) satellites, with their significant advantages of high bandwidth, high throughput, and low data transmission latency, are particularly suitable for the communication field. Extending mMIMO technology to LEO satellite communication systems can fully utilize the spatial degrees of freedom of mMIMO, significantly improving the spectral efficiency and power efficiency of LEO satellite communication systems. However, LEO satellite IoT faces multiple challenges, including time-varying propagation delays caused by high-speed motion, huge Doppler shifts, and limited communication payloads. In order to meet the needs of large-scale massive connections while reducing costs and transmission delays, unlicensed multiple access technology is commonly used. However, this unlicensed multi-user access method also brings the problem of user activity detection, and the computational complexity of joint user activity detection, multi-user detection and channel estimation schemes is too high.

[0003] Massive MIMO technology is one of the key technologies for satellite IoT. However, the increasing number of base station antennas leads to a significant increase in the dimensionality of the channel matrix. Furthermore, in large-scale IoT access scenarios, due to the lack of device scheduling information at the receiver, it is necessary to identify all active users, resulting in excessively high computational complexity for joint user activity detection, multi-user detection, and channel estimation schemes. In low Earth orbit (LEO) satellite communication, the channel suffers from various interferences such as Doppler shift and multipath fading, severely impacting reception performance. Simultaneously, primary users occupy only a small portion of the bandwidth in actual communication, exhibiting sparse characteristics in the frequency domain and low spectrum utilization. For the multi-user detection problem in mMIMO LEO satellite IoT scenarios, traditional linear detection methods require high computational complexity, while limited satellite payload and computing power lead to extremely high overhead.

[0004] Therefore, given the inherent contradiction between massive mMIMO connectivity and the limited computing resources of LEO satellites, further exploration is needed on how to reduce computational complexity while accelerating convergence and improving signal recovery accuracy to support massive user connections.

[0005] In recent years, sparse signal recovery methods such as compressed sensing have been proven to achieve good channel information acquisition performance with relatively small pilot overhead, and are gradually being used by researchers for large-scale MIMO wireless channel estimation on land. Summary of the Invention

[0006] The purpose of this invention is to provide a compressed sensing-based mMIMO user activity detection and channel estimation method. In the mMIMO LEO satellite IoT scenario, the number of active users in each time slot is much smaller than the number of access users, and the user activity status is unknown in the prior knowledge. This method can detect user activity before channel estimation and multi-user detection, thereby reducing computational overhead.

[0007] The objective of this invention is achieved through the following technical solution.

[0008] This invention discloses a compressed sensing-based mMIMO user activity detection and channel estimation method. For signals sparse on a set of known bases, compressed sensing technology can directly sample at frequencies lower than the Nyquist sampling rate. When the number of antenna arrays is large, based on the potential sparsity of the channel matrix in the transform domain, compressed sensing theory can be used to transform the joint user activity detection and channel estimation problem into a multi-vector measurement problem. This problem is then solved using synchronous orthogonal matched pursuit, which has a fast convergence speed and high signal recovery accuracy, thus completing channel estimation. This method achieves good channel information acquisition performance with relatively small pilot overhead, significantly reduces the size of the sampled data that needs to be stored, and improves spectral efficiency.

[0009] The compressed sensing-based mMIMO user activity detection and channel estimation method disclosed in this invention includes the following steps:

[0010] Step 1: Establish an uplink transmission channel model for mMIMO LEO satellites.

[0011] Utilizing the spatiotemporal frequency distribution characteristics of mMIMO LEO satellite communication services, the channel is modeled based on a probability distribution model;

[0012] The satellite is equipped with N a =N x N y A uniform planar array UPA consisting of N receiving antennas x and N yLet be the number of antennas along the x-axis and y-axis, respectively. Within a given time slot, the LEO satellite simultaneously serves K single antennas. Each user transmits a short packet containing D bits of information through N channels CU, occupying a bandwidth of B and a delay of t. D And N = Bt D The pilot length is m, and the channel coding block length is M = Nm;

[0013] The uplink channel impulse response CIR between the LEO satellite and user K at time t is expressed as:

[0014]

