A User-Centric Decellular Latency Analysis Method Considering Queuing

Abstract

This invention relates to the field of wireless communication technology, and more particularly to a user-centric decellularized network latency analysis method considering queuing. The method includes: constructing an uplink transmission system and signal model; employing maximum ratio combining (MRC) reception at access points; establishing a multi-access point cooperative reception model based on the user's corresponding service access point set; representing the service rate of the user queue as a function of the steady-state transmission success probability, and deriving the steady-state transmission success probability coupled with the state of interfering user queues; determining the fully unsaturated region that keeps the network in a stable state; calculating the average latency of user data packets, and determining the optimal transmission probability that minimizes the user's average latency based on the relationship between average latency and transmission probability. This method achieves a joint characterization of the transmission and queuing processes in user-centric decellularized networks, improving the accuracy and applicability of user latency analysis, and providing a basis for the design and optimization of latency-sensitive decellularized networks.

Description

Technical Field

[0001] This invention relates to the field of wireless communication technology, and more particularly to a user-centric decellular network latency analysis method that takes queuing into account. Background Technology

[0002] With the development of sixth-generation (6G) mobile communication technology, high-reliability low-latency communication (HRLLC) has become a core requirement for applications such as the Industrial Internet, autonomous driving, and telemedicine. Against this backdrop, user-centric decellularized network architectures are gaining increasing attention. This architecture provides services to users by coordinating multiple access points (APs), effectively improving link reliability and reducing transmission latency.

[0003] However, existing research mainly focuses on spectral efficiency optimization, access point selection, and beamforming design, and still has limitations in user latency analysis. Most performance analysis models only focus on the physical layer single packet transmission time, ignoring the packet queuing process in the Media Access Control (MAC) layer. In real-world networks, packet arrival and service are usually random, and without considering queuing buffer backlog, it is impossible to accurately predict the average user latency.

[0004] To accurately characterize the latency caused by queuing, some studies employ cross-layer analysis methods. Within this framework, instantaneous channel conditions determine whether data packets can be successfully transmitted, thus affecting queue length and queuing latency. However, in user-centric decellularized networks, each user is served by multiple access points (APs) simultaneously, and data packet transmission depends on the channel conditions between the user and all its serving APs. As the number of channels increases, the computational complexity of cross-layer analysis grows exponentially, making traditional cross-layer analysis methods difficult to apply directly.

[0005] To simplify calculations, some studies consider first calculating the physical layer steady-state transmission success probability for each user, and then further obtaining the queue service rate for each user, thereby calculating the latency. In this case, the steady-state transmission success probability of a user depends on the queue state of interfering users, and the queue state of each user is related to its own steady-state transmission success probability. Therefore, there is a serious nonlinear coupling relationship between the steady-state transmission success probabilities of each user, making traditional analysis methods difficult to apply directly. Although this problem has been solved in single-receiver scenarios, in user-centric decellularized networks, the multi-AP cooperative detection mechanism introduces large-scale fading of multiple links, making the analysis of steady-state transmission success probability more complex. In addition, in user-centric networks, to reduce interference between users, time-domain scheduling is usually used to schedule users to transmit on different time slots. At this time, the head-of-line (HOL) data packet must wait for the scheduled time slot before it can be sent. Its waiting time is affected by both the arrival time and the queue length at arrival, causing the service process to no longer be independently and identically distributed. Traditional analysis methods are difficult to use directly to calculate the queue service rate, thus increasing the complexity of latency analysis.

[0006] Therefore, how to establish a user-centric decellular network latency analysis method that can accurately characterize the queuing process while taking into account physical layer characteristics, so as to provide accurate latency prediction and guide system configuration, is a key technical problem that urgently needs to be solved in the field of wireless communication. Summary of the Invention

[0007] To address the limitations of existing user-centric latency analysis in decellularized networks, this invention provides a latency analysis and optimization method that considers both transmission and queuing processes.

