A QoS-CSMA / CA Improvement Method for Haptic Services

Abstract

This invention belongs to the field of wireless communication technology, specifically relating to an improved QoS-CSMA / CA method for haptic services. The invention utilizes Pareto distribution to model the data arrival process, and optimizes latency and throughput during network transmission by introducing a token bucket model and a dynamic backoff factor adjustment method. It dynamically adjusts the backoff factor in real-time, statistically analyzing link queue status and latency violations to ensure low latency and high throughput transmission performance. The invention implements a more efficient backoff mechanism and collision avoidance strategy at the MAC layer, meeting users' statistical latency requirements with less resource consumption while improving network throughput. It can efficiently handle the burstiness and heavy-tailed characteristics of haptic services, providing reliable QoS guarantees and higher transmission efficiency for wireless communication systems.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication technology, and specifically relates to an improved QoS-CSMA / CA method for haptic services. Background Technology

[0002] As a next-generation ultra-low latency and high-reliability communication technology, tactile internet is widely used in scenarios such as telemedicine, autonomous driving, industrial control, and virtual reality, placing extremely high demands on network latency, jitter, and stability. However, the data arrival process of tactile services often exhibits bursty and heavy-tailed distribution characteristics, mainly due to its high-frequency and latency-sensitive interactive characteristics. The system may generate a large amount of tactile information in a short period. To reduce the communication burden, compression techniques such as perceptual dead-zone coding are often used to reduce the average data rate. However, this type of coding concentrates the transmission of data packets when significant tactile changes are detected, thus exacerbating the volatility of instantaneous arrival rate and arrival interval. That is, a large amount of data may be generated in a short period, while the traffic volume is low at other times, leading to increased burstiness. Simultaneously, the uncertainty of different interaction intensities and user behavior may cause occasional but violent data bursts, making the probability of extremely large data flow events much higher than the conventional distribution, exhibiting typical heavy-tailed characteristics: the probability of large data packets or high arrival rates decays slowly, resulting in a heavy tail. Existing wireless network protocols struggle to provide stable QoS guarantees when facing bursty traffic and latency constraints.

[0003] WiFi (IEEE 802.11), with its advantages of high bandwidth, low cost, and ease of deployment, is very suitable for indoor network requirements and is one of the main application protocols for in-vehicle and indoor networks. Furthermore, with technological advancements, modern WiFi (especially WiFi 6 and later versions) has significantly improved in latency, reliability, and bandwidth, enabling it to support haptic services. WiFi 6 introduces key technologies such as OFDMA, MU-MIMO, and BSS Coloring, effectively reducing congestion, improving multi-user concurrency efficiency, and lowering transmission latency. Among these, the core CSMA / CA protocol access mechanism of WiFi coordinates channel access between terminals through carrier sense, random backoff, and acknowledgment mechanisms, effectively avoiding packet collisions in traditional network environments. However, the CSMA / CA mechanism has significant limitations in scenarios with bursty and heavy-tailed service characteristics. Bursting services can easily lead to a large number of terminals simultaneously competing for the channel, causing frequent collisions and retransmissions. The significantly increased probability of collisions triggers exponential backoff mechanisms, resulting in prolonged backoff times and increased transmission latency for critical services. Heavy-tailed traffic can cause some services to occupy the channel for extended periods, leading to uneven resource allocation and significant latency jitter. Furthermore, traditional CSMA / CA employs a fixed backoff strategy, which struggles to handle dynamic and complex access demands, limiting its performance in environments with bursty traffic and heavy-tailed traffic characteristics.

[0004] To address these challenges, several improvement schemes have been proposed. IEEE 802.11e introduces priority queues to differentiate service types, but its static parameter settings are difficult to adapt to dynamically changing network environments. While the next-generation WiFi 6 standard employs advanced technologies such as OFDMA and MU-MIMO to improve throughput, its underlying mechanism still relies on the traditional CSMA / CA mechanism, failing to fundamentally solve the backoff latency problem. Some research attempts to use machine learning algorithms to optimize backoff strategies, but due to high computational complexity, large-scale deployment on resource-constrained terminal devices is difficult. Therefore, optimizing the CSMA / CA mechanism in the WiFi environment to adapt to bursty and heavy-tailed traffic characteristics is crucial for QoS assurance of haptic services.

