Two-dimensional orthogonal graph division multiple access communication method for massive MIMO systems

Abstract

A two-dimensional orthogonal graph multiple access communication method for massive MIMO systems is proposed and applied to massive MIMO cellular mobile communication systems. It utilizes the ideal autocorrelation and cross-correlation characteristics of frequency hopping sequences to eliminate interference caused by multipath propagation and Doppler shift, as well as interference between users. For massive MIMO systems, channel estimation for equalization is eliminated, reducing the occupation of pilot resources and thus suppressing pilot pollution to a certain extent. Since equalization and channel estimation for equalization are unnecessary, receiver complexity is reduced, and pilot pollution is mitigated to some extent, allowing the superior performance of massive MIMO systems to be fully realized.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a two-dimensional orthogonal graph division multiple access communication method for a large-scale MIMO system. Background Art

[0002] Nowadays, 5G cellular mobile communication technology is gradually replacing 4G and becoming the mainstream technology of future mobile communications. Therefore, the research on 5G key technologies has become particularly important, and massive MIMO technology is one of the key technologies to improve channel capacity and spectrum efficiency, and is also an indispensable part of promoting the application of 5G technology.

[0003] Massive MIMO technology is an expansion and extension of traditional MIMO technology. In the 4G era, traditional MIMO technology systems usually use 4 or 8 antenna channels, while massive MIMO technology can achieve 32 channels, 64 channels or even more antenna channels in 5G, greatly expanding the capacity of the system. By deploying a large number of antennas in the base station, massive MIMO technology allows multiple users to communicate simultaneously in the same time-frequency resources, thereby significantly improving spectrum efficiency. In addition, the deployment of a large number of antennas also increases the array gain, reduces the transmission power required by users, and improves the energy efficiency of the system. The above advantages make massive MIMO technology occupy a key position in 5G and become the focus of researchers.

[0004] However, everything has two sides. In order to give full play to the above advantages of large-scale MIMO technology, it is necessary to obtain accurate channel state information (CSI), which relies on sending pilot signals for channel estimation. Due to the large number of antennas, a large number of channel estimates need to be performed, which increases the complexity of the receiving end. In addition, since CSI remains unchanged within a coherent time, it needs to be re-estimated after exceeding the coherent time interval. The coherent time interval is limited, and the pilot sequence is also limited. When the number of user devices exceeds the pilot resources, different devices will reuse the same pilot, resulting in some pilot sequences being reused and causing fading. Therefore, when the base station receives the pilot signal of the desired channel, it will also receive the pilot signal of the non-desired channel that reuses the pilot sequence, resulting in a decrease in the accuracy of channel estimation, that is, pilot contamination. Therefore, in the large-scale MIMO system, although the increase in the number of antennas at the base station end can bring corresponding gains, due to the gain loss caused by pilot contamination, after the system performance reaches a certain value, increasing the number of antennas again will not make it continue to improve. Pilot contamination has become one of the main reasons restricting the performance of large-scale MIMO systems. In addition, in large-scale MIMO systems, although traditional one-dimensional multiple access technology can solve the delay interference caused by multipath through diversity technology, equalization and channel estimation, it has no effective solution to the interference caused by Doppler frequency shift. Therefore, the interference caused by multipath propagation and Doppler frequency shift will also affect the communication quality.

[0005] In the prior art, patent CN114039625A, for example, introduces in detail the graph division multiple access technology based on the optimal frequency hopping graph and the construction method of the optimal frequency hopping graph based on the Golomb Costas sequence, but does not involve large-scale MIMO systems. The large increase in the number of users in the cell of a large-scale MIMO system has led to the reuse of a part of the pilots, resulting in pilot pollution, which restricts the system performance. In addition, since a series of factors such as pilot pollution affect the accuracy of channel estimation, the accuracy of the channel information obtained by the base station is reduced, which greatly affects the performance of large-scale MIMO. The large-scale antenna array makes the large-scale MIMO system have more stringent requirements on the deployment of the site and the density of base stations. At the same time, the interference caused by the small distance between antennas is also a new challenge. Large-scale MIMO has higher requirements for precoding and signal detection, and it is particularly critical to balance system performance and computational complexity. For large-scale MIMO systems, their graph division multiple access technology and the optimal frequency hopping pattern construction method need to be designed separately. Otherwise, the current large-scale MIMO systems have the problem of requiring a large amount of channel estimation and extremely complex systems, and cannot enter the real practical application stage. There is no relevant design in the current field to solve the above problems, which has caused the development of large-scale MIMO systems to stagnate. This situation urgently needs to be solved. Summary of the invention

[0006] In view of the problems existing in the above-mentioned background technology, the present invention proposes a two-dimensional orthogonal graph division multiple access communication method for a large-scale MIMO system. Based on the two-dimensional orthogonal graph division multiple access technology, an optimal frequency hopping pattern is designed for users in the large-scale MIMO system to realize orthogonal graph division multiple access communication, so as to creatively solve the technical problems faced by the large-scale MIMO system, and further provide technical support for the practical application of the large-scale MIMO technology.

