A Pilot Contamination Suppression Method under De-Cellular Massive MIMO

Abstract

The present invention discloses a method for suppressing pilot contamination in a cellular-free massive MIMO, including: establishing an uplink superimposed pilot transmission model in the cellular-free massive MIMO system, and deriving the CSI between the AP and the UE based on the LMMSE criterion; constructing an MRC receiver using LMMSE channel estimation, and deriving a closed-form expression of the uplink spectral efficiency of the user according to the decoded signal expression; aiming at maximizing the total spectral efficiency of all users, designing a pilot allocation method based on enhanced tabu search. Specifically, according to the total spectral efficiency of the users, determine which pilot allocation scheme to add to the tabu list to prevent the algorithm from falling into a local optimal solution, and then obtain the pilot allocation scheme that optimizes the total spectral efficiency of the users through iterative update. The method for suppressing pilot contamination in a cellular-free massive MIMO proposed by the present invention can better suppress pilot contamination and improve the total spectral efficiency of the system, and has wide application value and application prospects.

Description

Technical Field

[0001] The present invention relates to a mobile communication system, and more particularly to a method for suppressing pilot contamination in a cellular-free massive MIMO system. Background Art

[0002] Cellular-free massive MIMO (Multiple-Input Multiple-Output) is a key technology in the future communication field. Compared with traditional massive MIMO, cellular-free massive MIMO technology emphasizes more on the user (User Equipment, UE) as the center. By deploying far more access points (Access Points, APs) than UEs, diversity is utilized to combat shadow fading, providing a higher coverage rate. Since the distance between the AP and the UE is reduced, the path loss is significantly reduced. All access points within the area can serve each user, which eliminates the inter-cell interference and solves the problem of poor performance of UEs at the cell edge. Given its advantages of high spectral efficiency, low cost, and easy deployment, cellular-free massive MIMO is regarded as a potential key technology in the future 6G communication system.

[0003] Generally, high-quality data transmission relies on high-precision channel state information (Channel State Information, CSI), which can be obtained through uplink training. Uplink training means that the UE sends a pilot sequence to the AP, and the AP estimates the CSI based on the received pilot sequence. However, since the number of UEs is often more than the number of orthogonal pilots, multiple UEs will reuse the same pilot, which will cause pilot contamination and seriously reduce the channel estimation accuracy. To address this problem, there are currently two mainstream solutions. First, a pilot allocation scheme is adopted in the traditional pilot transmission model; second, a superimposed pilot transmission model is used to increase the number of pilot transmissions.

[0004] However, these methods do not completely eliminate the pilot contamination problem. The traditional superimposed pilot transmission model only increases the number of pilot transmissions, but there is still a situation where the number of pilots is less than the number of UEs. Therefore, based on the superimposed pilot transmission model, an enhanced tabu search allocation algorithm is designed to further improve the system performance while increasing the number of pilots and making the same pilot as unlikely to be used multiple times as possible. Summary of the Invention

[0005] Object of the Invention: To overcome the deficiencies of the prior art, the present application designs a pilot allocation method under the cellular-free massive MIMO superimposed pilot transmission model, which preferably solves the serious pilot contamination problem caused by the shortage of orthogonal pilot resources when the number of UEs is greater than the coherence interval length within a certain coherence interval, resulting in users having to share non-orthogonal pilots.

[0006] Technical solution: A uplink superimposed pilot transmission model of a cellular-free massive MIMO system based on an enhanced tabu search allocation algorithm according to the present invention includes the following steps:

[0007] (1) In a cellular-free massive MIMO system, establish a superimposed pilot transmission model, and obtain the channel state information between users and access points based on the LMMSE estimation criterion;

[0008] (2) Design an MRC receiver using LMMSE channel estimation, and derive a closed-form expression for the uplink spectral efficiency of users according to the decoded signal expression;

[0009] (3) With the goal of maximizing the total spectral efficiency of all users, design a pilot allocation method based on enhanced tabu search. Specifically, determine which pilot allocation scheme to add to the tabu list according to the total spectral efficiency of users to prevent the algorithm from falling into a local optimal solution, and then obtain the pilot allocation scheme that optimizes the total spectral efficiency of users through iterative update.

[0010] Further, it includes:

[0011] The specific steps of step (1) include:

[0012] Assume that all UEs send data signals and pilot signals to the AP simultaneously. The signal received by the l-th AP (i.e., AP l ) is:

[0013]

[0014] where l = 1, 2,..., L, k = 1, 2,..., K, L is the number of APs, and K is the number of users. g lk is the channel vector between the AP l and the k-th user (i.e., UE k ), P p is the transmit power allocated to send pilots, P u is the transmit power allocated to send data signals, P is the total transmit power of the user, and P = P p + P u . and represent the pilot signal and the data signal respectively. The superscript H represents the conjugate transpose operation. In addition, N l ∈ C N×τ is a Gaussian white noise matrix, where C represents the complex number field, N is the number of antennas of a single AP, and τ is the length of the pilot coherence interval.