[0008] The technical solution of this invention is as follows:

[0009] A user-centric decellular network latency analysis method that considers queuing includes the following steps:

[0010] Step 1: Construct the system and signal model. Considering the uplink transmission process, the uplink received signal is modeled based on the set of serving APs corresponding to the target user. In the model, the maximum ratio combining (MRC) receiving method is used to combine the received signal of the target user.

[0011] Step 2: User queue stability analysis. Based on the uplink reception results, calculate the service rate and steady-state transmission success probability for each user queue; according to the relationship between the service rate and the corresponding data packet arrival rate, determine the fully unsaturated region that can keep the network in a stable state.

[0012] Step 3: Average Latency Calculation and Optimization. Calculate the average latency of each user's data packets within the fully unsaturated region (i.e., the average latency of the user); based on the relationship between the average latency and the transmission probability, determine the optimal configuration parameters to minimize user latency.

[0013] The beneficial effects of this invention are as follows:

[0014] Compared with existing technologies, this invention can accurately characterize the queuing delay caused by dynamic changes in user queues and the transmission delay caused by channel randomness and multi-user interference, providing a reference for the design of delay-sensitive systems. Specifically, unlike methods that only focus on single-packet transmission time, this invention incorporates the queuing backlog of data packets in the buffer into delay modeling, resulting in a more accurate characterization of the delay for each user. Furthermore, compared with cross-layer analysis methods, this invention has lower computational complexity and is more suitable for practical communication system deployments. This is achieved by establishing the relationship between average user delay and transmission probability under stable conditions. Based on the relationship between the two, this invention determines the optimal transmission probability that ensures system stability while minimizing average user latency. The value of provides a reference for parameter design in actual systems. Attached Figure Description

[0015] Figure 1 A schematic diagram of a user-centric clustered decellularized network as considered in this invention;

[0016] Figure 2 This is a flowchart of the method of the present invention;

[0017] Figure 3 This is a graph showing the change in total throughput of 20 users in the same time slot as the transmission probability in an embodiment of the present invention;

[0018] Figure 4 This is a schematic diagram of the data packet delay composition according to an embodiment of the present invention;

[0019] Figure 5 This is an embodiment of the present invention. Schematic diagram;

[0020] Figure 6 This is a graph showing the success rate of steady-state transmission for users according to an embodiment of the present invention.

[0021] Figure 7 The following are graphs showing the relationship between the steady-state transmission success probability and the average delay as a function of the data packet arrival rate in embodiments of the present invention: (a) graph showing the change of the user's steady-state transmission success probability as a function of the data packet arrival rate; (b) graph showing the change of the user's average delay as a function of the data packet arrival rate. Detailed Implementation

[0022] The technical solution provided in this application will be further described below with reference to specific embodiments and accompanying drawings. The advantages and features of this application will become clearer from the following description.

[0023] The main parameters of this embodiment are annotated as shown in Table 1.

[0024] Table 1. Notes on main parameters.

[0025]

[0026] This invention considers user-centric clustered decellular network uplink transmission scenarios, such as... Figure 1 It includes several single-antenna users and access points (APs), each AP is equipped with multiple antennas, and users of the same color are assigned to the same time slot for transmission.

[0027] The following prerequisites must be met:

[0028] (1) The arrival process of user packets is assumed to be a Bernoulli process; each user is equipped with a data buffer to store newly arriving data packets, and each user's data packet will be kept in the user's buffer until successful transmission; assuming a frame contains Each user is assigned to one of the time slots for transmission; if the buffer is not empty, the user is assigned a probability... Transmission or retransmission is performed; each data packet transmission or retransmission occupies one time slot; the arrival of data packets can only occur at the beginning of a time slot, and the data packets complete their service and leave the queue at the end of the time slot.

[0029] (2) The AP adopts a capture model, assuming that each data packet is decoded independently at the receiving end. When the signal-to-interference-plus-noise ratio (SINR) of the target user is greater than a certain threshold, the data packet is determined to be successfully transmitted.

