A dynamic load balancing transmission method based on synchronous adjustment

Abstract

The present invention relates to the technical field of dynamic load balancing processing and communication, and discloses a dynamic load balancing transmission method based on synchronous adjustment, comprising: obtaining a synchronization parameter set of each communication node in real time; calculating the initial load weight of each communication node through a dynamic feedback algorithm; introducing a priority coefficient, and dynamically correcting the initial load weight according to the real-time level of the transmission request; generating a final load weight through normalization processing so that the sum of the load weights of all nodes is 1, and generating a dynamic load distribution strategy based on the final load weight; according to the dynamic load distribution strategy, using a weighted polling algorithm, dynamically distributes the transmission request of a terminal device to each communication node according to the load weight ratio of each node, thereby completing dynamic load balancing of the transmission request. The present invention obtains the synchronization parameter set of each communication node in real time, accurately calculates and dynamically corrects the node load weight, thereby realizing an efficient and accurate dynamic load balancing distribution strategy.

Description

Technical Field

[0001] The present invention relates to the technical field of dynamic load balancing processing and communication, and in particular to a dynamic load balancing transmission method based on synchronous adjustment. Background Art

[0002] With the rapid development of communication technology and information industry, the scale and complexity of data transmission are increasing, and data traffic is characterized by high concurrency, burstiness and multi-path, which leads to the increasingly prominent problem of network load imbalance. Therefore, load balancing technology has gradually become one of the important research directions in the field of network transmission. Dynamic load balancing transmission technology, with its good flexibility and adaptability, is gradually becoming an important supporting technology in the modern network communication environment. At present, research on dynamic load balancing technology is mostly focused on how to improve the utilization efficiency of network resources, reduce network congestion and improve service quality to meet the growing business needs.

[0003] CN105516328A discloses a dynamic load balancing method for a distributed storage system. This technology collects line load information in real time through a monitoring server and dynamically determines the data transmission path based on the line load information. However, in an actual network environment, it is difficult to comprehensively consider real-time network quality changes and sudden load conditions, which easily leads to a lag in the load balancing effect.

[0004] CN106789642A discloses a dynamic load balancing method based on software-defined networking. This method sets a dynamic load balancing threshold and triggers a parallel scheduling strategy based on real-time load parameters. However, the load adjustment strategy of this method is relatively passive. When the network load changes sharply, its response speed is insufficient, resulting in low timeliness of load balancing, which makes it difficult to meet application scenarios with high real-time requirements.

[0005] At present, although the existing technology has improved the load balancing performance of network transmission, it still has many shortcomings, such as the inability to respond quickly and accurately to real-time changes in the network, and the lack of refined control strategies for dynamic adjustment mechanisms.

[0006] Therefore, the main problem faced by the existing technology is that in the scenario where network parameters change in real time, it is impossible to accurately and quickly achieve dynamic load balancing, resulting in some nodes being overloaded, which in turn affects the overall network performance.

[0007] In response to the shortcomings of the above-mentioned existing technologies, the present invention provides a dynamic load balancing transmission method based on synchronous adjustment. By obtaining the synchronization parameter set of each communication node in real time, the node load weight is accurately calculated and dynamically corrected, thereby realizing an efficient and accurate dynamic load balancing distribution strategy. Summary of the Invention

[0008] The purpose of this section is to summarize some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the abstract of the specification and the title of the invention of this application to avoid blurring the purpose of this section, the abstract of the specification and the title of the invention, and such simplifications or omissions cannot be used to limit the scope of the invention.

[0009] In view of the above existing problems, the present invention is proposed.

[0010] To solve the above technical problems, the present invention provides the following technical solutions: real-time acquisition of the synchronization parameter set of each communication node, wherein the synchronization parameter set includes the current network delay, bandwidth utilization, packet loss rate, and channel quality score;

[0011] Based on the synchronization parameter set, calculating the initial load weight of each communication node through a dynamic feedback algorithm;

[0012] Introducing a priority coefficient to dynamically modify the initial load weight according to the real-time level of the transmission request, wherein the higher the real-time level, the greater the priority coefficient;

[0013] Generate the final load weight through normalization so that the sum of the load weights of all nodes is 1, and generate a dynamic load distribution strategy based on the final load weight;

[0014] According to the dynamic load distribution strategy, a weighted round-robin algorithm is adopted to dynamically distribute the transmission requests of the terminal devices to each communication node according to the load weight ratio of each node, thereby completing the dynamic load balancing of the transmission requests.

[0015] As a preferred solution of the dynamic load balancing transmission method based on synchronization adjustment described in the present invention, the synchronization parameter set includes the current network delay, bandwidth utilization, packet loss rate and channel quality score, wherein:

[0016] Each communication node periodically sends a probe data packet and records the round-trip response time of the data packet to calculate the network delay;

[0017] Monitor the data throughput of each communication node in real time and calculate the bandwidth utilization rate based on the maximum available bandwidth of the node;

[0018] Count the ratio of the number of node data packets sent to the number of successfully received data packets per unit time to determine the real-time packet loss rate;

[0019] The signal-to-noise ratio and signal-to-interference ratio between each communication node and the terminal device are measured, and a weighted sum is performed based on the signal-to-noise ratio and the signal-to-interference ratio. The predicted value is output in combination with the multipath fading prediction model to obtain a real-time dynamic channel quality score.