[0005] The performance of the CSMA / CA mechanism largely depends on the statistical characteristics of the backoff duration, with the average backoff duration called the backoff factor. Meanwhile, statistical latency QoS, composed of latency bounds and latency violation probabilities, is another important indicator. Since human perception of tactile stimuli is much faster than that of hearing and vision, typically perceiving latency within milliseconds, tactile services are particularly sensitive to latency. Average latency cannot effectively reflect tail performance, while statistical latency analysis can provide more explicit service guarantees, thus better meeting the stringent latency requirements of tactile services. Furthermore, in application scenarios such as high-frequency bidirectional interaction and complex multimodal feedback, tactile services also have high requirements for system throughput. Tactile information is typically generated continuously at a high sampling rate; once multiple degrees of freedom or multi-point tactile feedback are involved, the data volume increases significantly. Insufficient network throughput can lead to data loss or increased compression ratios, thus affecting tactile resolution and stability. Therefore, in addition to ensuring low latency, maintaining sufficient throughput is equally important for tactile services. Therefore, in order to effectively reduce the probability of latency violations and improve throughput in haptic services, dynamically adjusting the backoff factor based on statistical latency QoS is an effective optimization method.

[0006] In summary, although the CSMA / CA protocol has been improved in many ways with the rapid development of the times, it remains a challenge to provide the maximum throughput while meeting the diverse latency QoS requirements of tactile services with bursty traffic and heavy-tailed characteristics. Summary of the Invention

[0007] To overcome the above problems, this invention provides an improved QoS-CSMA / CA method for haptic services. By analyzing the bursty characteristics of network traffic, this invention uses Pareto distribution to model the data packet arrival process and introduces a token bucket model to shape the arriving traffic, which can better cope with bursty traffic and heavy-tailed characteristics. Combining effective capacity theory, effective bandwidth under the token bucket model, and delay violation probability, the backoff factor is dynamically adjusted to optimize the backoff process.

[0008] The objective of this invention is achieved through the following technical solution:

[0009] A QoS-CSMA / CA improvement method for haptic services includes the following steps:

[0010] Step 1, modeling the arrival process of tactile services, specifically includes the following:

[0011] Step 1.1: The Pareto distribution is used to accurately model the burst traffic and heavy-tail characteristics in the haptic service, where the probability density function of the Pareto distribution is:

[0012]

[0013] in, This represents the number of data packets arriving within a unit time slot, when the shape parameter... The mean of the Pareto distribution yes:

[0014]

[0015] According to shape parameters and average arrival rate of services Calculate scale parameters for:

[0016]

[0017] Random flow data is generated using the inverse transform method:

[0018]

[0019] in for Uniformly distributed random numbers Indicates link The number of data packets arriving per unit time slot;

[0020] Step 1.2, based on the link Calculate the delay bound and delay violation probability requirement for the link. Initial QoS parameters :

[0021]

[0022] in Indicates link The time delay boundary, Indicates link Long-term delays violate probability requirements;

[0023] Step 1.3, the token bucket parameter initialization method based on burst traffic and heavy-tailed distribution statistical characteristics, calculates the Pareto distribution mean. and variance and in combination with network load Determine the bucket capacity and token generation rate The initial bucket capacity and initial token generation rate for:

[0024]

[0025]

[0026] in Indicates the adjustment coefficient;

[0027]

[0028] Step 1.4, adjusting parameters adopts a two-stage iterative strategy:

[0029] The first stage involves gradually reducing the bucket capacity to find the minimum requirement that is met. Iteratively adjust bucket capacity The formula is as follows:

[0030]

[0031] in Indicates the number of iterations. The reduction factor representing the bucket capacity. Indicates the first The bucket capacity after each iteration is determined by stopping the iteration when the token bucket starts dropping packets, thus determining the minimum bucket capacity.

[0032] The second stage involves gradually reducing the token generation rate to find the minimum rate that meets the requirements. Iteratively adjust the token generation rate The formula is as follows:

[0033]

[0034] in The reduction factor representing the token generation rate. Indicates the first The token generation rate after each iteration is determined by stopping the iteration when the token bucket starts dropping packets, thus determining the minimum token generation rate.

