HARQ-assisted authorization-free access method in cellular-free large-scale MIMO system

Through the HARQ retransmission strategy and the Poisson cluster process model, combined with macro diversity and signal sparseness, the equipment transmission reliability and interference problems in authorization-free access are solved, and the efficient reliability of the large-scale MIMO system without cellular is achieved.

CN119093978APending Publication Date: 2024-12-06SOUTHEAST UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410858272.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-06-28
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

In a large-scale MIMO system without cellular, the equipment lacks resource scheduling connection under the authorization-free random access mechanism, which affects the transmission reliability. The traditional HARQ solution has significant interference between users under the cellular network, affecting the probability of data transmission success.

Method used

The HARQ retransmission strategy assists authorization-free access, combined with the Poisson cluster process to simulate the geographical location distribution of users and APs, deduce the maximum retransmission times and signal-to-interference noise ratio closed expression, and use macro diversity and signal-space sparsity to suppress interference , analyze the interrupt probability through the MRC receiver.

Benefits of technology

It effectively improves the reliability and transmission performance of the system, reduces interference, and achieves better system performance with fewer antennas and APs, verifying the superiority of authorization-free HARQ under large-scale MIMO without cellular.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119093978A_ABST
    Figure CN119093978A_ABST
Patent Text Reader

Abstract

The invention discloses an HARQ (Hybrid Automatic Repeat reQuest) retransmission strategy assisted authorization-free access method in a cellular-free large-scale MIMO (Multiple Input Multiple Output) system. According to the method, clustering is carried out based on Poisson cluster process space distribution characteristics of users and APs, and then the macro diversity gain and signal space sparsity of the non-cellular large-scale MIMO system are fully utilized to reduce interference to communication users. Then, an HARQ retransmission strategy is introduced, round-trip time delay is analyzed, and the maximum number of retransmissible times under the constraint of time delay is obtained; in addition, a closed expression of the uplink and downlink transmission signal to interference plus noise ratio and an approximate expression of the outage probability of the maximum ratio merging receiver are deduced, and the closed expression and the approximate expression are used for verifying the effectiveness of the scheme. Compared with an HARQ scheme under a traditional centralized cellular architecture, the method has the advantages that the macro diversity gain and signal space sparsity of the cellular-free large-scale MIMO system are fully utilized, the influence of interference on communication users can be effectively reduced, the system reliability is improved under the time delay constraint, and therefore the method has certain actual value.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of mobile communications and relates to a license-free access method for a non-cellular large-scale MIMO system. Background Art

[0002] The traditional authorization-based access method requires the terminal to establish a connection with the base station before sending data, and then transmit data after receiving feedback from the base station. This four-step handshake mechanism brings a lot of signaling overhead and additional delay to user communications, and is no longer suitable for the future 6G-enabled Internet of Things with strong contingency and sparseness. In order to meet the strict delay requirements in mURLLC, the unlicensed random access mechanism allows users to directly transmit randomly selected pilots and data to reduce access delay and signal overhead. In 3GPP Release 16, the unlicensed random access process with a two-step handshake mechanism was introduced to support MTC devices. Unlike the traditional four-step authorization access, the device in the unlicensed random access does not need to wait for the base station scheduling authorization before transmitting data, thereby reducing access delay and control signaling overhead. In the two-step unlicensed random access mechanism, once the device is active in the time slot, it directly sends the pilot and data packet to the base station, and then waits for the base station decoding feedback. Since the device lacks a resource scheduling connection with the base station under the unlicensed random access mechanism, active devices will inevitably conflict when randomly selecting transmission resources, thereby affecting their transmission reliability.