[0020] As a preferred solution of the dynamic load balancing transmission method based on synchronous adjustment described in the present invention, the initial load weight is calculated based on the current load of each communication node, the historical load change trend and the channel quality score, wherein the historical load change trend is obtained by periodically collecting node load data and analyzing and calculating using a weighted moving average algorithm.

[0021] As a preferred solution of the dynamic load balancing transmission method based on synchronous adjustment described in the present invention, the calculating of the initial load weight includes:

[0022] Establish a load history database to periodically record the load conditions of each node;

[0023] Use weighted moving average algorithm to analyze historical load data and determine the load change trend of each node;

[0024] The initial load weight of the node is calculated by combining the real-time load of the node and the historical load trend.

[0025] As a preferred solution of the dynamic load balancing transmission method based on synchronous adjustment described in the present invention, the normalization process includes:

[0026] Superimpose the corrected weights of each node and calculate the total weight;

[0027] Based on the total weight, calculate the weight ratio of each node so that the sum of all adjusted node weights is equal to 1.

[0028] As a preferred solution of the dynamic load balancing transmission method based on synchronous adjustment described in the present invention, the weighted round-robin algorithm includes:

[0029] Establish a node polling queue and distribute transmission requests to corresponding nodes in accordance with the node weight ratio;

[0030] After each transmission request allocation is completed, the node polling queue is reordered according to the real-time updated node load weight.

[0031] As a preferred solution of the dynamic load balancing transmission method based on synchronous adjustment described in the present invention, it also includes:

[0032] Pre-configure the standby node resource pool and monitor the resource usage of the standby nodes in real time;

[0033] When the node load exceeds the set threshold and the standby node activation mechanism is triggered, the priority order of the standby node in the queue is dynamically adjusted based on the load weight calculated in real time;

[0034] When the load decreases, the priority ranking of the standby nodes is re-evaluated and updated dynamically.

[0035] As a preferred solution of the dynamic load balancing transmission method based on synchronous adjustment described in the present invention, the dynamic load distribution strategy includes a scheduling table generation rule, an adaptive weight smoothing rule, and a burst traffic fallback rule, wherein the scheduling table generation rule includes:

[0036] Collect the final load weights of all communication nodes at a preset update period T;

[0037] Set the amplification factor, multiply the final load weight of each communication node by the amplification factor and round up to get the corresponding number of slots ;

[0038] Repeat the node identification in the schedule table times to construct a polling sequence so that the number of times a node identifier appears in the schedule is proportional to its load weight;

[0039] The scheduling table is stored in a circular queue manner, and when a terminal transmission request is received, the node identifier is taken out in queue order, and after one scheduling is completed, the node identifier is moved to the end of the queue.

[0040] As a preferred solution of the dynamic load balancing transmission method based on synchronous adjustment described in the present invention, the adaptive weight smoothing rule includes:

[0041] When the final load weight change rate of any node in two adjacent update cycles is greater than the threshold θ, the weight of the node is corrected using the exponential smoothing algorithm;

[0042] Using the revised weights Replace the original weight update schedule to suppress scheduling jitter caused by drastic changes in weights.

[0043] As a preferred solution of the dynamic load balancing transmission method based on synchronous adjustment described in the present invention, the burst traffic fallback rule includes:

[0044] Monitor the queue length of each communication node in real time. When the queue length exceeds twice the historical average queue length of the node, multiply the allocation weight of the node by the attenuation factor γ, where 0<γ<1;

[0045] Redistribute the attenuated weight proportionally to online nodes and standby nodes with lower loads;

[0046] When the queue length of a node drops below 2 times the historical average queue length, the original allocation weight of the node is restored and the scheduling table is updated synchronously.

[0047] Beneficial effects of the present invention:

[0048] 1. By acquiring the synchronized parameter sets of each communication node in real time, it achieves instantaneous quantitative perception of the network operating status, providing multi-dimensional, refined basic data for subsequent load assessment. This avoids evaluation distortion caused by a single performance indicator and captures network fluctuations within millisecond-level time resolution, enabling scheduling logic to be based on complete and accurate real-time information. This helps reduce monitoring latency and improve the real-time effectiveness of load decision-making.

[0049] 2. By using a dynamic feedback algorithm based on a synchronized parameter set to calculate the initial load weight of each communication node, an adaptive quantitative scoring system is implemented that comprehensively considers node processing capabilities, link quality, and historical load trends. This system converts dynamic network characteristics into measurable and comparable weight values, providing a computational basis for allocation strategies. With minimal computational complexity, a load benchmark that can be iterated in real time is obtained, avoiding node saturation or resource idleness caused by static weights, thereby improving resource utilization across the entire network.