[0035] Step 2: Dynamic QoS awareness, specifically including:

[0036] Step 2.1, calculate the backoff factor update cycle. Midlink Short-term latency violates probability requirements :

[0037]

[0038] in This indicates the period during the backsliding factor update cycle. In the middle, link The delay exceeds the binding The number of time slots; Indicates the total number of time slots;

[0039] Step 2.2, Dynamically update the backoff factor update cycle QoS parameters in :

[0040]

[0041] Step 3: Estimating the effective bandwidth based on the token bucket model, where the effective bandwidth... The calculation uses the following formula:

[0042]

[0043] in, It is an adjustment curve for the arrival process. , Indicates a time slot;

[0044] Step four, dynamic adjustment of the link backoff factor, specifically includes:

[0045] Step 4.1, Calculation of effective link capacity:

[0046] A two-state Markov chain model is established based on the service process. In the Markov chain model, state 1 represents the link. A transmission rate of 1 indicates a link with a state of 0. The transmission rate is 0; the duration of the fallback process and the transmission duration of the link both follow the mean value. Exponential distribution with mean 1; transition rate matrix for:

[0047]

[0048] According to avoidance factors and QoS parameters ,link Effective capacity is:

[0049]

[0050] in, Indicates link The average arrival rate;

[0051] Step 4.2, Parameters and Adaptive updates:

[0052] The parameters required to calculate the optimal backoff factor are obtained using the sub-gradient algorithm:

[0053]

[0054]

[0055] in, and In the sub-gradient algorithm, the first... The Lagrange multiplier in the next iteration, It is a variation of the avoidance factor. This transforms a non-convex optimization problem into a convex optimization problem. Indicates link Effective bandwidth and This indicates the step size of the corresponding Lagrange multiplier, and is set. and To ensure the rapid convergence of the sub-gradient method;

[0056] Step 4.3 comprehensively considers the current network throughput, queue status, and QoS requirements, and obtains the optimal backoff factor by solving for it, achieving the best balance between transmission efficiency and QoS guarantee. The backoff factor... The calculation is as follows:

[0057] First, calculate using the following formula. :

[0058]

[0059] in, , Indicates link In the The amount of data successfully transmitted within a single backoff factor adjustment cycle;

[0060] Then through To obtain the optimal backoff factor;

[0061] Step 4.4, the retreat time is calculated as follows:

[0062]

[0063] in, Indicates the length of a time slot. It is a random variable that follows an exponential distribution, and its mean is related to the backslip factor. related;

[0064] Step 5 involves real-time monitoring and dynamic scheduling of the link transmission status; specifically including:

[0065] Step 5.1: At the beginning of each time slot, update the queue length of each link in real time based on the number of newly arrived data packets and the number of successfully transmitted data packets. The formula for updating the queue status for each time slot is:

[0066]

[0067] in, This indicates the queue length at the end of the previous time slot. Indicates link In the time slot queue length, This indicates the amount of data successfully transmitted and removed from the queue in the current time slot. Ensure that the queue length does not become negative;

[0068] Step 5.2, Delay violation judgment criteria:

[0069] when hour, The value is increased by 1 if it is not increased, otherwise it remains unchanged. Indicates link Average arrival rate Indicates link The delay exceeds the binding Number of time slots;

[0070] Step 5.3: The system sorts each link queue in descending order according to its length in each time slot to determine the priority. Then, it sequentially checks whether each link simultaneously meets the two conditions of the backoff timer being zero and the channel being idle. If both conditions are met, transmission begins immediately; otherwise, the countdown continues until the transmission conditions are met.

[0071] 2. The QoS-CSMA / CA improvement method for haptic services according to claim 1, characterized in that, in step 5.2, the number of delay violations of the link within the current parameter adjustment period is accumulated. This estimates the actual latency violation probability of the link, reflecting the current latency violation situation of the link and providing feedback for subsequent parameter adjustments.

[0072] Compared with the prior art, the present invention has the following beneficial effects:

[0073] This invention addresses the challenges of handling bursty traffic and heavy-tailed characteristics in haptic services by proposing a dynamic backoff factor adjustment method based on latency violation probability. By analyzing the bursty characteristics of network traffic, this invention models the packet arrival process using a Pareto distribution and introduces a token bucket model to shape the arriving traffic, thus better handling bursty traffic and heavy-tailed characteristics. Combining effective capacity theory, effective bandwidth under the token bucket model, and latency violation probability, the backoff factor is dynamically adjusted to optimize the backoff process. Compared to traditional fixed backoff mechanisms, this invention effectively avoids excessively long backoff times and low throughput in haptic services, maintaining stable network operation. Attached Figure Description

[0074] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the content of the embodiments of the present invention and these drawings without creative effort.

[0075] Figure 1 This is a flowchart of the QoS-CSMA / CA process.

[0076] Figure 2 For when setting , , (unit: When ), the delay violation probability box plot of the present invention.

[0077] Figure 3 For when setting , , (unit: When ), the measured average arrival and service rates of each link, and the evolution of the average arrival and service rates of link 1.