[0003] In order to alleviate the above problems, the Hybrid Automatic Repeat Request (HARQ) machine under unlicensed random access can improve the communication reliability of the device. In the unlicensed random access assisted by HARQ retransmission, the active device can autonomously transmit multiple pilots and data, and meet the high-reliability communication requirements in URLLC by increasing time diversity. However, the research on traditional unlicensed HARQ schemes focuses on centralized cellular networks. As the number of users increases, the disadvantages of cellular networks will become apparent. For example, users at the edge of the cell will suffer from huge inter-user interference, which will affect the probability of successful transmission of user data.

[0004] In addition, cell-free massive MIMO, which combines the advantages of massive MIMO and distributed MIMO, is considered to be a key technology for the upcoming 6G network. In cell-free massive MIMO, each device user is served by multiple distributed access points (APs) in the same time-frequency resources, which transforms the system from network-centric to user-centric. At the same time, cell-free massive MIMO has the advantages of enhanced macro-diversity gain and signal power sparsity, which can effectively suppress interference between users and improve transmission reliability. Therefore, it is of great theoretical and practical significance to combine unlicensed NOMA with cell-free massive MIMO to design an unlicensed mechanism that allows massive users to access, thereby improving the reliability of the system. Summary of the invention

[0005] Technical problem: Based on the advantages of unlicensed NOMA and non-cellular massive MIMO, the present invention proposes an unlicensed access mechanism assisted by a HARQ retransmission strategy. This mechanism takes into account the actual user and AP aggregation scenario, uses the Poisson Cluster Process (PCP) to simulate the geographical distribution of users and APs, and performs clustering based on the PCP spatial distribution characteristics of users and APs. At the same time, for the introduced HARQ retransmission strategy, the round-trip delay is used to obtain the maximum number of retransmissions under the delay constraint. In addition, the present invention derives a closed expression for the uplink transmission signal to interference plus noise power ratio (SINR) and an approximate expression for the interruption probability under the maximum ratio combining receiver, which are used to measure the reliability performance of the system. The simulation results verify the accuracy of the closed expressions of SINR and the interruption probability, and prove the superiority of unlicensed HARQ in suppressing inter-user interference in the non-cellular massive MIMO scenario compared to the centralized massive MIMO scenario.

[0006] Technical solution: A HARQ-assisted unlicensed access method of the present invention includes the following contents:

[0007] Step 1: Construct a non-cellular massive MIMO system, which has K users and M access points (AP) connected to the central processing unit (CPU). Each user is equipped with a single antenna, and each AP is equipped with N antennas. It is assumed that multiple APs in the same AP cluster collaborate to complete the reception and transmission of user signals. The AP is connected to the edge distributed unit (EDU) in the cluster through a backhaul link, and the signal processing is completed on the EDU. The EDU is connected to the CPU through a backhaul link to realize information integration.

[0008] Step 1: Establish a spatial distribution model, consider EDU as the cluster center, and use density λ as p The uniform Poisson point process is uniformly distributed. Users and APs gather around EDU to form two independent Poisson cluster processes (PCP). The present invention simulates the actual distribution of users and APs through Thomas clustering. Thomas clustering consists of child points and parent points. In the present invention, the child points are users and APs, the parent point is EDU, and the probability density function of the child point to the parent point can be expressed as:

[0009]

[0010] in, and Represent the locations of users and APs respectively; and denote the variance of the user and AP respectively. At the same time, assuming that the standard deviation of the user and AP is u and η a Much smaller than half of the average distance between adjacent cluster centers. In each cluster, the density of users and APs is λ u and λ a .

[0011] Step 2: Establish the system model. The channel response vector between user k and APm is given by:

[0012]

[0013] in represents the large-scale fading coefficient between user k and APm, d represents the distance between user k and APm, and α is the path loss exponent; represents the small-scale fading coefficient, I N is the N-order identity matrix. Therefore, g k,m ~CN(0,λ k,m I N ).