[0030] (3) When determining the approximate transmission success probability, the corresponding data packet reception success probability is determined for different active states of users in the network; and the data packet reception success probability is weighted based on the active probability of each user to obtain the approximate steady-state transmission success probability.

[0031] (4) In the queuing model, the service rate corresponding to the user queue is determined based on the data packet arrival and transmission process, satisfying: User Service rate is expressed as

[0032] ;

[0033] Among them, if the user is in the scheduling time slot The queue is not empty, and will be... Transmit with a probability of [a certain value]; For users The probability of successful steady-state transmission on a scheduling time slot The arrival rate of data packets in each time slot. This indicates the number of time slots per frame.

[0034] (5) The fully unsaturated region Defined as the transmission probability that the arrival rate of all user queues is strictly less than the corresponding service rate. The set, i.e. ,in A collection consisting of all users.

[0035] (6) The latency of a user data packet is defined as the number of time slots elapsed from the time the packet arrives in the queue until it is successfully transmitted and leaves the queue. When At that time, the user The average delay is expressed as

[0036] ;

[0037] The determination of the optimal network configuration parameters specifically involves: taking all unsaturated regions while ensuring the stability of all user queues. The maximum transmission probability that minimizes the average latency for users. .

[0038] A user-centric decellular network latency analysis method that considers queuing, such as Figure 2 As shown, it includes the following steps:

[0039] Step 1: Construct the system and signal model: Configure the specific scenario of the user-centric decellular wireless communication network, including the number of users, the number of access points (APs), the number of antennas at each AP, and the number of APs serving each user; pre-define the serving AP clusters corresponding to each user, with each cluster containing... The nearest access point (AP) to the user is selected; the arrival and service process of user data packets are determined according to the time-domain scheduling method; MRC reception is used to calculate the SINR of each user.

[0040] Consider an uplink transmission scenario in a user-centric, decellularized network. Assume the network has... Single antenna user and Each AP is equipped with [number] APs. A single antenna. (Set) and These represent the sets consisting of all users and the access points (APs), respectively. For example... Figure 1 As shown, each user is assigned the nearest... One AP service To serve users The set of APs that satisfy .

[0041] The arrival process of data packets follows a Bernoulli distribution, and the arrival rate of data packets in each time slot is [missing information]. Furthermore, the arrival process can only occur at the beginning of each time slot. Each user is equipped with an infinitely large data buffer to store newly arriving data packets, which remain in the buffer until successfully transmitted. Assume a frame contains... Each user is assigned to one of the time slots for transmission, and within the assigned time slot, transmission occurs with probability. Transmit its HOL data packet (if the buffer is not empty), where It does not change over time. Each data packet transmission (or retransmission) occupies one time slot, and the data packet only leaves the buffer at the end of the time slot. Time slots are allocated. The user set is denoted as .

[0042] Assuming the AP receives data in capture mode, meaning each data packet is decoded independently and other interference is considered noise, when the SINR at the AP receiver exceeds a threshold... The data packet was determined to have been transmitted successfully.

[0043] Consider allocating to time slots Users on Assume that the transmit power of each user is... time slot active user group AP Received signal Represented as

[0044] (1);

[0045] in User The transmitted signal, Indicates user With AP The channels between them have ,in For users With AP Euclidean distance between them It is a path loss parameter. These are small-scale fading parameters. This indicates that the mean is 0 and the variance is . Additive white Gaussian noise (AWGN).

[0046] Assuming the receiver uses MRC for reception, the user The estimated signal is represented as

[0047] (2);

[0048] in Indicates user Channels between it and all APs serving it. Indicates user The combined noise vector has dimensions of . And the joint receive vector satisfies

[0049] (3);

[0050] Here and These represent the transpose and conjugate transpose of a vector, respectively.