[0007] A two-dimensional orthogonal graph division multiple access communication method for a large-scale MIMO system includes the following steps:

[0008] Step 1: For a massive MIMO system, obtain its basic data, including the number of cells in a cell group N1, the number of base station receiving antennas in each cell N2, the number of optimal frequency hopping pattern families used by each base station receiving antenna N3, the number of users in each cell M1 and the number of transmitting antennas at each user end M2, and calculate the reuse distance D of the frequency hopping pattern;

[0009] Step 2, determining the required optimal frequency hopping pattern family number S according to the number of cells N1 in a cell group, the number of base station receiving antennas N2 in each cell, and the number of optimal frequency hopping pattern families N3 used by each receiving antenna of the base station, and determining the number S1 of optimal frequency hopping patterns in each optimal frequency hopping pattern family according to the number of users M1 in each cell, the number of transmitting antennas M2 of each user terminal, the number of base station receiving antennas N2 in each cell, and the number N3 of optimal frequency hopping pattern families used by each receiving antenna of the base station;

[0010] Step 3: Based on the maximum Doppler shift d in the massive MIMO system max and the maximum delay τ within the cell max , obtain the minimum Doppler distance w between the best frequency hopping pattern families and the minimum delay distance w1 between the best frequency hopping patterns in each best frequency hopping pattern family;

[0011] Step 4, according to the number of optimal frequency hopping pattern families s, the number of optimal frequency hopping patterns in each optimal frequency hopping pattern family s1, the minimum Doppler distance w between the optimal frequency hopping pattern families, and the minimum delay distance w1 between the optimal frequency hopping patterns in each optimal frequency hopping pattern family obtained in the previous step, the range of the maximum number of frequency hopping patterns (q-1) in each family is obtained, wherein q is a prime number, that is, the optimal frequency hopping pattern is constructed based on a prime domain, at this time q=p, p is the maximum number of time slots of the optimal frequency hopping pattern, and according to the frequency domain width B and the time domain width T of the signal in the large-scale MIMO system, the frequency slot F is obtained. b With time slot T b , construct the q-2 order Golomb Costas sequence C according to the value of q-1, and determine its placement function;

[0012] Step 5: Assume that the frequency hopping operator of the lth optimal frequency hopping pattern of massive MIMO is s l (k), when s l When (k) is the placement function of the M=q-1 order optimal frequency hopping pattern, the complex envelope of the signal frequency hopping in the kth time slot corresponding to the lth optimal frequency hopping pattern of massive MIMO is

[0013]

[0014] in,

[0015]

[0016] If the kth time slot of the best frequency hopping pattern is a gap column, then a l,k =0, otherwise a l,k =1; f l,k =s l (k)F b , when θ l,k When (t) = 0, it indicates that the sub-pulse is the optimal frequency hopping pattern signal with constant carrier frequency;

[0017] Step 6: construct the frequency hopping pattern of the head of the best frequency hopping pattern family, add an empty column and an empty row to the rightmost and topmost of sequence C to obtain sequence C1; cyclically shift sequence C1 vertically by w, 2w, …, (s-1)w rows to obtain sequences C2, C3, …, C s , sequence C1, C2, ..., C s That is, the head of each optimal frequency hopping pattern family; the head frequency hopping patterns of each family are circularly shifted to the left horizontally by w1, 2w1, …, (s1-1)w1 columns, and s1-1 optimal frequency hopping patterns with one gap column and one gap row are obtained. The first frequency hopping pattern in each family is the head frequency hopping pattern of the family, and w1 satisfies

[0018] Step 7, applying the constructed Golomb Costas sequence to the massive MIMO system, so that the system's demand for pilot signals is reduced, thereby suppressing pilot contamination, while suppressing interference caused by multipath propagation and Doppler frequency shift and interference between users, reducing receiver complexity, and realizing multiple access communication of the massive MIMO system.

[0019] The beneficial effects achieved by the present invention are:

[0020] (1) Applied to large-scale MIMO cellular mobile communication systems, experimental verification shows that this method can effectively suppress interference caused by multipath propagation and Doppler frequency shift and interference between users by utilizing the ideal autocorrelation and cross-correlation characteristics of the frequency hopping sequence, thereby greatly improving the reliability of communication.

[0021] (2) For massive MIMO systems, the channel estimation for equalization is eliminated, which reduces the occupancy of pilot resources and can suppress pilot pollution to a certain extent.