[0014] Step 2: Cluster users and APs based on spatial distribution characteristics to obtain signal spatial sparsity and macro-diversity gain. Using the PCP spatial distribution characteristics, UEs and APs are clustered according to the EDUs selected by them. Among them, all UEs that select the same EDU are considered to be in the same cluster. Since the concept of cell division in cellular networks is eliminated, after different user signals experience different degrees of large-scale fading, the AP adjacent to the user usually captures more significant energy than other APs. Therefore, only neighboring APs within the communication range of the user device (i.e., APs in the same cluster) have non-negligible signal gain, which leads to the characteristic of signal sparsity in CF-mMIMO, which has the advantage of significantly reducing interference between users. In addition, in the CF-mMIMO scenario, since APs are geographically distributed, each device can select multiple adjacent APs as its service AP set, which enables users to have good macro-diversity gain.

[0015] Step 3: Non-orthogonal pilot allocation based on clustering information. Considering the intra-cluster and inter-cluster interference caused by Grant Free Random Access (GFRA) terminal sharing resources, the pilot signal received by APm in the uplink channel estimation stage can be expressed as:

[0016]

[0017] Among them, p k represents the transmission power of user k; Φ k represents the set of users in cluster k; represents the orthogonal pilot of length τ used by user k in the cluster; represents the Gaussian white noise AWGN vector, where each element obeys CN(0,σ 2 ). The received pilot matrix Y p,m It will be used to estimate all channels from APm to the users it serves.

[0018] Considering the huge number of users in the mURLLC scenario, it is extremely unrealistic to equip each user with an orthogonal pilot sequence, which will cause huge pilot overhead, although the use of orthogonal pilots will help alleviate the problem of pilot pollution. Therefore, based on the clustering result in step 2, orthogonal pilots are used in the same cluster, and non-orthogonal pilots can be used for users in different clusters. That is, in equation (3), the pilot sequence of UEi in the same cluster as UEk is The pilot signal selected by UEj in a different cluster from UEk is Can be equal to The reason for this allocation is based on the signal sparseness of the CF-mMIMO network. The interference of users with geographically distant locations on the target user's communication can be ignored. Out-of-cluster users with sparse characteristics for the target user can choose the same pilot resources as the target user without interfering with the target user's communication.

[0019] Then, for subsequent channel estimation, APm multiplies the received signal by the pilot sequence To perform a spread spectrum operation:

[0020]

[0021] in, According to the analysis provided above, due to the sparseness of the signal, the second term of the above equation can be approximated to zero. Therefore,

[0022] Step 4: Using the minimum mean square error channel estimation method, the channel estimation of user k at APm can be expressed as:

[0023]

[0024] in, Also equivalent to represents the estimated value of the large-scale fading factor, is the estimated small-scale fading factor, σ 2 is the noise power; represents the channel estimation error, η k,m =λ k,m -β k,m .

[0025] Step 5: Analyze the maximum number of retransmissions under the delay constraint. When interference prevents the user's SINR at the EDU from reaching the decoding threshold, the data sent by the user will not be successfully decoded, resulting in access failure. HARQ technology can ensure the reliability of authorization-free policy transmission, but because URLLC has strict delay requirements, the time for retransmission is very limited. Therefore, by analyzing the system delay, the maximum number of retransmissions can be obtained. The present invention adopts an active HARQ strategy. In this strategy, retransmission is only allowed when the device receives a negative acknowledgment (NegativeAcknowledge, NACK).

[0026] At the beginning of the transmission, frame alignment is required. Then, the user starts to transmit data. Next, the AP receives and processes the data. If the data packet is successfully decoded, the AP sends an ACK feedback, indicating that the transmission is over. If the data packet decoding fails, the AP will send a NACK feedback, which indicates that the user needs to retransmit the data within the maximum number of retransmissions allowed. After receiving the feedback, the user decides whether to perform a retransmission. Considering the short packet communication in the URLLC scenario, the time unit is the Transmission Time Interval (TTI).