[0051] According to formula (3), in a given set of active users At that time, the user The receiver's SINR is represented as

[0052] (4);

[0053] Step 2: User queue stability analysis: Establish the mapping relationship between user queue service rate and steady-state transmission success probability; use saddle point approximation to handle multi-user interference, and adopt fixed-point iterative solution to obtain the steady-state transmission success probability coupled with the state of the interfering user queues; by comparing the data packet arrival rate and service rate, obtain the fully unsaturated region that keeps all user queues stable.

[0054] When user queues are in an unstable state, the latency of user data packets will tend towards infinity. Therefore, to ensure that latency analysis is meaningful, this invention first analyzes the stability of each user queue. According to existing research, for any queue, the queue is stable when its user arrival rate is strictly less than the service rate, at which point the user queue is not saturated. The system is in a stable state if and only if all user queues in the system are not saturated. Assume users... The service rate of the queue According to the system model considered in this invention, when a given data packet arrival rate At that time, service rate Depends on transmission probability Therefore, when Given a given state, ensuring that all user queues are in a stable state is equivalent to determining the transmission probability that ensures all user queues are not saturated. The range of values ​​for is defined as the fully unsaturated region. It can be expressed by the following formula:

[0055] (5);

[0056] Step 2.1 User queue service rate: based on the frequency of user data packet arrival. and the number of time slots per frame Establish user queue service rate With steady-state transmission success probability The mapping relationship between them.

[0057] To obtain the fully unsaturated region, the service rate for each user must first be calculated. For any data packet, the service process begins when it becomes a HOL (House of Excellence) packet and ends when it is successfully transmitted and leaves the queue. If the queue is empty when the packet arrives, it is an HOL packet, and the service process begins immediately; the packet will begin its first transmission attempt in the next scheduling slot. If the queue is not empty when the packet arrives, it will become an HOL packet after the previous packet is successfully transmitted and will wait. Its first transmission attempt begins after one time slot. Therefore, for the user The data packets, their service time for

[0058] (6);

[0059] in This indicates the number of attempts required for a successful transmission. This is the number of time slots required for a data packet to reach the next scheduled time slot. Assume the user... The probability that the queue is not empty is The average service time for data packets is

[0060] (7);

[0061] When the transmission probability is At that time, the user The probability of each transmission attempt succeeding is ,in Indicates user The probability of a successful steady-state transmission. Therefore, the number of attempts required for a successful transmission. Obtain the parameter as The geometric distribution, i.e. ,and

[0062] (8);

[0063] Since the arrival process of data packets follows a Bernoulli process, given that at least one data packet has arrived, its probability of arriving in any time slot within a frame is the same. Therefore, obey The uniform distribution between them, i.e.

[0064] (9);

[0065] Note Combining formulas (7)-(9), the user queue service rate for

[0066] (10);

[0067] Step 2.2 User Steady-State Transmission Success Probability: For the capture model, the saddle point approximation method and fixed point iteration are used to traverse all possible combinations of user queue states to obtain the actual steady-state transmission success probability of each user.

[0068] Substituting formula (3) into formula (4), given the active user set... At that time, the user The receiver's SINR is represented as

[0069] (11);

[0070] The interference term can be approximated as:

[0071] (12);

[0072] in subscript They represent AP respectively ,user and users .make

[0073] (13);

[0074] Given a set of active users The probability of successful conditional transmission at that time is expressed as

[0075] (14);

[0076] in These are auxiliary parameters, determined by the network topology and the receiver's SINR threshold. Number of antennas per AP and system signal-to-noise ratio Jointly decided, to satisfy:

[0077] (15);

[0078] because , It is a light-tailed random variable, therefore, according to the saddle point approximation method, Approximately

[0079] (16);

[0080] in It is a distractor. The cumulative generating function (CGF) The second derivative, Satisfy the equation ,in yes The first derivative.