[0022] (3) Since there is no need for equalization and channel estimation for equalization, the complexity of the receiver is reduced and pilot contamination is alleviated to a certain extent, so that the excellent performance of the large-scale MIMO system can be fully utilized.

[0023] (4) It solves the problem of requiring a large amount of channel estimation and extremely complex system in current large-scale MIMO systems, simplifies the receiver, reduces interference, and enables large-scale MIMO systems to be truly applied in practice. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 The figure is a schematic diagram of uplink and downlink in a massive MIMO system in an embodiment of the present invention.

[0025] Figure 2 Schematic diagram of Golomb Costas sequence C in an embodiment of the present invention.

[0026] Figure 3 It is the first image of the optimal frequency hopping pattern family in the embodiment of the present invention.

[0027] Figure 4 It is the optimal frequency hopping pattern within the cluster in the embodiment of the present invention.

[0028] Figure 5 FIG. 2 is a schematic diagram of orthogonal graph division multiple access in a massive MIMO system in an embodiment of the present invention.

[0029] Figure 6 Schematic diagram of the bit error rate of a single user of OPDMA in an embodiment of the present invention.

[0030] Figure 7 FIG. 4 is a schematic diagram of OPDMA multi-user bit error rate in an embodiment of the present invention.

[0031] Figure 8 It is a block diagram of an orthogonal graph division multiple access system in an embodiment of the present invention. DETAILED DESCRIPTION

[0032] The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings.

[0033] The Costas sequence is a frequency hopping sequence with ideal autocorrelation characteristics. Although the Costas sequence performs well in terms of autocorrelation, it does not have ideal cross-correlation characteristics. Therefore, in scenarios where multiple frequency hopping patterns with ideal autocorrelation and cross-correlation characteristics are required, especially when a large number of such frequency hopping patterns are required in mobile communication systems, the Costas sequence cannot meet the requirements.

[0034] The present invention proposes the concept of optimal frequency hopping patterns (OFHPs), and proposes a method of constructing an optimal frequency hopping pattern with ideal autocorrelation and cross-correlation using a two-dimensional cyclic shift method, and applies the optimal frequency hopping pattern to a large-scale MIMO system for the first time, so that users in a mobile communication system can reduce or avoid the interference caused by multipath propagation and Doppler frequency shift and the influence of interference between users, and more effectively solve the problems of increased receiver complexity and pilot pollution. In this way, orthogonal pattern division multiple access (OPDMA) communication based on the optimal frequency hopping pattern is realized.

[0035] The performance of a frequency hopping communication system depends on the frequency hopping pattern, so designing a good frequency hopping pattern is the key. A good frequency hopping pattern should have the following characteristics: (1) Strong confidentiality. The more frequency hopping patterns there are and the difficulty of deciphering them is high enough, the stronger the anti-interception and anti-interference capabilities of the communication system will be, and the more secure the information transmitted will be. (2) Randomness. The more random the pattern, the stronger the anti-interference capability. (3) Good orthogonality, that is, try to ensure that there is no frequency overlap between patterns.

[0036] The optimal frequency hopping pattern is based on the algebraically constructed Costas sequence. The algebraic construction methods of Costas sequences include the Welch method and the Golomb method. The Welch construction method is based on the prime field, while the Golomb construction method can be based on both the prime field and the extended field. The following first focuses on how to construct the optimal frequency hopping pattern based on the Golomb Costas sequence constructed in the prime field for the users in the architecture design of the massive MIMO system to implement the orthogonal graph division multiple access technology.

[0037] The steps to construct the optimal frequency hopping pattern based on the prime field Golomb Costas sequence are as follows:

[0038] Suppose the restricted domain is GF(q), where q=p m , m is a positive integer, p is a prime number, let α and β be two primitive elements of GF(q), and sequence C be a permutation matrix of order N = q-2. Then the necessary and sufficient condition that sequence C is a Golomb Costas sequence is that the placement function of sequence C is:

[0039] y(k)=log β (1-α k ),1≤k≤q-2 (1)

[0040] Assume that the cell coordinates of sequence C are (i, j). When equation (2) is satisfied, the cell can be turned black:

[0041] α i +β j =1 (2)

[0042] When m=1, the Golomb Costas sequence is constructed based on the prime field, and equation (2) represents the modulo p congruence. The maximum autocorrelation sidelobe of the autocorrelation function of the Golomb Costas sequence is 1, which has an ideal autocorrelation property.

[0043] Then, the corresponding optimal frequency hopping pattern is constructed for the multi-antenna system in the massive MIMO system.

[0044] Uplink refers to the direction in which data is transmitted from end-user devices (such as mobile phones, computers, etc.) to the network or base station, and downlink refers to the direction in which data is transmitted from the network or base station to the end-user device, such as Figure 1 shown.