[0027] Therefore, the HARQ round trip time (RTT) is The delay after the qth HARQ round trip is:

[0028]

[0029] Among them, T fa Indicates frame alignment delay; T tx Indicates the data packet transmission time; T dp Indicates the processing time of AP; T fb Indicates the time when AP sends feedback to the user; T up represents the user's processing time. For ease of analysis, assume that T fa =T tx =T dp =T fb =T up =1TTI.

[0030] In the URLLC scenario, the data packet needs to be delayed for a certain period of time T latency Therefore, the maximum number of retransmissions is:

[0031]

[0032] Among them, T latency is the delay constraint threshold. Therefore, the number of retransmissions for each UE will not exceed N max .

[0033] Step 6: System reliability performance analysis. During the retransmission process, the signal to interference plus noise power ratio (SINR) and the interruption probability will change dynamically. Once a terminal successfully transmits, the interference will be reduced, and once a user retransmits, the interference will increase. Therefore, the present invention derives the closed expression of the user uplink SINR under the MRC receiver of the proposed unlicensed access strategy, and further derives the approximate expression of the interruption probability of user access with the help of the log-normal distribution theory, so as to analyze the reliability of the system.

[0034] Step 1: Closed-form expression of uplink SINR based on MRC receiver.

[0035] Define the set of APs serving user k as S k Assuming that the EDU has the AP information serving each user, consider the MRC receiver Then the signal received by user k on EDU is:

[0036]

[0037] Therefore, the SINR expression of user k is:

[0038]

[0039] in,

[0040]

[0041] Based on equation (9), the uplink SINR closed-form expression is derived:

[0042]

[0043] in,

[0044]

[0045] Step 2: Approximate expression for outage probability based on the lognormal moment matching method.

[0046] The end-to-end SINR is approximated by using the moment matching technique for the Gamma or lognormal random variable, and the user outage probability is further derived. k Denominator Y k is approximately a lognormal random variable, that is, Based on the above approximation, for the delay constraint threshold T latency , the interruption probability of user k can be approximated as:

[0047]

[0048] where erfc(·) is the complementary error function;

[0049]

[0050] E[Y k ]=E[A k ]+E[B k ]+E[C k ]+E[D k ], (twenty three)

[0051]

[0052]

[0053] Beneficial effects: The HARQ-assisted unlicensed access method proposed in the present invention in a non-cellular massive MIMO system takes into account the user PCP spatial distribution model, utilizes macro diversity and signal spatial sparsity, can effectively suppress interference with fewer antennas and APs, and achieve better system performance. And by deriving the SINR closed expression under the MRC receiver and the analysis of the approximate expression of the interruption probability, the interruption performance of the scheme proposed in the present invention can be effectively verified, which is used to measure the system reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is the spatial distribution model of the present invention.

[0055] Figure 2 This is a HARQ delay analysis diagram.

[0056] Figure 3 It is the fitting curve of the theoretical value and simulation value of SINR changing with the number of antennas under the MRC receiver.

[0057] Figure 4 It is the fitting curve of the theoretical value and simulation value of the outage probability as the number of antennas changes.

[0058] Figure 5 It is the curve of the interruption probability changing with the number of retransmissions under the condition of SINR threshold change.

[0059] Figure 6 It is the interruption probability curve of the density ratio of users and APs when the number of service APs changes. Figure 7 The performance of a cell-free massive MIMO system versus a centralized system is shown as the number of antennas and serving APs vary. DETAILED DESCRIPTION

[0060] The present invention is described in detail below in conjunction with examples:

[0061] Assume that a non-cellular massive MIMO system is located in a square area with a side length of 1 km and the average density of EDUs (clusters) is λ p =10 / km 2 , unless otherwise specified, the average density of AP is λ ap =5 / cluster, the average density of active users is λ ue=5 / cluster. The actual number of EDUs, APs, and users all follow Poisson distribution. To ensure that each cluster is far enough away, the distance between EDUs must be greater than 200 meters. The path loss model between user k and APm is where d k,m is the distance between user k and APm, and α = 3.7 is the path loss exponent. For ease of analysis, each user selects several APs with the largest large-scale fading factors as service APs. The packet length is L = 128, and the allowed block error rate is ε = 10 -6 All results are averages of 10,000 simulations.