[0081] Consider all possible sets of active users ,user Probability of successful steady-state transmission It can be represented as

[0082] (17);

[0083] in Indicates user The probability of being active on a scheduling slot. If the user The queue is in an unsaturated state, and the probability that the queue is not empty in the scheduling time slot is: At this time there is If the queue is saturated, then the queue will never be empty in the scheduling slot. Combining formulas (16) and (17), the user The probability of successful steady-state transmission is

[0084] (18);

[0085] in

[0086] (19);

[0087] As can be seen from formulas (18) and (19), the steady-state success transmission probability depends on the state of other user queues scheduled to the same time slot. Since the user queue state cannot be obtained when the transmission success probability is unknown, it is necessary to enumerate all... The possible states are determined by exhaustively searching through all possible queue states to obtain the actual queue states of all users.

[0088] (20);

[0089] Specifically, using Indicates user The queue state, where and These represent queue unsaturated and queue saturated, respectively. For each state... Probability of successful transmission for all users The queue state can be obtained through fixed-point iteration according to formula (20). Next, the actual service rate of the user is calculated according to formula (10), and it is checked whether the arrival rate of each user is strictly less than its service rate, thereby obtaining a new queue state. .when When established, the exhaustive search process terminates, at which point the actual steady-state transmission success probability for all users can be obtained. Service rate .

[0090] Step 2.3 Unsaturated Area: Based on the arrival rate of each user Compared with the service rate calculated above Numerical search or optimization algorithms are used to determine the range of transmission probability values ​​that ensures all user queues remain stable, i.e., the defined unsaturated region. .

[0091] To ensure system stability, transmission probability It is necessary to start from the fully unsaturated region defined by formula (5) The two boundary values ​​of this region can be obtained by solving the following problem:

[0092] (twenty one);

[0093] This problem can be solved numerically, such as through evolutionary algorithms. Ultimately, the fully unsaturated region... It can be represented as

[0094] (twenty two);

[0095] Figure 3 The diagram illustrates the total throughput of 20 users scheduled to the same time slot in a randomly generated network with 80 users. The fully unsaturated interval of the network is calculated using formula (22). It can be seen that when When the total throughput equals the sum of the packet arrival rates, it proves that all user queues are in a stable state. or When the total throughput is less than the sum of the packet arrival rates, it proves that at least one user's queue is saturated and the system is in an unstable state.

[0096] Step 3 Average Latency Calculation and Optimization: Decompose the target latency based on the queuing process, and calculate the average latency of each user in the unsaturated area by combining Little's theorem and the service rate derived above. On this basis, with the goal of minimizing user latency, explore the optimal transmission probability value in the unsaturated area. Under the premise of ensuring network stability, achieve low-latency communication by reasonably adjusting the transmission probability to improve the overall transmission efficiency of the system and user experience.

[0097] like Figure 4 As shown, the latency of a data packet is defined as the number of time slots elapsed from the start of its arrival in the queue to its successful transmission. The latency of data packets can be divided into two parts:

[0098] (1) : The number of time slots from when a data packet arrives in the queue until its first transmission attempt.

[0099] (2) : The number of time slots from the first attempt to transmit the data packet until it is successfully transmitted.

[0100] When the arrival process follows a Bernoulli distribution, the average latency observed by any newly arriving data packet in the queue is equal to the average latency of all data packets for that user. Therefore, taking a newly arriving data packet as the observation object, the user... The average delay can be expressed as

[0101] (twenty three);

[0102] Next, calculate separately. and .

[0103] (1) Any newly arrived data packet in the queue must wait for all data packets that arrived before it to complete transmission. If a data packet in the queue is not transmitted or fails to transmit in a scheduled time slot, it must wait. Each time slot, therefore It can be represented as

[0104] (twenty four);

[0105] in It represents the time interval from the arrival time of the data packet to the first scheduling time slot after the currently being served data packet is successfully transmitted, and is used to assist in the calculation of latency; This indicates the number of packets queued in the queue when a packet arrives (excluding packets being served). According to Little's theorem, we have...

[0106] (25);

[0107] like Figure 5 As shown, It can be further expressed as

[0108] (26);

[0109] in This indicates the remaining number of transmissions for the data packet currently being served, satisfying...