[0015] in, L k Let a represent the number of multipaths for user k, l be the l-th propagation path for user k, and a k,l v k,l and τ ut k,l These represent the smaller transmission delays caused by channel gain, Doppler shift, and scattering angle, respectively. It is the array response vector on the satellite side;

[0016] Doppler shift in satellite communications is mainly caused by two parts: satellite-side shift and user-side shift. However, in satellite IoT scenarios, the Doppler frequency shift caused by user movement is much smaller than the Doppler frequency shift caused by satellite movement, i.e. The Doppler shift caused by the motion of the LEO satellite is the same for different propagation paths for the same user, that is... In addition, the transmission delay τ kl Represented as τ kl =τ kl sat +τ kl ut , where τ kl sat To account for the larger transmission delay caused by longer communication distances, τ kl ut To minimize transmission delay caused by factors such as scattering angle;

[0017] The signal transmitted after compensation at the transmitting end is represented as follows: Where v k cps =v k sat τ k cps =τ k sat The signal received by the satellite is represented as

[0018] When the user terminal is located in an unobstructed area, it can be equivalent to line-of-sight (LoS) propagation; otherwise, it will lead to non-LoS propagation caused by multipath effects. The unobstructed area includes suburbs and rooftops. Using the Rician channel model, the uplink channel is represented as...

[0019]

[0020] in It is the average channel power, κ k It is the Rician factor. and These represent the LosS propagation part and the non-LoS propagation part, respectively.

[0021] Step 2: After the user terminal sends the signal, the pilot signal is received using compressed sensing theory based on the mMIMO LEO satellite IoT uplink channel transmission model.

[0022] Based on the signal y received on the satellite side according to equation (1) cps The pilot signal is written using compressed sensing theory as follows:

[0023] Y (p) =HX (p) +Z (p) (3)

[0024] in Let K be the channel coefficient matrix between K users and the satellite. Its elements are independent and identically distributed. Additive complex Gaussian noise.

[0025] Step 3: In satellite IoT scenarios, user activity status is unknown prior to the data, so user activity detection needs to be performed before channel estimation and multi-user detection. Based on the user activity status, the pilot signal received in Step 2 is added to the user activity factor.

[0026] In step three, based on the user's activity level, the pilot signal received in step two is written as...

[0027] Y (p) =AX (p) +Z (p) (4)

[0028] in b k ∈{0,1} is the user activity factor, when b k =1 indicates that user k is active, and vice versa. This represents the set of active users.

[0029] Step 4: Addressing the sparsity of the user activity matrix and the CIR sparsity in the Internet of Things, the Y received in Step 3... (p) The sparse signal is recovered using the SOMP algorithm and then initialized.

[0030] In step four, the signal residual R (0) =Y (p) Support set The iteration count is i = 1. First, the inner product of multiple observation vectors and the measurement matrix is ​​calculated, and the multiple inner products are summed. Then, the inner product values ​​of different column numbers are compared, and the column number corresponding to the largest inner product value is selected as the first element of the support set.

[0031] Step 5: Based on the elements of the support set, obtain the least squares solution of multiple sparse vectors in the current iteration and calculate their respective residuals. In the next iteration, use the residuals from this iteration to calculate the inner product and obtain the support set elements for the next iteration, until the iteration ends.

[0032] In step five,

[0033] In the i-th iteration, multiple users are first selected using the maximum relevance criterion.

[0034]

[0035] Where C i =X H R (i-1) R (i) It is the residual signal of the i-th iteration; let η i As the active user set in this iteration, then using η i And the set of active users Λ selected in the (i-1)th iteration i-1 Form a new set Λ i .

[0036] Step 6: Use this set to select the i-th column corresponding to the channel response matrix A, and use the least squares method to solve for the channel estimation matrix corresponding to the selected active user set. pass Update residual signal R (i) And determine whether it satisfies or i≥i max , where i max This represents the maximum number of iterations. If the iteration condition is met, the iteration terminates; otherwise, iteration continues.

[0037] Step 7: The user activity detection problem is transformed into determining which columns in the channel response matrix A are zero vectors using the SMOP algorithm. The channel estimation problem is then transformed into determining which columns in the received signal Y are zero vectors. (p)The problem is to recover the channel response matrix A; the set selected through iteration is the set of active users detected by the SMOP algorithm. and channel estimation matrix This enables joint user activity detection and channel estimation.