[0062] The channel response vector between user k and APm is given by:

[0063]

[0064] in, represents the small-scale fading coefficient, I N is the N-order identity matrix.

[0065] The implementation method of the present invention in the system is specifically as follows:

[0066] Step 1: Cluster users and APs based on spatial distribution characteristics to obtain signal spatial sparsity and macro-diversity gain. Using the PCP spatial distribution characteristics, UEs and APs are clustered according to the EDUs selected by them. Among them, all UEs that select the same EDU are considered to be in the same cluster. Since the concept of cell division in cellular networks is eliminated, after different user signals experience different degrees of large-scale fading, the AP adjacent to the user usually captures more significant energy than other APs. Therefore, only neighboring APs within the communication range of the user device (i.e., APs in the same cluster) have non-negligible signal gain, which leads to the characteristic of signal sparsity in CF-mMIMO, which has the advantage of significantly reducing interference between users. In addition, in the CF-mMIMO scenario, since APs are geographically distributed, each device can select multiple adjacent APs as its service AP set, which enables users to have good macro-diversity gain.

[0067] Step 2: Non-orthogonal pilot allocation based on clustering information. Considering the intra-cluster and inter-cluster interference caused by Grant Free Random Access (GFRA) terminal sharing resources, the pilot signal received by APm in the uplink channel estimation stage can be expressed as:

[0068]

[0069] Among them, p k represents the transmission power of user k; Φk represents the set of users in cluster k; represents the orthogonal pilot of length τ used by user k in the cluster; represents the Gaussian white noise AWGN vector, where each element obeys CN(0,σ 2 ). The received pilot matrix Y p,m It will be used to estimate all channels from APm to the users it serves.

[0070] Considering the huge number of users in the mURLLC scenario, it is extremely unrealistic to equip each user with an orthogonal pilot sequence, which will cause huge pilot overhead, although the use of orthogonal pilots will help alleviate the problem of pilot pollution. Therefore, based on the clustering result in step 2, orthogonal pilots are used in the same cluster, and non-orthogonal pilots can be used for users in different clusters. That is, in equation (3), the pilot sequence of UEi in the same cluster as UEk is The pilot signal selected by UEj in a different cluster from UEk is Can be equal to The reason for this allocation is based on the signal sparseness of the CF-mMIMO network. The interference of users with geographically distant locations on the target user's communication can be ignored. Out-of-cluster users with sparse characteristics for the target user can choose the same pilot resources as the target user without interfering with the target user's communication.

[0071] Then, for subsequent channel estimation, APm multiplies the received signal by the pilot sequence To perform a spread spectrum operation:

[0072]

[0073] in, According to the analysis provided above, due to the sparseness of the signal, the second term of the above equation can be approximated to zero. Therefore,

[0074] Step 3: Using the minimum mean square error channel estimation method, the channel estimation of user k at APm can be expressed as:

[0075]

[0076] in, Also equivalent to represents the estimated value of the large-scale fading factor, is the estimated small-scale fading factor, σ 2 is the noise power; represents the channel estimation error, η k,m =λk,m -β k,m .

[0077] Step 4: Analyze the maximum number of retransmissions under the delay constraint. When interference prevents the user's SINR at the EDU from reaching the decoding threshold, the data sent by the user will not be successfully decoded, resulting in access failure. HARQ technology can ensure the reliability of authorization-free policy transmission, but because URLLC has strict delay requirements, the time for retransmission is very limited. Therefore, by analyzing the system delay, the maximum number of retransmissions can be obtained. The present invention adopts an active HARQ strategy. In this strategy, retransmission is only allowed when the device receives a negative acknowledgment (NegativeAcknowledge, NACK).