[0045] Assume that the sequence C is a (q-2)-order Golomb Costas sequence. Take the uplink as an example, that is, the user sends information to the base station through the wireless channel, such as data, control signaling, etc. Assume that in a large-scale MIMO cellular system, a cell group in the large-scale MIMO system has N1 cells, the number of base station receiving antennas in each cell is N2, the number of optimal frequency hopping pattern families used by each receiving antenna is N3, the number of users in each cell is M1, and the number of transmitting antennas at each user end is M2. Then the number of optimal frequency hopping pattern families to be designed is at least s, where represents the set of positive integers, which can be expressed by the following formula:

[0046] s ≥ N1×N2×N3 (3)

[0047] At the same time, the optimal number of frequency hopping patterns in each group should be s1, and we can get:

[0048]

[0049] Let w represent the minimum Doppler distance between the best frequency hopping patterns, And must meet Where q-1 represents the maximum number of frequency hopping patterns in each family. Add an empty column and an empty row to the rightmost and topmost sides of sequence C to obtain sequence C1, and then cyclically shift sequence C1 upward in the vertical direction by w, 2w, …, (s-1)w rows to obtain sequences C2, C3, …, C s The sequence C1, C2, ..., C s That is, the head of each optimal frequency hopping pattern family. Their placement functions are:

[0050]

[0051] Let w1 represent the minimum delay distance between frequency hopping patterns in a group, and circularly shift the head frequency hopping patterns of each group to the left horizontally by w1, 2w1, …, (s1-1)w1 columns, respectively, to obtain (s1-1) optimal frequency hopping patterns with one gap column and one gap row. The first frequency hopping pattern in each group is the frequency hopping pattern of the head of the group. w1 satisfies

[0052] Therefore, the lth cell in the i-th cell in massive MIMO i The placement function of the optimal frequency hopping pattern is:

[0053]

[0054] where i=1,2,…,s,l i =1,2,…,s1.

[0055] Assume that the frequency hopping operator of the lth optimal frequency hopping pattern of massive MIMO is s l (k), when s l When (k) is the placement function of the M=q-1 order optimal frequency hopping pattern, the complex envelope of the signal frequency hopping in the kth time slot corresponding to the lth optimal frequency hopping pattern of massive MIMO is

[0056]

[0057] in,

[0058]

[0059] If the kth time slot of the best frequency hopping pattern is a gap column, then a l,k =0, otherwise a l,k =1.

[0060] f l,k =s l (k)F b , when θ l,kWhen (t) = 0, it indicates that the sub-pulse is the optimal frequency hopping pattern signal with constant carrier frequency. Let the Doppler shift be normalized by the frequency hopping frequency interval Δf. If the maximum Doppler shift in the system is less than the Doppler distance between the two optimal frequency hopping patterns, then no matter how large the multipath delay is, the two frequency hopping patterns have ideal cross-correlation characteristics. Similarly, let the multipath delay be normalized by the frequency hopping time slot T b =1 / Δf normalized, if the maximum multipath delay in the system is less than the delay distance between the two best frequency hopping patterns, then no matter how large the Doppler frequency shift is, the two frequency hopping patterns have ideal cross-correlation characteristics. The specific method for designing the best frequency hopping pattern that meets the conditions is as follows:

[0061] (1) Construct a Golomb Costas sequence. First, determine the order of the sequence. Take the maximum Doppler frequency shift d in the system as max (d max is a non-negative integer) to determine the minimum Doppler distance w between the best frequency hopping patterns, the condition is based on formula (11):

[0062] w≥d max +1 (11)

[0063] The number s of the optimal frequency hopping pattern clusters required for a cell group is determined based on N1, N2, and N3. Then, the maximum multipath delay τ in the cell is determined based on max (τ max is a non-negative integer) to determine the minimum delay distance w1 of the frequency hopping diagram in the cluster, and the condition is based on formula (12):

[0064] w1 ≥ τ max +1 (12)

[0065] The number s1 of the frequency hopping patterns in each cluster is determined based on M1, M2, N2, and N3. After obtaining the above four parameters w, w1, s, and s1, the value of q can be determined as:

[0066] q-1≥max(w·s,w1·s1) (13)

[0067] Then we get the order N of the Golomb Costas sequence (N = q-2). In GF(q), q = p m , p is a prime number, m is a positive integer, when m=1, a Golomb Costas sequence is constructed based on the prime field.

[0068] (2) Performing cyclic shift on the generated Golomb Costas sequence in the vertical direction to obtain the first frequency hopping pattern of each optimal frequency hopping pattern cluster.

[0069] (3) Performing cyclic shift in the horizontal direction on the first frequency hopping pattern of each cluster respectively obtains other optimal frequency hopping patterns in the cluster.