[0078] Figure 4 For when setting , , (unit: When ), the network latency performance.

[0079] Figure 5 For when setting And the statistical latency QoS requirement for the first 30% of links is , The average throughput trend under a fixed network size when other links have no QoS constraints. Detailed Implementation

[0080] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present invention and not the entire structure.

[0081] Example 1

[0082] An improved CSMA / CA protocol method for haptic services, based on latency QoS guarantees, dynamically adjusts the backoff factor through token bucket flow control and real-time statistics of link queue status and latency violations. This allows for distributed satisfaction of diverse latency QoS requirements without incurring signaling overhead for traffic information from other devices or scheduling information from access points (APs). It also optimizes the performance of the traditional CSMA / CA protocol in network environments with bursty and heavy-tailed traffic characteristics. The flowchart of this invention is shown below. Figure 1 As shown, the method includes the following steps:

[0083] Step 1: The arrival process of haptic data often exhibits bursty and heavy-tailed distribution characteristics, making it difficult to effectively describe these characteristics using the traditional uniform traffic assumption. Therefore, this invention employs a Pareto distribution to model the data packet arrival process of haptic services (not limited to the Pareto distribution). An independent token bucket model is constructed for each link, initializing the bucket capacity and token generation rate based on the statistical characteristics of bursty traffic and heavy-tailed distribution. The token bucket parameters are gradually decreased until the token bucket begins to drop data packets, thus determining the minimum token bucket parameters. By adjusting the token generation rate and bucket capacity, common network problems such as bursty traffic and heavy-tailed traffic can be effectively addressed, ensuring stable and efficient network operation.

[0084] Modeling the arrival process of haptic services includes the following:

[0085] (1) In this invention, the Pareto distribution is used to accurately model the burst traffic and heavy-tail characteristics in tactile services (not limited to the Pareto distribution). Compared with the traditional Poisson distribution, the Pareto distribution can more accurately characterize the burst traffic characteristics and heavy-tail characteristics commonly found in real networks. The probability density function of the Pareto distribution is:

[0086]

[0087] in, This represents the number of data packets arriving per unit time slot (i.e., the amount of upper-layer service data flowing to the radio transmission queue within one time slot length), when the shape parameter The mean of the Pareto distribution yes:

[0088]

[0089] According to shape parameters and average arrival rate of services Calculate scale parameters for:

[0090]

[0091] To ensure that the traffic modeling matches actual network conditions, the packet arrival data is generated using the inverse transformation method, an efficient and reliable random number generation technique. This method first generates uniformly distributed random numbers. Then, by applying the inverse function transformation of the Pareto distribution, data with bursty flow and heavy-tailed characteristics that meet the requirements are obtained. This implementation method is computationally efficient and ensures that the statistical properties of the generated data are consistent with the theoretical model.

[0092] Random flow data is generated using the inverse transform method:

[0093]

[0094] in for Uniformly distributed random numbers Indicates link The number of data packets arriving per unit time slot;

[0095] This invention utilizes the inverse transform method to generate random flow data conforming to a Pareto distribution. This method can effectively reflect different flow demands, sudden flow events, and heavy-tailed characteristics.

[0096] (2) According to the link Calculate the delay bound and delay violation probability requirement for the link. Initial QoS parameters :

[0097]

[0098] in Indicates link The time delay boundary, Indicates link Long-term delays violate probability requirements;

[0099] (3) A token bucket parameter initialization method based on burst traffic and heavy-tailed distribution statistical characteristics is proposed, which calculates the mean of the Pareto distribution. and variance and in combination with network load Determine the bucket capacity and token generation rate The initial bucket capacity and initial token generation rate for:

[0100]

[0101]

[0102] in This represents the adjustment coefficient. Indicates network load;

[0103] Pareto distribution statistic:

[0104]

[0105]

[0106] (4) A two-stage iterative strategy is adopted to adjust the parameters:

[0107] The first stage involves gradually reducing the bucket capacity to find the minimum requirement that is met. The second stage involves gradually reducing the token generation rate to find the minimum value that meets the requirements. This phased approach avoids interference between parameters; the iterative method for adjusting the token bucket parameters is as follows:

[0108]

[0109]

[0110] in Indicates the number of iterations. and This represents the reduction factor of the token bucket parameter. Indicates the first Bucket capacity after the next iteration. Indicates the first The token generation rate after each iteration stops when the token bucket starts dropping packets, thus determining the minimum token bucket parameter. The goal is to find the minimum token bucket parameter that meets business requirements without packet loss, achieving effective traffic control, minimizing the resource overhead required for traffic control, and ensuring stable and reliable business performance.