[0078] At the beginning of the transmission, frame alignment is required. Then, the user starts to transmit data. Next, the AP receives and processes the data. If the data packet is successfully decoded, the AP sends an ACK feedback, indicating that the transmission is over. If the data packet decoding fails, the AP will send a NACK feedback, which indicates that the user needs to retransmit the data within the maximum number of retransmissions allowed. After receiving the feedback, the user decides whether to perform a retransmission. Considering the short packet communication in the URLLC scenario, the time unit is the Transmission Time Interval (TTI).

[0079] Therefore, the HARQ round trip time (RTT) is The delay after the qth HARQ round trip is:

[0080]

[0081] Among them, T fa Indicates frame alignment delay; T tx Indicates the data packet transmission time; T dp Indicates the processing time of AP; T fb Indicates the time when AP sends feedback to the user; T up represents the user's processing time. For ease of analysis, assume that T fa =T tx =T dp =T fb =T up =1TTI.

[0082] In the URLLC scenario, the data packet needs to be delayed for a certain period of time T latency Therefore, the maximum number of retransmissions is:

[0083]

[0084] Where T is the delay constraint threshold. Therefore, the number of retransmissions per UE will not exceed N max .

[0085] Step 5: System reliability performance analysis. During the retransmission process, the signal to interference plus noise power ratio (SINR) and the interruption probability will change dynamically. Once a terminal successfully transmits, the interference will be reduced, and once a user retransmits, the interference will increase. Therefore, the present invention derives the closed expression of the user uplink SINR under the MRC receiver of the proposed unlicensed access strategy, and further derives the approximate expression of the interruption probability of user access with the help of the log-normal distribution theory, so as to analyze the reliability of the system.

[0086] Step 1: Closed-form expression of uplink SINR based on MRC receiver.

[0087] Define the set of APs serving user k as S k Assuming that the EDU has the AP information serving each user, consider the MRC receiver Then the signal received by user k on EDU is:

[0088]

[0089] Therefore, the SINR expression of user k is:

[0090]

[0091] in,

[0092]

[0093]

[0094] Based on equation (8), the closed-form expression of uplink SINR is derived:

[0095]

[0096] in,

[0097]

[0098] Step 2: Approximate expression for outage probability based on the lognormal moment matching method.

[0099] The end-to-end SINR is approximated by using the moment matching technique for the Gamma or log-normal random variable, and the user outage probability is further derived. k Denominator Yk is approximately a lognormal random variable, that is, Based on the above approximation, for the delay constraint threshold T latency , the interruption probability of user k can be approximated as:

[0100]

[0101] where erfc(·) is the complementary error function;

[0102]

[0103]

[0104] E[Y k ]=E[A k ]+E[B k ]+E[C k ]+E[D k ],(twenty two)

[0105]

[0106]

[0107] The above shows the whole process of a HARQ-assisted unlicensed access method in a non-cellular massive MIMO system.

[0108] The technical solution of the present invention utilizes the macrodiversity and signal space sparsity of the non-cellular massive MIMO system, which can effectively suppress interference with fewer antennas and service APs, and achieve better system performance. At the same time, the SINR closed expression under the MRC receiver and the approximate expression of the outage probability analysis verify the outage performance of the proposed solution, which is used to measure the system reliability. The specific performance analysis will be described in detail in the simulation results.

[0109] Figure 3 and Figure 4 It shows that the theoretical and simulated values ​​of SINR and outage probability have good fitting results when the number of AP antennas and the user-to-AP density ratio change, which verifies the accuracy of the SINR closed expression and the approximate expression of outage probability. As the number of AP antennas increases, the macro diversity gain of the system is enhanced, that is, the useful signal gain increases, and the interference signal is effectively suppressed, so the SINR increases, resulting in a good outage probability. In addition, as the density ratio of user APs increases, users can get closer to the AP, which enhances the useful signal, while the increase in the interference signal is relatively smaller than it, so the SINR also increases, making the outage probability lower.