[0110] (27);

[0111] Because a successful transmission requires a certain number of attempts. ,have

[0112] (28);

[0113] Combining formula (9) and formulas (25)-(27),

[0114] (29);

[0115] (2) In the system considered in this invention, each failed attempt, starting from the first transmission attempt, will result in... The delay is one time slot, and the last successful transmission also occupies one time slot. Therefore, there is

[0116] (30);

[0117] Combining formulas (23), (28)-(30), the user Average latency It can be represented as

[0118] (31);

[0119] It is evident that the user's average latency is affected by the data packet arrival rate. Number of time slots per frame Transmission probability The steady-state transmission success probability of each user is affected by both factors, and the steady-state transmission success probability is further affected by the combined factors of these factors. , and This is jointly determined. Therefore, given a packet arrival rate... and the number of time slots per frame At that time, the transmission probability can be adjusted. To optimize latency. Therefore, in practical systems, in order to ensure system stability while minimizing the average latency for each user, the transmission rate... The fully unsaturated region described in step 2 should be considered based on the actual situation of the system. Appropriate selection of internal components is necessary to improve the overall performance of the system.

[0120] It should be noted that the latency analysis in this invention is limited to the case where all user queues are not saturated. In this case, according to formulas (18) and (19), the steady-state success transmission probability is... With transmission probability Irrelevant. Combining formula (31), we can see that the average delay... With transmission probability The probability decreases monotonically as the probability increases. Therefore, in a practical system, the optimal transmission probability should be selected. To minimize the total system latency.

[0121] Example Result Analysis

[0122] Figure 6 The diagram illustrates the topology of a randomly generated network. All users and access points (APs) are randomly distributed within a 1m square area, with circles representing users and triangles representing APs. Users of the same color are assigned to the same time slot for transmission. Each user's steady-state transmission success probability is labeled with both the theoretical approximation and simulation results (format: theoretical approximation / simulation result). The diagram shows that the theoretical approximation and simulation results are the same or very close, indicating that formula (18) can accurately approximate the steady-state transmission success probability of a user. Furthermore, it can be observed that the steady-state transmission success probability of red users is much higher than that of blue users. This is because there are more blue users than red users, thus blue users experience stronger interference due to the larger number of users transmitting simultaneously.

[0123] Figure 7 Shown in Figure 6 The figure shows the relationship between the steady-state transmission success probability and average delay of four randomly selected users and the data packet arrival rate under the same network topology. As can be seen from the figure, the theoretical analysis results of the steady-state transmission success probability and average delay for each user are in excellent agreement with the simulation results, verifying the accuracy of the analysis in this invention. Furthermore, from... Figure 7 (a) It can be seen that the user's steady-state transmission success probability decreases as the packet arrival rate increases. As shown in formula (19), a higher arrival rate increases the user's active probability, leading to more interference and thus reducing the steady-state transmission success probability. Furthermore, according to formula (31), the average delay... Follow The decrease increases with the increase, and Figure 7 The trend is consistent with that in (b).

[0124] The above description is merely a description of preferred embodiments of this application and is not intended to limit the scope of this application in any way. Any changes or modifications made by those skilled in the art based on the above-disclosed technical content should be considered as equivalent and valid embodiments and fall within the scope of protection of the technical solution of this application.