[0070] Before the actual design begins, the reuse distance D of the frequency hopping diagram can be obtained according to formula (14):

[0071]

[0072] Where N1 = i 2 +j 2 +ij(i=0,1,…;j=1,2…), represents the number of cells in a cell group in a large-scale MIMO system, and R is the cell radius. Then plan the size of each cell and obtain the maximum multipath propagation delay in the cell, estimate the maximum Doppler frequency shift and the maximum number of users in the system, and allocate and reserve the best frequency hopping pattern to cope with the problem of a sudden increase in the number of users under special circumstances, so that the number of designed frequency hopping patterns is greater than the number of users. After a large number of generated best frequency hopping patterns are allocated to each antenna, there will be surplus. The remaining best frequency hopping patterns in these used best frequency hopping pattern families and the unused best frequency hopping pattern families can be reserved, but the best frequency hopping patterns of the same family can only be used for users in the cell that has used the family of frequency hopping patterns.

[0073] A method to mitigate interference between multiple antennas in massive MIMO systems using graph division multiple access technology.

[0074] Different from the maximum 8 antenna channels of traditional 4G MIMO, massive MIMO implements 64 / 128 / 256 or even more antenna channels in 5G. In fact, massive MIMO is the beamforming of a large number of antennas, that is, a technology that adaptively adjusts the radiation pattern of the antenna array according to specific scenarios. Through advanced antenna technology on mobile devices and network base stations, wireless signals can be focused in a specific direction instead of spreading over a wide area. Now, an example is used to briefly explain the process of constructing the optimal frequency hopping pattern based on the Golomb Costas sequence generated in the prime field for use in massive MIMO systems. Assume that there are N1=3 cells in a cell cluster in a massive MIMO system, the number of base station receiving antennas in each cell is N2=2, the number of optimal frequency hopping pattern families used by each receiving antenna is N3=1, the number of users in each cell is M1=5, and the number of transmitting antennas at each user end is M2=2. Then, according to equation (3), the number of optimal frequency hopping pattern families to be designed is at least s=6. According to equation (4), the number of optimal frequency hopping patterns in each family should be at least s1=5. Assume that the maximum Doppler shift in the system is d max =1, the maximum delay within the cell is τ max=1. From formula (11), we can get the minimum Doppler distance between the best frequency hopping pattern clusters as w=2. When C1 is used to form the head of each best frequency hopping pattern cluster, C1 is cyclically shifted every 2 rows in the vertical direction. From formula (12), we can get the minimum delay distance between the frequency hopping patterns in the same cluster as w1=2. Then the frequency hopping pattern of the head of each cluster is cyclically shifted to the left every 2 columns in the horizontal direction to obtain the best frequency hopping pattern of this cluster. From formula (13), we can get q-1≥max(6·2,5·2), that is, q≥13. Here, q=p=13 is taken to further construct the Golomb Costas sequence. According to φ(12)=4, the prime field GF(13) has 4 primitive elements, namely 2, 6, 7, and 11. Let α=2, β=6 to generate an 11-order Golomb Costas sequence C, and the placement function is y(k)=log6(1-2 k ), 1≤k≤11, the resulting sequence is {6,2,1,11,3,5,8,9,4,10,7}, such as Figure 2 shown.

[0075] Now construct the frequency hopping pattern of the head of the best frequency hopping pattern family, add an empty column and an empty row to the rightmost and topmost of sequence C to get sequence C1. Circularly shift sequence C1 vertically by 2 rows, 4 rows, 6 rows, 8 rows, and 10 rows in sequence, and get a total of 5 sequences C2, C3, C4, C5, and C6 with one gap row and one gap column. The placement functions of sequences C1, C2, C3, C4, C5, and C6 are: like Figure 3 (a), (b), (c), (d), (e), (f).

[0076] Sequences C1 and C2 are used as the head frequency hopping patterns of frequency hopping pattern clusters 1 and 2, respectively, and are cyclically shifted to the left in the horizontal direction to obtain the best frequency hopping pattern in the cluster. Figure 4 Several patterns of frequency hopping pattern families 1 and 2 are given, and the placement function corresponding to each pattern is: a (k) = log6(1-2 k ), y b (k) = log6(1-2 2 2 k ), y c (k) = log6(1-2 4 2 k ), y d (k) = 2 + log6 (1-2 k ), y e (k) = 2 + log6 (1-2 2 2 k ), y f(k) = 2 + log6(1 - 2 4 ·2 k ).