[0110] Figure 5It shows that the outage probability increases with the increase of the access SINR threshold. When the SINR threshold is 5dB, the outage probability is still high even if the retransmission time increases. This shows that a reasonable SINR threshold must be set, otherwise, even if the allowed transmission delay is large, the reliability requirement cannot be met.

[0111] In order to compare the performance of GF-HARQ in the cell-free massive MIMO system and the centralized massive MIMO system and ensure the fairness of the comparison, the base station in the centralized system is located at the center of each cluster, and the number of antennas in the centralized system is equal to the sum of the AP antenna numbers in the cell-free massive MIMO system.

[0112] Figure 6 It shows that when the number of service APs for each user is M ap =1 and M ap =3, the interruption probability varies with the user to AP density ratio. As the user density increases or the AP density decreases, the performance of both the cell-free massive MIMO system and the centralized system decreases rapidly. When the number of retransmissions is small, the performance of the cell-free massive MIMO system is better due to the system sparsity. However, as the retransmission time increases, the performance of the centralized system even gradually exceeds that of the cell-free massive MIMO system. This is because the density of users and APs is large, and a large number of antennas in the centralized system base station can provide sufficient signal gain, while in the cell-free massive MIMO system, the gain provided by a small number of antennas is not enough to resist strong interference. Increasing the number of serving APs can effectively improve the strength of useful signals and suppress interference. Therefore, as the number of serving APs increases, the gap between the centralized system and the cell-free massive MIMO system can be rapidly narrowed. In addition, it should be noted that as the number of serving APs increases, the performance of the cell-free massive MIMO system is greatly improved under an upper limit, which is close to the performance of the centralized system when the retransmission time is large. Therefore, the number of serving APs for a user should also be set to a reasonable value.

[0113] Figure 7 The performance of the cell-free massive MIMO system and the centralized system is shown as the number of antennas and the number of serving APs vary. As the number of antennas increases, the system performance of both the cell-free massive MIMO system and the centralized system improves. When the number of antennas is small, the performance of the cell-free massive MIMO system is better than the centralized system due to the macro-diversity gain and channel hardening effect of the cell-free massive MIMO system. However, when the number of antennas and retransmissions increases, the performance of the centralized system gradually outperforms the cell-free massive MIMO system. This is because when there are enough base station antennas, the signal gain of the user is large enough, and with retransmissions, the interference signal gradually decreases, thereby improving the performance of the centralized system.

Claims

1. A HARQ-assisted unlicensed access method in a cell-free massive MIMO system. It is characterized in that Based on the user-centric network characteristics of the non-cellular architecture, on the basis of user clustering, users use the unlicensed random access mechanism to transmit data packets. At the same time, in order to alleviate the inherent user collision defects of the unlicensed access mechanism, the HARQ strategy is used for retransmission.

2. A HARQ-assisted unlicensed access method in a non-cellular massive MIMO system according to claim 1, It is characterized in that The specific embodiment of the HARQ strategy in the access method includes: before the user accesses the uplink, the frame calibration operation is first performed, and then the unlicensed random access mechanism is used to transmit the data packet. Subsequently, the AP receives and decodes the data packet and sends a feedback frame. After receiving the feedback frame, the user decides whether to perform a retransmission operation; if the AP decodes successfully, the user will receive an ACK feedback and does not need to perform a retransmission operation; conversely, if the AP data packet decoding fails, the user will receive a NACK feedback, which indicates that the user needs to retransmit the data within the maximum allowed retransmission number, and the maximum retransmission number is T latency is the delay constraint threshold, The round trip delay required for HARQ to perform a retransmission, T fa Indicates the frame alignment delay.