[0015] To estimate g lk , multiply the received signal Y l on the right by Based on Y lAnd the channel g is obtained using the LMMSE criterion lk The LMMSE estimate of where

[0016]

[0017] β lk represents the large-scale fading coefficient between the AP l and the UE k and ρ k is the set using the same pilot

[0018] Furthermore, it includes:

[0019] Step (2) includes:

[0020] Using to construct an MRC receiver, and adopting the UatF technology to derive a closed-form expression for the total spectral efficiency of the system, which is:

[0021]

[0022] where B is the channel bandwidth

[0023] Furthermore, it includes:

[0024] Step (3) specifically includes:

[0025] S1 Determine whether the length of the superimposed pilot is greater than or equal to 1 / 2K;

[0026] S1.1 If it is greater, allocate pilots to the first 1 / 2 of the users in sequence, and randomly allocate pilots to the remaining users;

[0027] S1.2 Otherwise, select this new K-means clustering method, and the specific steps are as follows:

[0028] S1.2.1 First, let x = ceiling(user number / superimposed pilot length), divide the users into x groups, and initialize x centroids;

[0029] S1.2.2 Calculate the distance of each user from all centroids, select the centroid with the closest distance, and divide this user into the clustering of this centroid. Average the distances of all users in each clustering, recalculate the centroid, and re-divide the clustering;

[0030] S1.2.3 Allocate pilots to each clustering. If the number of users in the clustering is less than or equal to the number of pilots, allocate them in sequence, otherwise allocate pilots to the first number of leading pilots in sequence, and randomly allocate pilots to the remaining users as the initial pilot allocation scheme;

[0031] Set the tabu list to empty, set the tabu length to infinity, initialize the iteration number \(i = 1\) and \(m = 1\). Calculate the spectral efficiency of each user and sort them in ascending order of spectral efficiency;

[0032] S3 Use the enhanced tabu search allocation algorithm to generate the spectral efficiency of each user in the corresponding neighborhood, and select the neighborhood with the largest total spectral efficiency of users. The specific steps are as follows:

[0033] S3.1 Allocate a pilot with the least number of uses to UE m and call this neighborhood the \(m\)-th neighborhood. If there are two or more pilots with the least number of uses, we choose the pilot allocation scheme that maximizes the total spectral efficiency;

[0034] S3.2 Judge whether \(m\) is less than \(K\);

[0035] S3.2.1 If it holds, \(m = m + 1\); continue to execute step S3.1;

[0036] S3.2.2 If it does not hold, calculate the total spectral efficiency of all neighborhoods;

[0037] S4 Judge whether the total spectral efficiency of S3 is the largest total spectral efficiency in the previous \(i\) iterations;

[0038] S4.1 If it is the maximum value in the previous \(i\) iterations, then for the next iteration, regardless of whether it exists in the tabu list, select the pilot allocation scheme that maximizes the total spectral efficiency as the initialization pilot allocation scheme for the next iteration, and reduce the tabu length by 1. Next, judge whether this pilot allocation scheme already exists in the tabu list. If it does not exist in the tabu list, add it and modify the tabu length to 5. If it already exists, re-initialize the tabu length to 5;

[0039] S4.2 If it is not the maximum value in the previous \(i\) iterations, then judge whether the pilot allocation scheme that maximizes the total spectral efficiency in this iteration exists in the tabu list. If it already exists, we judge whether the pilot allocation scheme with the second largest total spectral efficiency exists in the tabu list until we find a pilot allocation scheme that does not exist in the tabu list. Reduce the tabu length by 1, add this pilot allocation scheme to the tabu list and set the tabu length to 5, and use this pilot allocation scheme as the initialization pilot allocation scheme for the next iteration;

[0040] S5 Judge whether \(i\) is less than the maximum number of iterations;

[0041] S5.1 If it does not hold, let \(i = i + 1\), \(m = 1\), and repeat steps (S3)-(S4);

[0042] If S5.2 holds, select the pilot allocation scheme with the maximum total spectral efficiency among all iterations as the final pilot allocation scheme of this algorithm.