[0038] Beneficial effects:

[0039] 1. The compressed sensing-based mMIMO user activity detection and channel estimation method disclosed in this invention extends the application of massive MIMO technology to low-Earth orbit satellite communication systems, enabling satellites to implement flexible beamforming, fully utilize the spatial degrees of freedom of massive MIMO, and significantly improve the spectral efficiency and power efficiency of LEO satellite communication systems.

[0040] 2. Compared with traditional pilot-based channel estimation algorithms, the compressed sensing-based mMIMO user activity detection and channel estimation method disclosed in this invention performs channel estimation based on compressed sensing. It can take advantage of the sparsity of the channel in the transform domain to maximize the reduction of the number of pilots and improve the spectrum utilization, thereby reducing the dimension of the matrix and the complexity of the algorithm.

[0041] 3. Compared with the traditional OMP algorithm, the mMIMO user activity detection and channel estimation method based on compressed sensing disclosed in this invention uses the SOMP algorithm for sparse signal recovery. In each iteration, the SOMP algorithm uses a set of residuals of the joint sparse signal to jointly select the best matching support set elements. By taking advantage of the joint sparsity of the signal, the number of iterations is reduced, the computational complexity is reduced, and the accuracy of support set element selection is increased, thereby improving the accuracy of the recovered signal. Attached Figure Description

[0042] Figure 1 This is a schematic diagram of the mMIMO LEO satellite communication system.

[0043] Figure 2 This is a schematic diagram of the pilot transmission strategy in the channel estimation method.

[0044] Figure 3 This is a flowchart of the mMIMO user activity detection and channel estimation method based on compressed sensing of the present invention. Detailed Implementation

[0045] The method for mMIMO user activity detection and channel estimation based on compressed sensing described in this invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0046] like Figure 3 As shown in the figure, the specific implementation steps of the compressed sensing-based mMIMO user activity detection and channel estimation method disclosed in this embodiment are as follows:

[0047] Step 1: Establish an uplink transmission channel model for mMIMO LEO satellites.

[0048] To address the issues of dynamic and variable satellite-to-ground links and the severe impact of Doppler frequency shift on network communication, and considering the spatiotemporal frequency distribution characteristics of mMIMOLEO satellite communication services, a probability distribution model is used to model the channel.

[0049] The satellite is equipped with N a =N x N y A uniform planar array (UPA) consisting of N receiving antennas. x and N y denoted by , respectively, the number of antennas along the x-axis and y-axis. Within a given time slot, the LEO satellite simultaneously serves K single antennas. Each user transmits a short packet containing D bits of information through N channels (Channel Use, CU), occupying a bandwidth of B and a delay of t. D And N = Bt D The pilot length is m, and the channel coding block length is M = Nm.

[0050] Unlike ground-based mMIMO systems, LEO satellite communication suffers from high transmission latency and is significantly affected by Doppler shift. Focusing on transmission latency and Doppler shift, the uplink channel impulse response (CIR) between the LEO satellite and user K at time t is expressed as:

[0051]

[0052] in, L k Let a represent the number of multipaths for user k, l be the l-th propagation path for user k, and a k,l v k,l and τ ut k,l These represent the smaller transmission delays caused by channel gain, Doppler shift, and scattering angle, respectively. It is the array response vector on the satellite side.

[0053] Doppler shift in satellite communications is mainly caused by two parts: satellite-side shift and user-side shift. However, in satellite IoT scenarios, the Doppler frequency shift caused by user movement is much smaller than the Doppler frequency shift caused by satellite movement, i.e. Since LEO satellites are at relatively high altitudes, it can be assumed that the Doppler shift caused by the motion of LEO satellites is the same for different propagation paths for the same user. In addition, the transmission delay τ kl It can be represented as τ kl =τ kl sat +τ kl ut , where τ kl sat For longer communication distances, the larger transmission delay τ kl ut This reduces the transmission delay caused by factors such as scattering angle.