[0111] Step Two: Dynamic QoS Awareness Method. This method decomposes long-term QoS constraints into a series of progressive QoS constraints with varying parameter update cycles. By real-time statistics of link queue status and latency violations, the progressive QoS requirements are recursively updated within each parameter update cycle. To better characterize the stringency of current latency service quality requirements, this invention introduces a progressive QoS parameter. Based on the large deviation theory, by analyzing the tail probability of queue length or queuing latency distribution, this parameter is determined by both the latency limit and the latency violation probability within the current cycle. When the latency violation probability is low or the latency limit is small, the corresponding QoS parameter will increase, indicating a more stringent latency control requirement for the current cycle. The QoS parameter within each cycle is dynamically adjusted based on the latency limit and latency violation probability to better meet the QoS requirements of haptic services; specifically including:

[0112] (1) Calculate the update cycle of the parameter (set backoff factor). Midlink Short-term latency violates probability requirements :

[0113]

[0114] in This indicates the parameter (backoff factor) update cycle. In the middle, link The delay exceeds the binding The number of time slots, via the link Local measurements of the transmitter; This represents the total number of time slots; obviously, in each link... Able to base on its Local measurements are updated recursively in each parameter update cycle;

[0115] (2) Dynamic update parameter update cycle QoS parameters in :

[0116] Calculate the delay violation probability requirement in the parameter update cycle. Then calculate the QoS parameters using the following formula. :

[0117]

[0118] This dynamic update process can adapt to changes in the network environment in real time, avoiding inefficiency or instability caused by fixed parameter settings, and improving the network's latency guarantee capability and overall performance stability in haptic services.

[0119] Step three involves estimating the effective bandwidth based on the token bucket model. This involves dynamically adjusting the token generation rate and bucket capacity to find suitable parameters, and then combining these with QoS requirements to estimate the effective capacity of the link. Unlike traditional static estimation methods based on average rate or instantaneous bandwidth, this invention fully considers the burstiness and heavy-tail distribution characteristics of traffic. It utilizes the token bucket model to shape and rate-control burst traffic, adjusting the token bucket parameters according to real-time traffic conditions to ensure they match actual traffic demand, thereby achieving dynamic estimation of the link's effective bandwidth. Specifically, this invention dynamically adjusts the token generation rate and bucket capacity by statistically analyzing the status of each link in the tactile service, balancing the token generation rate with actual traffic demand, and then combining this with the service's QoS requirements to estimate the effective bandwidth of the current link under latency QoS constraints. This not only improves the accuracy of effective bandwidth estimation but also reduces the impact of bursty and heavy-tail traffic on network performance, thereby improving the overall network latency stability and throughput performance.

[0120] Specifically include:

[0121] (1) A token bucket model-based effective bandwidth estimation method is proposed. This method considers the statistical characteristics of burst traffic and heavy-tailed distribution, and can more accurately estimate the actual bandwidth resources required by the network. The effective bandwidth... The calculation uses the following improved formula:

[0122]

[0123] in, It is an adjustment curve for the arrival process. , It is the token generation rate. Indicates time slot, Indicates the token bucket capacity;

[0124] Step 4: Dynamic adjustment method for link backoff factor, based on real-time statistics of link queue status and latency violations, as well as QoS parameters within the parameter update cycle. By combining effective bandwidth and effective capacity theories, the backoff factor and transmission strategy are dynamically adjusted to optimize channel occupancy and improve transmission stability. During operation, the backoff factor... Dynamic optimization is performed using gradient descent to balance latency and transmission efficiency under normal load; when queue congestion or increased latency violation probability is detected, the system automatically reduces... To extend the retreat time; while under light load conditions, it is directly... Set it to 0 to avoid unnecessary backoff delays.