[0043] Beneficial effects:

[0044] Compared with the prior art, the remarkable advantages of the present invention are as follows: First, the present invention derives a closed - form expression of the uplink total rate based on the MRC receiver under the superimposed pilot transmission model; then, under the superimposed pilot transmission model, the enhanced tabu search allocation algorithm is used to optimize the pilot allocation, and the optimal pilot allocation scheme that maximizes the total spectral efficiency is solved. The transmission method designed by the present invention can greatly improve the total spectral efficiency of the system and optimize the spectral efficiency of users with low spectral efficiency, having broad application value and application prospects. Description of the drawings

[0045] Figure 1 It is the flowchart of the method described in the embodiment of the present invention;

[0046] Figure 2 It is the relationship diagram between the uplink total spectral efficiency of the system described in the embodiment of the present invention and different coherence intervals. Detailed implementation manners

[0047] The technical method of the present invention will be described in detail below.

[0048] As Figure 1 shown, the main technical problem of the present invention is to propose an uplink superimposed pilot transmission method for a cellular - free massive MIMO system based on an enhanced tabu search allocation algorithm. The designed transmission method can optimize the pilot allocation method while increasing the number of available pilots, so as to achieve the purpose of optimizing the total spectral efficiency of the system. The following further details the present invention with reference to the accompanying drawings of the specification.

[0049] Step (1): In the cellular - free massive MIMO scenario, establish a transmission model under the superimposed pilot scheme, and adopt LMMSE to obtain the channel estimation information between users and access points in the cellular - free massive MIMO communication scenario

[0050] The present invention studies the uplink of a cellular - free massive MIMO system. The considered cellular - free massive MIMO system has M APs, K UEs, and 1 CPU. Each AP is equipped with N antennas, and each UE is equipped with a single antenna.

[0051] Assume that all UEs simultaneously send data signals and pilot signals to the APs. The signal received by the l - th AP (i.e., AP l ) is:

[0052]

[0053] where \(l = 1, 2, \ldots, L\), \(k = 1, 2, \ldots, K\), \(L\) is the number of APs, and \(K\) is the number of users. \(g\) lk is the channel vector between the \(l\)-th AP l and the \(k\)-th user (i.e., UE k ). \(P\) p is the transmit power allocated to the pilot transmission, \(P\) u is the transmit power allocated to the data signal transmission, \(P\) is the total transmit power of the user, and \(P = P\) p + \(P\) u . and represent the pilot signal and the data signal respectively. The superscript \(H\) represents the conjugate transpose operation. In addition, \(N\) l \(\in \mathbb{C}\) N×τ is the Gaussian white noise matrix, where \(\mathbb{C}\) represents the complex domain, \(N\) is the number of antennas of a single AP, and \(\tau\) is the length of the pilot coherence interval.

[0054] To estimate \(g\) lk , the received signal \(Y\) l is right-multiplied by Based on \(Y\) l and the LMMSE criterion is used to obtain the LMMSE estimate of the channel \(g\) lk as where

[0055]

[0056] \(\beta\) lk represents the large-scale fading coefficient between the \(l\)-th AP l and the UE k , and \(\rho\) k is the set of users using the same pilot.

[0057] Step (2) includes:[[]]

[0058] Using to construct an MRC receiver and adopting the UatF technique to derive a closed-form expression for the total spectral efficiency of the system, which is:[[]]

[0059]

[0060] where \(B\) is the channel bandwidth,

[0061] Step (3) specifically includes:[[]]

[0062] S1 Judge whether the length of the superimposed pilot is greater than or equal to \(1 / 2K\);

[0063] S1.1 If it is greater, allocate pilots to the first \(1 / 2\) of the users in sequence, and randomly allocate pilots to the remaining users;

[0064] S1.2 Otherwise, select this new K-means clustering method. The specific steps are as follows:

[0065] S1.2.1 First, let x = ceiling(user number / superimposed pilot length), divide users into x groups, and initialize x centroids;

[0066] S1.2.2 Calculate the distances from each user to all centroids, select the centroid with the shortest distance, and assign this user to the cluster of this centroid. Calculate the average of the distances of all users in each cluster, recalculate the centroid, and reassign the clusters;

[0067] S1.2.3 Allocate pilots for each cluster. If the number of users in the cluster is less than or equal to the number of pilots, allocate them sequentially. Otherwise, allocate pilots to the first several users sequentially, and randomly allocate pilots to the remaining users as the initial pilot allocation scheme;

[0068] S2 Set the taboo list to empty, set the taboo length to infinity, initialize the iteration number i = 1, m = 1. Calculate the spectral efficiency of each user and sort them in ascending order of spectral efficiency;