[0054] To address the severe Doppler shift and high transmission delay issues in LEO satellite communication, the transmitted signal, after compensation at the transmitting end, is represented as follows: Where v k cps =v k sat τ k cps =τ k sat The signal received by the satellite can be represented as

[0055] When the user terminal is located in an unobstructed area such as a suburb or rooftop, it can be considered as line-of-sight (LoS) propagation. Conversely, it will lead to non-LoS propagation due to multipath effects. Using the Rician channel model, the uplink channel is represented as...

[0056]

[0057] in It is the average channel power, κ k It is the Rician factor. and These represent the LosS propagation part and the non-LoS propagation part, respectively.

[0058] Step 2: After the user terminal sends the signal, it receives the pilot signal based on the mMIMO LEO satellite IoT uplink channel transmission model.

[0059] Based on the signal y received on the satellite side according to equation (1) cps The pilot signal is written using compressed sensing theory as follows:

[0060] Y (p) =HX (p) +Z (p) (3)

[0061] in Let K be the channel coefficient matrix between K users and the satellite. Its elements are independent and identically distributed. Additive complex Gaussian noise.

[0062] Step 3: In satellite IoT scenarios, user activity status is unknown prior to the data, requiring user activity detection before channel estimation and multi-user detection. Based on the user activity status, the pilot signal received in Step 2 is written as...

[0063] Y (p) =AX (p) +Z (p) (4)

[0064] in b k ∈{0,1} is the user activity factor, when b k =1 indicates that user k is active, and vice versa. This represents the set of active users.

[0065] Step 4: Addressing the sparsity of the user activity matrix and the CIR sparsity in the Internet of Things, the Y received in Step 3... (p) The sparse signal is recovered using the SOMP algorithm.

[0066] Initialization: Signal residual R (0) =Y (p) Support set The iteration number i = 1. First, calculate the inner product of multiple observation vectors and the measurement matrix, and sum the multiple inner products. Then, compare the inner product values ​​of different column numbers, and select the column number corresponding to the largest inner product value as the first element of the support set.

[0067] Step 5: Perform iterations. Based on the elements of the support set, obtain the least squares solutions of multiple sparse vectors in the current iteration and calculate their respective residuals. In the next iteration, use the residuals from this iteration to calculate the inner product and obtain the support set elements for the next iteration, until the iteration ends.

[0068] In the i-th iteration, multiple users are first selected using the maximum relevance criterion.

[0069]

[0070] Where C i =X H R (i-1) R (i) It is the residual signal of the i-th iteration. Let η i As the active user set in this iteration, then using η i And the set of active users Λ selected in the (i-1)th iteration i-1 Form a new set Λ i .

[0071] Step Six: Next, use this set to select the i-th column corresponding to the channel response matrix A, and use the least squares method to solve for the channel estimation matrix corresponding to the selected active user set.

[0072]

[0073] Through the channel estimation matrix Update residual signal R (i) And determine whether it satisfies or i≥i max , where i max This represents the maximum number of iterations. If the iteration condition is met, the iteration terminates; otherwise, iteration continues.

[0074] Step 7: The user activity detection problem is transformed into determining which columns in the channel response matrix A are zero vectors using the SMOP algorithm. The channel estimation problem is then transformed into determining which columns in the received signal Y are zero vectors. (p) The problem is to recover the channel response matrix A; the set selected through iteration is the set of active users detected by the SMOP algorithm. and channel estimation matrix This means achieving joint user activity detection and channel estimation;