[0125] Specifically include:

[0126] (1) Calculation of effective link capacity:

[0127] A two-state Markov chain model is established based on the service process. In the Markov chain model, state 1 represents the link. A transmission rate of 1 indicates a link with a state of 0. The transmission rate is 0; in this invention, the duration of the fallback process and the transmission duration of the link respectively follow the mean value of 0. Exponential distribution with mean 1; transition rate matrix for:

[0128]

[0129] According to avoidance factors and QoS parameters ,link Effective capacity is:

[0130]

[0131] in, Indicates link The average arrival rate;

[0132] (2) Parameters and Adaptive updates:

[0133] The parameters required to calculate the optimal backoff factor are obtained using the sub-gradient algorithm. and :

[0134]

[0135]

[0136] in, and In the sub-gradient algorithm, the first... The Lagrange multiplier in the next iteration, This is a variation of the backoff factor, designed to better solve the optimization problem, whereby... This transforms a non-convex optimization problem into a convex optimization problem. Indicates link Effective bandwidth , and This indicates the step size of the corresponding Lagrange multiplier, and is set. and To ensure the rapid convergence of the sub-gradient method;

[0137] (3) A novel backoff factor calculation method is proposed. Taking into account current network throughput, queue status, and QoS requirements, the optimal backoff factor is obtained through calculation, achieving the best balance between transmission efficiency and QoS guarantee. The backoff factor... The calculation is as follows:

[0138] First, calculate using the following formula. :

[0139]

[0140] in, , Indicates link In the The parameter represents the amount of data successfully transmitted within the backoff factor adjustment period;

[0141] Then through To obtain the optimal backoff factor;

[0142] (4) Backoff time calculation (used for subsequent QoS-CSMA / CA backoff):

[0143] In CSMA / CA mechanisms, to reduce channel access conflicts, nodes delay transmission by using backoff time when they detect a busy channel. Traditional CSMA / CA (such as IEEE 802.11) uses a binary exponential backoff mechanism. To improve network performance, researching the optimal backoff factor aims to optimize the distribution of backoff time, thereby reducing conflicts, increasing throughput, and lowering latency.

[0144] The retreat time is calculated as follows:

[0145]

[0146] in, Indicates the length of a time slot. It is a random variable that follows an exponential distribution, and its mean is related to the backslip factor. related;

[0147] The backoff time is used to initialize the backoff timer. During the backoff process, the backoff timer only decrements once if the channel remains idle for a given backoff time slot. The node only starts transmitting when the backoff timer reaches zero and the channel remains idle. The backoff time is calculated using an exponential distribution and... The mapping relationship. By... By converting the expected value of the backoff time, an adaptive match between the collision probability and network load is achieved, effectively reducing the probability of network collisions. This dynamic adjustment method can cope with sudden high loads in the network, reduce collisions and latency, and improve channel utilization.

[0148] Step 5 involves real-time monitoring and dynamic scheduling of link transmission status. A multi-dimensional network monitoring system is constructed to statistically analyze link queue status and latency violations in real time. When performance indicators exceed thresholds, parameter reconfiguration is automatically triggered. By monitoring network status in real time and dynamically adjusting flow control strategies, data transmission requests from high-load links are prioritized based on current link load conditions to improve overall network efficiency and ensure stable communication under different conditions.

[0149] Specifically include:

[0150] (1) An efficient queue state update method was designed. At the beginning of each time slot, the queue length of each link is updated in real time based on the number of newly arrived data packets and the number of successfully transmitted data packets. This method has low computational cost but can accurately reflect the current network congestion status. The queue state update formula for each time slot is as follows:

[0151]

[0152] in, This indicates the queue length at the end of the previous time slot. Indicates link In the time slot queue length, This indicates the amount of newly arrived data in the current time slot. This indicates the amount of data successfully transmitted and removed from the queue in the current time slot. Ensure that the queue length does not become negative;

[0153] (2) Delay violation criteria:

[0154] when hour, The value is increased by 1 if it is not increased, otherwise it remains unchanged. Indicates link Average arrival rate Indicates link The delay exceeds the binding Number of time slots;

[0155] The current queue length is compared with the theoretical maximum allowed queue length (calculated from the arrival rate and latency threshold) to quickly determine whether a latency violation has occurred.

[0156] The number of latency violations by the cumulative link within the current parameter adjustment period. Estimate the actual latency violation probability of the link, reflect the current latency violation situation of the link, provide feedback for subsequent parameter adjustments, and help dynamically optimize the transmission strategy;

[0157] (3) The system sorts each link queue in descending order in each time slot to determine the priority (the longer the queue length, the higher the priority). Then, it sequentially checks whether each link simultaneously meets the conditions of "backoff timer reset to zero and channel idle". If it does, transmission starts immediately; otherwise, it continues to count down and wait until the transmission conditions are met.

[0158] Dynamic scheduling priorities are sorted in descending order of queue length: The dynamic scheduling algorithm adopts a descending sorting strategy based on queue length. High-load links are scheduled first, thereby reducing the risk of queue overflow and ensuring fairness among links.