[0069] S3 Use the enhanced taboo search allocation algorithm to generate the spectral efficiency of each user in the corresponding neighborhood, and select the neighborhood with the largest total spectral efficiency of users. The specific steps are as follows:

[0070] S3.1 Allocate a pilot with the least number of uses to UE m and call this neighborhood the m-th neighborhood. If there are two or more pilots with the least number of uses, we select the pilot allocation scheme that maximizes the total spectral efficiency;

[0071] S3.2 Determine whether m is less than K;

[0072] S3.2.1 If it holds, m = m + 1; continue to execute step S3.1;

[0073] S3.2.2 If it does not hold, calculate the total spectral efficiency of all neighborhoods;

[0074] S4 Determine whether the total spectral efficiency of S3 is the largest total spectral efficiency in the previous i iterations;

[0075] S4.1 If it is the maximum value in the previous i iterations, then in the next iteration, regardless of whether it exists in the taboo list, select the pilot allocation scheme that maximizes the total spectral efficiency as the initial pilot allocation scheme for the next iteration, and reduce the taboo length by 1. Next, determine whether this pilot allocation scheme already exists in the taboo list. If it does not exist in the taboo list, add it and modify the taboo length to 5. If it already exists, re-initialize the taboo length to 5;

[0076] S4.2 If it is not the maximum value in the previous i iterations, then determine whether the pilot allocation scheme that maximizes the total spectral efficiency in this iteration exists in the taboo list. If it already exists, we then determine whether the pilot allocation scheme with the second largest total spectral efficiency exists in the taboo list, until we find a pilot allocation scheme that does not exist in the taboo list. Decrease the taboo length by 1, add this pilot allocation scheme to the taboo list and set the taboo length to 5, and use this pilot allocation scheme as the initial pilot allocation scheme for the next iteration;

[0077] S5 Determine whether i is less than the maximum number of iterations;

[0078] S5.1 If not, let i = i + 1, m = 1, and repeat steps S3 - S4;

[0079] S5.2 If so, select the pilot allocation scheme with the largest total spectral efficiency in all iterations as the final pilot allocation scheme of this algorithm.

[0080] The performance of the technical solution of the present invention is further described below in conjunction with simulation experiments.

[0081] Figure 1 A relationship diagram between the total uplink spectral efficiency of the system and different coherence intervals is given, where the abscissa is different coherence intervals and the ordinate is the total spectral efficiency of the system. Figure 1 The simulation parameters in [reference] are set as: L = 100, N = 8, B = 20 MHz, K = 65, τ sp varies with the change of the coherence interval, and τ sp is always equal to the length of the coherence interval, τ rp = 0.3τ sp。The red curve is the total spectral efficiency of the uplink superimposed pilot transmission model in the cell-free massive MIMO system based on the enhanced tabu search allocation algorithm. The black curve is the tabu search algorithm under the superimposed pilot transmission model. Obviously, compared with other single pilot contamination suppression methods, such as the blue superimposed pilot transmission model (Zhang Y, Qiao X, Yang L, et al. Superimposed pilots are beneficial for mitigating pilot contamination in cell-free massive MIMO[J]. IEEE Communications Letters, 2020, 25(1): 279-283.) and the pink tabu search allocation algorithm under the traditional pilot transmission model (Liu H, Zhang J, Zhang X, et al. Tabu-search-based pilot assignment for cell-free massive MIMO systems[J]. IEEE Transactions on Vehicular Technology, 2019, 69(2): 2286-2290.), it brings more significant performance improvement. At the same time, the uplink superimposed pilot transmission model of the cell-free massive MIMO system under the enhanced tabu search allocation algorithm of the red curve compared with the black curve that directly applies the tabu search allocation algorithm to the superimposed pilot transmission model further improves the total spectral efficiency, enhances the user experience, and has broad application prospects and use value.