Claims

1. A user-centric latency analysis method for decellular networks that considers queuing, characterized in that, Includes the following steps: Step 1: Construct system and signal model; Considering the uplink transmission process, model the uplink received signal according to the set of service access points corresponding to the target user, and use the maximum ratio combining reception method in the model to combine the received signal of the target user; Step 1 is as follows: Configure specific scenarios for user-centric decellular wireless communication networks, including the number of users, the number of access points (APs), the number of antennas at each AP, and the number of APs serving each user; pre-define the serving AP clusters corresponding to each user, with each cluster containing... The AP closest to the user; Based on the time-domain scheduling method, the arrival and service process of user data packets are determined; MRC reception is used to calculate the signal-to-interference-plus-noise ratio (SINR) for each user; Consider an uplink transmission scenario in a user-centric decellularized network; assume the network has Single antenna users and Each AP is equipped with [number] APs. Antenna; Set and Let these represent the sets of all users and the AP, respectively; each user is represented by the nearest AP. One AP service To serve users The set of APs that satisfy ; The arrival process of data packets follows a Bernoulli distribution, and the arrival rate of data packets in each time slot is [missing information]. Furthermore, the arrival process can only occur at the beginning of each time slot; each user is equipped with an infinitely large data buffer to store newly arriving data packets, which remain in the buffer until successfully transmitted; assuming a frame contains Each user is assigned one time slot for transmission, and if the buffer is not empty, transmission occurs in the assigned time slot with probability. Transmit its HOL data packet, in which It does not change over time; each data packet transmission or retransmission occupies one time slot, and the data packet only leaves the buffer at the end of the time slot; allocated to a time slot The user set is denoted as ; Assuming the AP receives data in capture mode, meaning each data packet is decoded independently and other interference is considered noise; when the SINR at the AP receiver exceeds a threshold... The data packet was determined to have been successfully transmitted. Consider allocating to time slots Users on Assume that the transmit power of each user is time slot active user group AP Received signal Represented as: (1) in User The transmitted signal, Indicates user With AP The channels between them have ,in For users With AP Euclidean distance between them It is a path loss parameter. These are small-scale fading parameters; This indicates that the mean is 0 and the variance is . Additive white Gaussian noise (AWGN); Assuming the receiver uses MRC for reception, the user The estimated signal is represented as (2) in Indicates user Channels between it and all APs serving it. Indicates user The combined noise vector has dimensions of . And the joint receive vector satisfies (3) Here and These represent the transpose and conjugate transpose of a vector, respectively. According to formula (3), in a given set of active users At that time, the user The receiver's SINR is represented as (4); Step 2: User queue stability analysis; Based on the uplink reception results, calculate the service rate and steady-state transmission success probability corresponding to each user queue; Based on the relationship between the service rate and the corresponding data packet arrival rate, determine the fully unsaturated region that can keep the network in a stable state; In step 2, the fully unsaturated region: Assuming user The service rate of the queue When a given data packet arrival rate At that time, service rate Depends on transmission probability Ensuring that all user queues are in a stable state is equivalent to determining the transmission probability that ensures all user queues are not saturated. The range of values ​​for is defined as the fully unsaturated region. It can be expressed by the following formula: (5); In step 2, the user queue service rate analysis method is as follows: Based on the frequency of user data packets arriving and the number of time slots per frame Establish user queue service rate With steady-state transmission success probability The mapping relationship between them; For users The data packets, their service time for (6) in This indicates the number of attempts required for a successful transmission. It is the number of time slots that need to be waited for from the time a data packet is sent to the time slot to reach the next scheduled time slot; Assuming user The probability that the queue is not empty is The average service time for data packets is (7) user The probability of each transmission attempt succeeding is ,in Indicates user The probability of successful steady-state transmission; Number of attempts required for a successful transmission Obtain the parameter as The geometric distribution, i.e. ,and (8) obey The uniform distribution between them, i.e. (9) Combining formulas (7)-(9), the user queue service rate for (10); In step 2, the method for analyzing the probability of successful steady-state transmission by the user is as follows: For the capture model, a saddle point approximation method and fixed point iteration are used to traverse all possible combinations of user queue states to obtain the actual steady-state transmission success probability of each user. Substituting formula (3) into formula (4), given the active user set... At