3. A HARQ-assisted unlicensed access method in a non-cellular massive MIMO system according to claim 1, It is characterized in that The specific steps of the unlicensed random access mechanism include: Establish spatial distribution models of users and APs and cell-free massive MIMO system models; Users and APs are clustered based on spatial distribution characteristics, and all users who select the same EDU are considered to be in the same cluster; Pilot allocation based on user clustering information; The AP receives the data signal containing pilot information sent by the user; The AP performs a spread spectrum operation by multiplying the received data signal containing pilot information by a pilot sequence; According to the obtained spread spectrum signal, the user's channel estimation is obtained using the minimum mean square error channel estimation method.

4. A HARQ-assisted unlicensed access method in a non-cellular massive MIMO system according to claim 3, It is characterized in that In the spatial distribution model, the EDU is the cluster center and is evenly distributed according to the uniform Poisson point process; users and APs gather around the EDU to form two independent Poisson cluster processes.

5. A HARQ-assisted unlicensed access method in a non-cellular massive MIMO system according to claim 4, It is characterized in that The actual spatial distribution of users and APs is simulated by the Thomas clustering process in the Poisson clustering process, where the child points are users and APs, the parent point is EDU, and it is assumed that the standard deviation of users and APs in the probability density function of the child points to the parent points is less than half of the average distance between adjacent cluster centers, and all users who choose the same EDU are considered to be in the same cluster.

6. A HARQ-assisted unlicensed access method in a non-cellular massive MIMO system according to claim 3, It is characterized in that Taking advantage of the user-centric network characteristics of the cellular-free architecture, each user can select multiple adjacent APs as its service AP set.

7. A HARQ-assisted unlicensed access method in a non-cellular massive MIMO system according to claim 5, It is characterized in that Based on the clustering result, users in the same cluster use orthogonal pilots, and non-orthogonal pilots may be used for users in different clusters.

8. A HARQ-assisted unlicensed access method in a non-cellular massive MIMO system according to claim 3, It is characterized in that For the pilot signal received by APm, by performing pilot correlation operations and utilizing the signal space sparseness of the non-cellular system, the received signal of user k at APm is approximately: Among them, p k represents the transmission power of user k, represents the channel response vector between user k and APm, I N represents the N-dimensional identity matrix, represents the additive white Gaussian noise vector, and represents the orthogonal pilot of length τ used by user k in the cluster.

9. A HARQ-assisted unlicensed access method in a non-cellular massive MIMO system according to claim 1, It is characterized in that During the retransmission process, based on the MRC receiver, the closed-form expression of the uplink SINR of user k is: in, S k represents the set of APs serving user k, p k 、p i 、p j denote the transmission power of users k, i, and j respectively, Φ p , Φ o Respectively represent the user sets in cluster p and o, σ 2 is the noise power, N is the number of antennas, represents the estimated value of the large-scale fading factor between user k and APm. Similarly, β i,m represents the estimated value of the large-scale fading factor between user i and APm, λ k,m , j,m denote the large-scale fading coefficients between user k, user j and APm, respectively, and η i,m =λ i,m -β i,m represents the channel estimation error coefficient between user i and APm.

10. A HARQ-assisted unlicensed access method in a non-cellular massive MIMO system according to claim 9, It is characterized in that During the retransmission process, the approximate expression of the interruption probability based on the lognormal moment matching method is: Among them, T latency is the delay constraint threshold, erfc(·) is the complementary error function, It represents the mean and variance of the log-normal distribution that SINR approximately obeys. The calculation results are as follows: in, And E[·] represents the expectation operator, (·) H represents the conjugate transpose operator, |·| represents the size of the determinant of the matrix; represents the estimated value of the channel response between user k and APm, is the estimated value of the small-scale fading coefficient between user k and APm; Let Y k =A k +B k +C k +D k ,but E[Y k ]=E[A k ]+E[B k ]+E[C k ]+E[D k ], in, and represents the estimated value of the channel response between user i and APm, represents the channel estimation error between user i and APm, g j,m represents the channel response vector between user j and APm;