[0159] Step 5 involves real-time monitoring and dynamic sorting of the link queue status, combined with backoff time and queue priority sorting, to achieve effective scheduling decisions within the current cycle. It also provides real-time link status feedback for subsequent dynamic adjustment of the backoff factor, thereby improving overall transmission efficiency and adaptive performance.

[0160] Example 2

[0161] The parameter values ​​are set as shown in Table 1;

[0162] Table 1. Simulation Parameter Table

[0163]

[0164] When setting , , (unit: When ), the delay violation probability box plot of the present invention is as follows: Figure 2 As shown, the evolution of the measured average arrival and service rates of each link, as well as the average arrival and service rates of link 1, is as follows: Figure 3 As shown, the network latency performance under the same conditions is as follows: Figure 4 As shown. Figure 5 The average throughput variation trend of the present invention under different loads is shown, wherein the average arrival rate of each link is set to The average throughput in the network initially increases with increasing load, then slows down. Simulation results show that, facing bursty traffic and heavy-tailed characteristics in haptic services, this invention can ensure low latency and high throughput transmission performance by dynamically adjusting the backoff factor.

[0165] The preferred embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the scope of protection of the present invention is not limited to the specific details in the above embodiments. Within the scope of the technical concept of the present invention, any person skilled in the art can make equivalent substitutions or changes based on the technical solution and inventive concept of the present invention within the scope of the technology disclosed in the present invention. These simple modifications are all within the scope of protection of the present invention.

Claims

1. A QoS-CSMA / CA improvement method for haptic services, characterized in that, Includes the following steps: Step 1: Modeling the arrival process of tactile services: Pareto distribution is used to model burst traffic and heavy-tail characteristics in haptic services; initial QoS parameters are calculated based on link delay boundaries and delay violation probability requirements; a token bucket parameter initialization method based on the statistical characteristics of burst traffic and heavy-tail distribution is used to determine the bucket capacity by calculating the Pareto distribution mean and variance and combining it with network load. and token generation rate ; The parameter adjustment adopts a two-stage iterative strategy: the first stage finds the minimum by gradually reducing the bucket capacity. The second phase involves gradually reducing the token generation rate to find the minimum. ; Step 2: Dynamic QoS Awareness: Calculate the short-term latency violation probability requirement of the link during the backoff factor update cycle; dynamically update the QoS parameters during the backoff factor update cycle. Step 3: Estimating effective bandwidth based on the token bucket model; Step 4, Dynamic Adjustment of Link Backoff Factor: Calculate the effective link capacity; parameters and Adaptive update; considering current network throughput, queue status, and QoS requirements, the optimal backoff factor is obtained; Calculate the retreat time; Step 5: Real-time monitoring and dynamic scheduling of link transmission status: At the beginning of each time slot, the length of each link queue is updated in real time based on the number of newly arrived data packets and successfully transmitted data packets; Determine if delay violation criteria are met; sort each link queue in descending order of length in each time slot, and determine in sequence whether each link simultaneously meets the conditions of backoff timer being zero and channel being idle. If it does, transmit immediately; otherwise, continue counting down and waiting until the transmission conditions are met.

2. The QoS-CSMA / CA improvement method for haptic services according to claim 1, characterized in that, Step 1, modeling the arrival process of tactile services, specifically includes the following: Step 1.1: The Pareto distribution is used to accurately model the burst traffic and heavy-tail characteristics in the haptic service, where the probability density function of the Pareto distribution is: ； in, This represents the number of data packets arriving within a unit time slot, when the shape parameter... The mean of the Pareto distribution yes: ； According to shape parameters and average arrival rate of services Calculate scale parameters for: ； Random flow data is generated using the inverse transform method: ； in for Uniformly distributed random numbers Indicates link The number of data packets arriving per unit time slot; Step 1.2, based on the link Calculate the delay bound and delay violation probability requirement for the link. Initial QoS parameters : ； in Indicates link The time delay boundary, Indicates link Long-term delays violate probability requirements; Step 1.3, the token bucket parameter initialization method based on burst traffic and heavy-tailed distribution statistical characteristics, calculates the Pareto distribution mean. and variance and in combination with network load Determine the bucket capacity and token generation rate The initial bucket capacity and initial token generation rate for: ； ； in Indicates the adjustment coefficient; ； Step 1.4, adjusting parameters adopts a two-stage iterative strategy: The first stage involves gradually reducing the bucket capacity to find the minimum requirement that is met. Iteratively adjust bucket capacity The formula is as follows: ； in Indicates the number of iterations. The reduction factor representing the bucket capacity. Indicates the first The bucket capacity after each iteration is determined by stopping the iteration when the token bucket starts dropping packets, thus determining the minimum bucket capacity. The second stage involves gradually reducing the token generation rate to find the minimum rate that meets the requirements. Iteratively adjust the token generation rate The formula is as follows: ； in The reduction factor representing the token generation rate. Indicates the first The token generation rate after each iteration is determined by stopping the iteration when the token bucket starts dropping packets, thus determining the minimum token generation rate.