[0077] According to Figure 5 as shown, from the above steps, it can be set that the maximum number of users in the system cell M1 = 50, the number of user transmitting antennas M2 = 2, the number of cells in the cluster N1 = 3, the number of cell base station receiving antennas N2 = 20, the number of families used for each base station antenna N3 = 1. After calculation, the number of families required for the system is s = 60, s1 = 5. Let the maximum normalized Doppler shift and multipath delay in the system be d max = 1 and τ max = 1 respectively. The minimum Doppler distance between families and the minimum delay distance within families are w = 2 and w1 = 2 respectively. Based on the prime field, a Golomb Costas array is constructed, BPSK modulation is adopted, and the bit error rate of a single cell is simulated. Figure 6 shows the bit error rate in the AWGN channel and the multipath channel in the single-user scenario. Among them, each curve represents communication in the multipath channel with a constant carrier frequency method (without using any anti-interference technology), communication in the multipath channel using orthogonal graph division multiple access, and communication in the AWGN channel using orthogonal graph division multiple access. From Figure 6 it can be seen that when in the AWGN channel, due to being unaffected by multipath interference, the bit error rate decreases rapidly as the signal-to-noise ratio increases. When in the multipath channel, there are multipath delay and Doppler shift. Since there is only one user, at this time, there is only signal self-interference caused by multipath delay and Doppler shift, and no interference between users. In the multipath channel, if communication is carried out using the constant carrier frequency method, then as the signal-to-noise ratio increases, the bit error rate decays relatively slowly. When using the best frequency hopping graph based on the Golomb Costas array for frequency hopping communication, compared with the constant carrier frequency, the bit error rate is significantly improved, and the bit error rate curve is closer to the AWGN channel. Thus, it can be seen that the ideal autocorrelation of the best frequency hopping graph can effectively suppress the signal self-interference caused by multipath delay and Doppler shift.

[0078] Figure 7 shows the bit error rate in the AWGN channel and the multipath channel when the number of cell users is 50. Among them, each curve represents the case of constant carrier frequency in the multipath channel, the case when the maximum normalized Doppler shift and multipath delay are within the minimum Doppler distance w between families and the minimum delay distance w1 within families (d max < w, τ max < w1), the case when d max ≥ w, τ max ≥ w1, and the case of orthogonal graph division multiple access communication in the AWGN channel. In the multi-user scenario, in addition to the signal self-interference caused by multipath delay and Doppler shift, there will also be interference between users. From Figure 7 It can be seen that in the constant carrier frequency mode, the communication quality becomes worse. As the signal-to-noise ratio increases, the bit error rate decreases very slowly. When the users in the cell adopt the optimal frequency hopping pattern designed for this system, the communication quality will be significantly improved. Let d max = 1, τ max = 1, the optimal frequency hopping pattern designed for this system satisfies w = 2, w1 = 2, that is, it satisfies d max < w, τ max < w1. At this time, it can ensure that there is an ideal cross-correlation between the optimal frequency hopping patterns. From Figure 7 in the curve "Multipath Channel (OFHP-in)", it can be seen that the ideal cross-correlation of the optimal frequency hopping pattern greatly suppresses the interference between users, and greatly improves the communication reliability. When d max ≥ w, τ max ≥ w1, let d max = 2, τ max = 2. At this time, it does not guarantee that there is an ideal cross-correlation between the frequency hopping patterns. From Figure 7 in the curve "Multipath Channel (OFHP-out)", it can be seen that compared with "Multipath Channel (OFHP-in)", the bit error rate increases, but compared with the constant carrier frequency mode, it still has a certain anti-interference ability. Ensuring the optimal frequency hopping pattern can make the signal have an "immune" ability in a multi-user and multi-antenna scenario, can overcome the interference between users. At the receiving end, there is no need to perform equalization and channel estimation for equalization anymore. The channel estimation for equalization is omitted, and the pilot occupation becomes lower, thereby reducing pilot contamination.

[0079] Use the graph division multiple access technology to reduce the pilot contamination in the large-scale MIMO system and reduce the receiver complexity, so that the large-scale MIMO technology can enter practical applications.

[0080] In wireless communication, when signals are transmitted through channels, interference caused by phenomena such as multipath propagation and Doppler shift often occurs, which seriously affects the accuracy of the signals at the receiving end. In order to ensure the reliability of communication, the receiving end usually adopts equalization and other methods to combat multipath interference and improve the communication quality. In order to fully reflect the gain brought by large-scale MIMO, it is generally necessary to perform channel estimation at the receiving end, and using pilot sequences to achieve channel estimation is one of the current mainstream channel estimation methods. For the uplink of large-scale MIMO, the terminal usually sends a pilot signal to the base station, and the base station performs channel estimation after receiving the pilot signal. Ideally, the pilot sequences allocated to each user are mutually orthogonal. However, since the pilot length is limited by the coherence interval, if the coherence interval is not long enough or the number of terminals is too large, some pilot sequences will be reused and cause fading. Therefore, when the base station receives the pilot signal of the desired channel, it will also receive the pilot signal of the non-desired channel that reuses the pilot sequence, thereby affecting the accuracy of the channel estimation. This phenomenon is pilot contamination. Therefore, when the number of antennas N2 at the base station end increases, the gain brought by the large-scale MIMO system will also increase further. However, due to the gain loss caused by pilot contamination, once the system performance reaches a certain value, increasing the number of antennas will not further improve it. Pilot contamination has become one of the main reasons restricting the performance of large-scale MIMO systems. Therefore, in order to fully enjoy the gain brought by large-scale MIMO, it is necessary to suppress pilot contamination as much as possible while overcoming the interference caused by multipath and Doppler effect.