Claims

1. A method for suppressing pilot contamination in cellular massive MIMO, characterized in that The specific steps are as follows: (1) In a cellular massive MIMO system, establish a superimposed pilot transmission model, and obtain the channel state information between users and access points based on the LMMSE channel estimation criterion; (2) Use LMMSE channel estimation to design an MRC receiver, and derive a closed-form expression for the total spectral efficiency of the system; (3) With the goal of maximizing the total spectral efficiency of all users, design a pilot allocation method based on enhanced tabu search; Determine which pilot allocation scheme to add to the tabu list according to the total spectral efficiency of users to prevent the algorithm from falling into a local optimal solution, and then obtain the pilot allocation scheme that optimizes the total spectral efficiency of users through iterative updates; Step (3) specifically includes: S1 Judge whether the length of the superimposed pilot is greater than or equal to 1 / 2K; S1.1 If it is greater, sequentially allocate pilots to the first 1 / 2 of the users, and randomly allocate pilots to the remaining users; S1.2 Otherwise, select a new K-means clustering method; S2 Set the tabu list to be empty, set the tabu length to infinity, initialize the iteration number i = 1, m = 1; calculate the spectral efficiency of each user, and sort them from smallest to largest according to the spectral efficiency; S3 Use the enhanced tabu search allocation algorithm to generate the spectral efficiency of each user in the corresponding neighborhood, and select the neighborhood with the largest total spectral efficiency of users; S4 Judge whether the total spectral efficiency in step S3 is the largest total spectral efficiency in the previous i iterations; S5 Judge whether i is less than the maximum number of iterations; S5.1 If not, let i = i + 1, m = 1, and repeat steps S3 - S4; S5.2 If so, select the pilot allocation scheme with the largest total spectral efficiency in all iterations as the final pilot allocation scheme of this algorithm; Step S4 specifically includes: S4.1 If it is the maximum value in the previous i iterations, then in the next iteration, regardless of whether it exists in the tabu list, select the set of pilot allocation schemes that maximizes the total spectral efficiency as the initialization pilot allocation scheme for the next iteration, and reduce the tabu length by 1. Next, judge whether this pilot allocation scheme already exists in the tabu list; if it does not exist in the tabu list, add it and modify the tabu length to 5. If it already exists, re-initialize the tabu length to 5; S4.2 If it is not the maximum value in the previous i iterations, then judge whether the pilot allocation scheme that maximizes the total spectral efficiency in this iteration exists in the tabu list. If it already exists, judge whether the pilot allocation scheme with the second largest total spectral efficiency exists in the tabu list until a pilot allocation scheme that does not exist in the tabu list is found. Reduce the tabu length by 1, add this pilot allocation scheme to the tabu list and set the tabu length to 5, and use this pilot allocation scheme as the initialization pilot allocation scheme for the next iteration.

2. The method for suppressing pilot contamination in a de-cellularized massive MIMO according to claim 1, wherein, The specific content of step (1) includes: Assume that all UEs send data signals and pilot signals to the AP simultaneously. The l-th AP, i.e., AP l receives the following signals: where \(l = 1,2,\cdots,L\), \(k = 1,2,\cdots,K\); \(L\) is the number of APs, \(K\) is the number of users; \(g\) lk is the AP l and the \(k\)-th user, i.e., UE k channel vector between them, \(P\) p is the transmit power allocated to transmit pilots, \(P_u\) is the transmit power allocated to transmit data signals, \(P\) is the total transmit power of the user, and \(P = P\) p + \(P\) u ; and represent the pilot signal and the data signal respectively, the superscript \(H\) represents the conjugate transpose operation. In addition, \(N\) l \(\in\mathbb{C}\) N×τ is the Gaussian white noise matrix, where \(\mathbb{C}\) represents the complex number field, \(N\) is the number of antennas of a single AP, and \(\tau\) is the length of the pilot coherence interval; To estimate g lk , multiply the received signal Y l on the right by Based on Y l and use the LMMSE criterion to obtain the LMMSE estimate of the channel g lk as where β lk represents the large-scale fading coefficient between the AP l and the UE k , and ρ k is the set using the same pilot.

3. The method for suppressing pilot contamination in a de-cellularized massive MIMO according to claim 2, wherein, The new K-means clustering method includes: S1.2.1 First, let x = the number of users / the length of the superimposed pilot, round x up, divide the users into x groups, and initialize x centroids; S1.2.2 Calculate the distances from each user to all centroids, select the centroid with the shortest distance, and divide this user into the cluster of this centroid. Calculate the average of the distances of all users in each cluster, recalculate the centroid, and re-divide the clusters; S1.2.3 Assign pilots to each cluster. If the number of users in the cluster is less than or equal to the number of pilots, assign them sequentially. Otherwise, assign pilots sequentially to the first number of pilots of users, and the remaining users are randomly assigned pilots as the initial pilot assignment scheme.

4. The method for suppressing pilot contamination in a cell-free massive MIMO according to claim 3, characterized in that, The specific steps of the enhanced tabu search allocation algorithm are as follows: S3.1 is for the UE m Allocate a pilot with the least number of uses, and call this neighborhood the m-th neighborhood. If there are two or more pilots with the least number of uses, select the pilot allocation scheme that maximizes the total spectral efficiency; S3.2 Determine whether m is less than K; S3.2.1 If it holds, m = m + 1; continue to execute step S3.1; S3.2.2 If it does not hold, calculate the total spectral efficiency of all neighborhoods.

Images