that time, the user The receiver's SINR is represented as (11) The interference term can be approximated as: (12) in subscript They represent AP respectively ,user and users ;make (13) Given a set of active users The probability of successful conditional transmission at that time is expressed as (14) in These are auxiliary parameters, determined by the network topology and the receiver's SINR threshold. Number of antennas per AP and system signal-to-noise ratio Jointly decided, to satisfy: (15) because , It is a light-tailed random variable, therefore, according to the saddle point approximation method, Approximately (16) in It is a distractor. Cumulative generating function The second derivative, Satisfy the equation ,in yes The first derivative; Consider all possible sets of active users ,user Probability of successful steady-state transmission It can be represented as (17) in Indicates user The probability of being active in a scheduling slot; If user The queue is in an unsaturated state, and the probability that the queue is not empty in the scheduling time slot is: At this time there is If the queue is saturated, then the queue will never be empty in the scheduling slot. ; Combining formulas (16) and (17), the user The probability of successful steady-state transmission is (18) in (19) The probability of a steady-state successful transmission depends on the state of other user queues scheduled to the same time slot. The actual queue states of all users are obtained through exhaustive search. (20) use Indicates user The queue state, where and These represent queue unsaturated and queue saturated, respectively; for each state... Probability of successful transmission for all users The formula (20) is obtained through fixed-point iteration; Next, the actual service rate of each user is calculated according to formula (10), and it is checked whether the arrival rate of each user is strictly less than its service rate, thereby obtaining a new queue state. ;when Upon establishment, the exhaustive search process terminates, and the actual steady-state transmission success probability of all users is obtained. Service rate ; In step 2, based on the arrival rate of each user With service rate Numerical search or optimization algorithms are used to determine the range of transmission probability values ​​that ensures all user queues remain stable, i.e., the fully unsaturated region. ; To ensure system stability, transmission probability It is necessary to start from the fully unsaturated region defined by formula (5) The two boundary values ​​of this region can be obtained by solving the following problem: (21) This problem was solved numerically, ultimately finding the fully unsaturated region. Represented as (22); Step 3: Average Latency Calculation and Optimization; Calculate the average latency of each user data packet within the fully unsaturated region; Based on the relationship between the average latency and the transmission probability, determine the optimal configuration parameters to minimize user latency; In step 3, the target latency is decomposed according to the queuing process, and the average latency of each user in the unsaturated area is calculated by combining Little's theorem and the service rate. On this basis, with the goal of minimizing user latency, the optimal transmission probability value is explored in the unsaturated area. Under the premise of ensuring network stability, low-latency communication is achieved by reasonably adjusting the transmission probability, so as to improve the overall transmission efficiency of the system and user experience. The latency of a data packet is defined as the number of time slots elapsed from the start of its arrival in the queue to its successful transmission. The latency of the data packets is divided into two parts: : The number of time slots from when a data packet arrives in the queue until its first transmission attempt; The number of time slots from the first attempt to transmit the data packet until it is successfully transmitted; Using a newly arrived data packet as the observation object, the user The average delay is expressed as (23) in, (1) : Represented as (24) in It represents the time interval from the arrival time of the data packet to the first scheduling time slot after the currently being served data packet is successfully transmitted, and is used to assist in the calculation of latency; This indicates the number of packets in the queue that are not currently being served when the packet arrives; according to Little's theorem, we have (25) Further expressed as (26) in This indicates the remaining number of transmissions for the data packet currently being served, satisfying... (27) Because a successful transmission requires a certain number of attempts. ,have (28) Combining formula (9) and formulas (25)-(27), (29) (2) as follows: (30) Combining formulas (23), (28)-(30), the user Average latency Represented as (31)。 2. The user-centric decellular network latency analysis method considering queuing as described in claim 1, characterized in that, When given packet arrival rate and the number of time slots per frame At that time, by adjusting the transmission probability To optimize latency; in practical systems, in order to ensure system stability while minimizing the average latency for each user, the transmission rate... Based on the actual situation of the system, in the fully unsaturated region described in step 2 Internal selection is used to improve the overall performance of the system.

3. The user-centric decellular network latency analysis method considering queuing as described in claim 2, characterized in that, The optimal transmission probability that minimizes the total system delay. .

Images