[0081] As can be seen from the above, orthogonal graph division multiple access technology can effectively reduce the interference between multiple antennas in large-scale MIMO systems, so that the system's demand for pilot signals is reduced. In large-scale MIMO systems, pilot pollution is caused by the fact that limited pilot resources cannot meet actual needs, resulting in some pilots being reused. For large-scale MIMO systems that use orthogonal graph division multiple access, channel estimation for equalization is omitted, reducing the occupancy of pilot resources, which can inhibit pilot pollution to a certain extent. Since there is no need for equalization and channel estimation for equalization, the complexity of the receiver is reduced, and pilot pollution is also reduced to a certain extent, so that the excellent performance of large-scale MIMO systems can be fully utilized.

[0082] The orthogonal graph division multiple access system based on the best frequency hopping pattern is similar to the traditional frequency hopping communication system, that is, both the transmitting and receiving ends need to ensure the same frequency hopping pattern, while the former uses a two-dimensional orthogonal best frequency hopping pattern, while the latter uses a one-dimensional orthogonal frequency hopping code. Figure 8As shown, Figure (a) is the transmitting end. On the one hand, the input signal is loaded by the digital modulator through the subcarrier to become a modulated signal, and then mixed with the frequency generated by the optimal frequency hopping pattern generator to become a frequency hopping signal that changes according to the optimal frequency hopping pattern, and then the signal is transmitted. Figure (b) is the receiving end. The receiving end uses the same optimal frequency hopping pattern as the transmitting end, and demodulates the signal received from the antenna according to the corresponding frequency hopping pattern to obtain a modulated signal, and finally restores the original input signal through the demodulator.

[0083] The above description is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiment. Any equivalent modifications or changes made by ordinary technicians in this field based on the contents disclosed by the present invention should be included in the protection scope recorded in the claims.

Claims

1. A two-dimensional orthogonal graph division multiple access communication method for a large-scale MIMO system, characterized by: The method comprises the following steps: Step 1: For a massive MIMO system, obtain its basic data, including the number of cells in a cell group N1, the number of base station receiving antennas in each cell N2, the number of optimal frequency hopping pattern families used by each base station receiving antenna N3, the number of users in each cell M1 and the number of transmitting antennas at each user end M2, and calculate the reuse distance D of the frequency hopping pattern; Step 2, determining the required optimal frequency hopping pattern family number s according to the number of cells N1 in a cell group, the number of base station receiving antennas N2 in each cell, and the number of optimal frequency hopping pattern families N3 used by each receiving antenna of the base station, and determining the number s1 of optimal frequency hopping patterns in each optimal frequency hopping pattern family according to the number of users M1 in each cell, the number of transmitting antennas M2 of each user terminal, the number of base station receiving antennas N2 in each cell, and the number of optimal frequency hopping pattern families N3 used by each receiving antenna of the base station; Step 3: Based on the maximum Doppler shift d in the massive MIMO system max and the maximum delay τ within the cell max , obtain the minimum Doppler distance w between the best frequency hopping pattern families and the minimum delay distance w1 between the best frequency hopping patterns in each best frequency hopping pattern family; Step 4: According to the number of optimal frequency hopping pattern families s, the number of optimal frequency hopping patterns in each optimal frequency hopping pattern family s1, the minimum Doppler distance w between the optimal frequency hopping pattern families, and the minimum delay distance w1 between the optimal frequency hopping patterns in each optimal frequency hopping pattern family obtained in the previous step, the range of the maximum number of frequency hopping patterns q-1 in each family is obtained, where q is a prime number or a power of a prime number, that is, q=p m , where p is a prime number and m is a positive integer; q-1 is the maximum number of time slots for the optimal frequency hopping pattern. According to the frequency domain width B and time domain width T of the signal in the massive MIMO system, the frequency slot F is obtained. b With time slot T b , construct the q-2 order Golomb Costas sequence C according to the value of q-1, and determine its placement function; Step 5: Assume that the frequency hopping operator of the lth optimal frequency hopping pattern of massive MIMO is s l (k), when s l When (k) is the placement function of the M=q-1 order optimal frequency hopping pattern, the complex envelope of the signal frequency hopping in the kth time slot corresponding to the lth optimal frequency hopping pattern of massive MIMO is in, If the kth time slot of the best frequency hopping pattern is a gap column, then a l,k =0, otherwise a l,k =1; f l,k =s l (k)F b , when θ l,k When (t) = 0, it indicates that the sub-pulse is the optimal frequency hopping pattern signal with constant carrier frequency; Step 6: construct the frequency hopping pattern of the head of the best frequency hopping pattern family, add an empty column and an empty row to the rightmost and topmost of sequence C to obtain sequence C1; cyclically shift sequence C1 vertically by w, 2w, …, (s-1)w rows to obtain sequences C2, C3, …, C s , sequence C1, C2, ..., C s That is, the head of each optimal frequency hopping pattern family; the head frequency hopping patterns of each family are circularly shifted to the left horizontally by w1, 2w1, …, (s1-1)w1 columns, and s1-1 optimal frequency hopping patterns with one gap column and one gap row are obtained. The first frequency hopping pattern in each family is the head frequency hopping pattern of the family, and w1 satisfies Step 7, applying the constructed GolombCostas sequence to the massive MIMO system, so that the system's demand for pilot signals is reduced, thereby suppressing pilot contamination, while suppressing interference caused by multipath propagation and Doppler frequency shift and interference between users, reducing receiver complexity, and realizing multiple access communication of the massive MIMO system.