3. The QoS-CSMA / CA improvement method for haptic services according to claim 2, characterized in that, Step two, dynamic QoS awareness, specifically includes the following: Step 2.1, calculate the backoff factor update cycle. Midlink Short-term latency violates probability requirements : ； in This indicates the period during the backsliding factor update cycle. In the middle, link The delay exceeds the binding The number of time slots; Indicates the total number of time slots; Step 2.2, Dynamically update the backoff factor update cycle QoS parameters in : 。 4. The QoS-CSMA / CA improvement method for haptic services according to claim 3, characterized in that, Step 3: Effective bandwidth estimation based on the token bucket model, where the effective bandwidth... The calculation uses the following formula: ； in, It is an adjustment curve for the arrival process. , Indicates a time slot.

5. The QoS-CSMA / CA improvement method for haptic services according to claim 4, characterized in that, Step four, dynamic adjustment of the link backoff factor, specifically includes the following: Step 4.1, Calculation of effective link capacity: A two-state Markov chain model is established based on the service process. In the Markov chain model, state 1 represents the link. A transmission rate of 1 indicates a link with a state of 0. The transmission rate is 0; the duration of the fallback process and the transmission duration of the link both follow the mean value. Exponential distribution with mean 1; transition rate matrix for: ； According to avoidance factors and QoS parameters ,link Effective capacity is: ； in, Indicates link The average arrival rate; Step 4.2, Parameters and Adaptive updates: The parameters required to calculate the optimal backoff factor are obtained using the sub-gradient algorithm: ； ； in, and In the sub-gradient algorithm, the first... The Lagrange multiplier in the next iteration, It is a variation of the avoidance factor. This transforms a non-convex optimization problem into a convex optimization problem. Indicates link Effective bandwidth and This indicates the step size of the corresponding Lagrange multiplier, and is set. and To ensure the rapid convergence of the sub-gradient method; Step 4.3 comprehensively considers the current network throughput, queue status, and QoS requirements, and obtains the optimal backoff factor by solving for it, achieving the best balance between transmission efficiency and QoS guarantee. The backoff factor... The calculation is as follows: First, calculate using the following formula. : ； in, , Indicates link In the The amount of data successfully transmitted within a single backoff factor adjustment cycle; Then through To obtain the optimal backoff factor; Step 4.4, the retreat time is calculated as follows: ； in, Indicates the length of a time slot. It is a random variable that follows an exponential distribution, and its mean is related to the backslip factor. related.

6. The QoS-CSMA / CA improvement method for haptic services according to claim 5, characterized in that, Step 5 involves real-time monitoring and dynamic scheduling of the link transmission status; this includes the following: Step 5.1: At the beginning of each time slot, update the queue length of each link in real time based on the number of newly arrived data packets and the number of successfully transmitted data packets. The formula for updating the queue status for each time slot is: ； in, This indicates the queue length at the end of the previous time slot. Indicates link In the time slot queue length, This indicates the amount of data successfully transmitted and removed from the queue in the current time slot. Ensure that the queue length does not become negative; Step 5.2, Delay violation judgment criteria: when hour, The value is increased by 1 if it is not increased, otherwise it remains unchanged. Indicates link Average arrival rate Indicates link The delay exceeds the binding Number of time slots; Step 5.3: The system sorts each link queue in descending order according to its length in each time slot to determine the priority. Then, it sequentially checks whether each link simultaneously meets the two conditions of the backoff timer being zero and the channel being idle. If both conditions are met, transmission begins immediately; otherwise, the countdown continues until the transmission conditions are met.

7. The QoS-CSMA / CA improvement method for haptic services according to claim 6, characterized in that, In step 5.2, the number of delay violations of the link within the current parameter adjustment period is accumulated. This estimates the actual latency violation probability of the link, reflecting the current latency violation situation of the link and providing feedback for subsequent parameter adjustments.

Images