[0075] The above detailed description further illustrates the purpose and technical solution of the invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for mMIMO user activity detection and channel estimation based on compressed sensing, characterized in that: Includes the following steps, Step 1: Establish an uplink transmission channel model for mMIMO LEO satellites; Step 2: After the user terminal sends the signal, it uses compressed sensing theory to receive the pilot signal based on the mMIMO LEO satellite uplink transmission channel model. Step 3: In satellite IoT scenarios, the user activity status is unknown beforehand, so user activity detection needs to be performed before channel estimation and multi-user detection. Based on the user's activity level, the pilot signal received in step two is added to the user activity factor. In step three, Based on the user's activity level, the pilot signal received in step two is written as... （4） in , As a user activity factor, when Indicates user It is active, otherwise it is inactive. Represents the set of active users; Step 4: Addressing the sparsity of the user activity matrix and the CIR sparsity in the Internet of Things (IoT), the data received in Step 3... The signal is recovered using the SOMP algorithm and then initialized. Step 5: Solve the least squares solutions of multiple sparse vectors in the current iteration based on the support set elements and calculate their respective residuals. In the next iteration, use the residuals from this iteration to calculate the inner product and obtain the support set elements for the next iteration, until the iteration ends. Step 6: Use this support set to select the channel response matrix. The corresponding The channel estimation matrix corresponding to the selected active user set is obtained by using the least squares method. ,pass Update residual signal And determine whether it satisfies or ,in This represents the maximum number of iterations; if the iteration condition is met, the iteration terminates; otherwise, the iteration continues. Step 7: Transform the user activity detection problem into judging the channel response matrix using the SOMP algorithm. The question of which columns are zero vectors transforms the channel estimation problem into identifying the zero vectors in the received signal. Recovery of the channel response matrix The problem is that the set selected through iteration is the set of active users detected by the SOMP algorithm. and channel estimation matrix This means achieving joint user activity detection and channel estimation.

2. The method for mMIMO user activity detection and channel estimation based on compressed sensing as described in claim 1, characterized in that: The implementation method for step one is as follows: Utilizing the spatiotemporal frequency distribution characteristics of mMIMO LEO satellite communication services, the channel is modeled based on a probability distribution model; The satellite is equipped with A uniform planar array UPA consisting of 1 receiving antenna and Antenna edges shaft and The number of antennas on the axis; the number of LEO satellites simultaneously serving within a given time slot. Each user has a single antenna and accesses via... Each channel CU transmits a containing Short packets of bit information occupy bandwidth. The time delay is ,and The pilot length is The channel coding block length is ; LEO satellites and users Between moments The uplink channel impulse response (CIR) is expressed as: （1） in, , Indicates user The number of multipaths, For users The One transmission path, , and These represent the small transmission delays caused by channel gain, Doppler shift, and scattering angle, respectively. It is the array response vector on the satellite side; Doppler shift in satellite communications is mainly caused by two parts: satellite-side shift and user-side shift. However, in satellite IoT scenarios, the Doppler frequency shift caused by user movement is much smaller than the Doppler frequency shift caused by satellite movement, i.e. The Doppler shift caused by the motion of the LEO satellite is the same for different propagation paths for the same user, that is... In addition, transmission delay Represented as ,in To account for the large transmission delay caused by long communication distances, The small additional transmission delay is due to the scattering angle; The signal transmitted after compensation at the transmitting end is represented as follows: ,in , ; The signal received by the satellite is represented as ; When the user terminal is located in an unobstructed area, it can be equivalent to line-of-sight (LoS) propagation; otherwise, it will lead to non-LoS propagation caused by multipath effects. The unobstructed area includes suburbs and rooftops. Using the Rician channel model, the uplink channel is represented as... （2） in It is the average channel power. It is the Rician factor. and These represent the LosS propagation part and the non-LoS propagation part, respectively.

3. The method for mMIMO user activity detection and channel estimation based on compressed sensing as described in claim 2, characterized in that: Based on the signal received on the satellite side according to equation (1) The pilot signal is written using compressed sensing theory as follows: （3） in ,for Channel coefficient matrix between each user and the satellite , , Its elements are independent and identically distributed. Additive complex Gaussian noise.

4. The method for mMIMO user activity detection and channel estimation based on compressed sensing as described in claim 3, characterized in that: In step four, signal residual Support set The iteration number i=1; first, calculate the inner product of multiple observation vectors and the measurement matrix, and sum the multiple inner products. Then, compare the inner product values ​​of different column numbers, and select the column number corresponding to the largest inner product value as the first element of the support set.

5. The method for mMIMO user activity detection and channel estimation based on compressed sensing as described in claim 4, characterized in that: In step five, In the In this iteration, multiple users are first selected using the maximum relevance criterion. （5） in , It is the first The residual signal of the next iteration; As the active user set in this iteration, then utilize And the previous The set of active users selected in the next iteration Form a new set .

Images