2. The two-dimensional orthogonal graph division multiple access communication method for a massive MIMO system according to claim 1, characterized in that: In step 1, the reuse distance D of the frequency hopping diagram is obtained according to formula (14): Where N1 = i 2 +j 2 +ij, i = 0, 1, ...; j = 1, 2 ..., represents the number of cells in a cluster in a massive MIMO system, (i, j) is the cell coordinate of sequence C, and R is the cell radius.

3. The two-dimensional orthogonal graph division multiple access communication method for a massive MIMO system according to claim 1, characterized in that: In step 2, it is assumed that the number of optimal frequency hopping patterns to be designed is at least s, where represents the set of positive integers, expressed by the following formula: s ≥ N1×N2×N3(3) At the same time, the optimal number of frequency hopping patterns in each group should be s1, so:

4. The two-dimensional orthogonal graph division multiple access communication method for a massive MIMO system according to claim 1, characterized in that: In step 3, the maximum Doppler frequency shift d in the system is max To determine the minimum Doppler distance w,d between the best frequency hopping patterns max is a non-negative integer, and the conditions are as follows: w≥d max +1 (11) According to the maximum multipath delay τ in the cell max To determine the minimum delay distance w1, τ in the frequency hopping diagram of the family max is a non-negative integer, and the condition is as follows: w1≥τ max +1 (12)。 5. The two-dimensional orthogonal graph division multiple access communication method for a massive MIMO system according to claim 1, characterized in that: In step 4, after obtaining the four parameters w, w1, s, and s1, the value of q-1 is determined as: q-1≥max(w·s,w1·s1) (13).

6. The two-dimensional orthogonal graph division multiple access communication method for a massive MIMO system according to claim 1, characterized in that: In step 5, for the constructed GolombCostas sequence, the restricted domain is GF(q), where q = p m , m is a positive integer, p is a prime number, let α and β be two primitive elements of GF(q), and sequence C be a permutation matrix of order N = q-2. Then the necessary and sufficient condition that sequence C is a GolombCostas sequence is that the placement function of sequence C is: y(k)=log β (1-a k ),1≤k≤q-2 (1) Assume that the cell coordinates of sequence C are (i, j). When equation (2) is satisfied, the cell is changed to black: α i +β j ＝1 (2) When m=1, the GolombCostas sequence is constructed based on the prime field, and equation (2) represents the modulo p congruence; the maximum autocorrelation sidelobe of the autocorrelation function of the GolombCostas sequence is 1, which has an ideal autocorrelation property.

7. The two-dimensional orthogonal graph division multiple access communication method for a massive MIMO system according to claim 1, characterized in that: In step 6, the sequence C1, C2, ..., C s The placement functions are:

8. The two-dimensional orthogonal graph division multiple access communication method for a massive MIMO system according to claim 1, characterized in that: In step 6, the lth cell in the i-th cell in the obtained massive MIMO is constructed i The placement function of the optimal frequency hopping pattern is: where i=1,2,…,s,l i =1,2,…,s1; Let the Doppler shift be normalized by the frequency hopping interval Δf. If the maximum Doppler shift in the system is less than the Doppler distance between the two best frequency hopping patterns, then the two frequency hopping patterns have ideal cross-correlation characteristics regardless of the multipath delay. Similarly, let the multipath delay be normalized by the frequency hopping time slot T b =1 / Δf normalized, if the maximum multipath delay in the system is less than the delay distance between the two best frequency hopping patterns, then no matter how large the Doppler frequency shift is, the two frequency hopping patterns have ideal cross-correlation characteristics.

Images