[0050] 3. By introducing a priority coefficient to dynamically adjust the initial weight according to the real-time level of the transmission request, service sensitivity is coupled with network resource allocation logic. This ensures that high-real-time services still have sufficient bandwidth and low-latency paths in congested environments. This ensures the quality of key services without sacrificing overall throughput, improves the differentiated service level experience, and reduces retransmissions caused by congestion.

[0051] 4. By forming a dynamic load distribution strategy, the mathematical consistency of the weight vector and the closed-loop self-verification of the scheduling strategy are achieved. This allows the impact of newly added or offline nodes on the global allocation ratio to be immediately and smoothly absorbed, avoiding imbalances in the overall allocation ratio caused by local weight drift. While keeping the load fluctuations of each node within a controllable threshold, it also ensures system stability and algorithm convergence.

[0052] 5. By adopting a weighted round-robin algorithm to dynamically allocate terminal transmission requests to each node according to the final weight ratio, efficient execution of the mapping of scheduling strategies to specific message flows is achieved. Combining the fairness of polling with the differentiation of weights, it not only avoids long-term node starvation, but also can tilt high-performance links on demand, dynamically balance the load of each node in multi-business concurrent scenarios, reduce the probability of single-point overload, and maintain business continuity and overall throughput steady state under link jitter or burst traffic. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort. Among them:

[0054] Figure 1This is a flow chart of a dynamic load balancing transmission method based on synchronous adjustment shown in the present invention. DETAILED DESCRIPTION

[0055] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, but not all of the embodiments.

[0056] Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without making any creative work should fall within the scope of protection of the present invention.

[0057] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0058] According to an embodiment of the present invention, Figure 1 The flowchart shown is a dynamic load balancing transmission method based on synchronous adjustment, which specifically includes the following steps:

[0059] S1. Acquire the synchronization parameter set of each communication node in real time. The synchronization parameter set includes the current network delay, bandwidth utilization, packet loss rate, and channel quality score;

[0060] S2. Based on the synchronization parameter set, the initial load weight of each communication node is calculated through a dynamic feedback algorithm;

[0061] S3. Introducing a priority coefficient to dynamically modify the initial load weight based on the real-time level of the transmission request. The higher the real-time level, the greater the priority coefficient.

[0062] S4. Generate a final load weight through normalization so that the sum of the load weights of all nodes is 1, and generate a dynamic load distribution strategy based on the final load weight;

[0063] S5. According to the dynamic load distribution strategy, a weighted round-robin algorithm is used to dynamically distribute the transmission requests of the terminal devices to each communication node according to the load weight ratio of each node, thereby completing the dynamic load balancing of the transmission requests.

[0064] It should be noted that the method of the present invention ensures the high sensitivity of scheduling decisions to the instantaneous state of the network on the one hand, and embeds business real-time differentiation and algorithm self-calibration mechanisms in the entire process on the other hand, thereby realizing continuous, explainable, low-latency dynamic load balancing of transmission requests in complex heterogeneous network environments, providing large-scale distributed systems with a transmission solution that takes into account high availability, high throughput and high QoS (quality of service).

[0065] The following describes in more detail the implementation process and / or effects of certain embodiments of the present invention in conjunction with some preferred or optional examples of the present invention.

[0066] Get the synchronization parameter set:

[0067] In a fixed monitoring period Internal node To the rest of the nodes Send K UDP Echo probe packets;

[0068] The round trip time of the kth probe packet is recorded as ;

[0069] The one-way average delay is calculated as:

[0070]

[0071] in, is the i-th communication node, K is the number of detection packets for each pair of nodes (such as 3), is the round trip time between nodes i and j for the kth detection packet, N is the total number of nodes, is the one-way average network delay of node i;

[0072] like Exceeds the average of the previous period , then replace it with the mean value of the previous cycle to suppress instantaneous jitter;

[0073] Then in the sliding window Read the network card counter internally to get the node's real-time throughput ;

[0074] Normalize the bandwidth utilization:

[0075]

[0076] in, is the maximum available bandwidth of the node obtained through link negotiation, is the actual throughput of node i in the window, is the bandwidth utilization of node i;

[0077] like , then define And record abnormal logs to prevent weight imbalance;

[0078] Furthermore, the number of packets sent in the same window W is counted separately. Number of successfully received packets ;

[0079] The packet loss rate is:

[0080]

[0081] in, is the real-time packet loss rate of node i;

[0082] when When the node historical mean Backfill to avoid division by zero errors;

[0083] Synchronous acquisition of signal-to-noise ratio between nodes and terminals Signal-to-interference ratio ;

[0084] Input the real-time measurement values ​​into the prediction model based on Rayleigh multipath fading , predicting the future Channel availability, and obtain weighted normalized score:

[0085]

[0086] in, 、 、 is the weight coefficient of channel quality score, , 、 are the system reference signal-to-noise ratio and signal-to-interference ratio, respectively. Score the dynamic channel quality of node i;

[0087] like , then the marked node is a weak channel;

[0088] Will 、 、 、 Write 32-bit floating-point structures in sequence, with a 64-bit UNIX nanosecond timestamp attached to the header;

[0089] Define the synchronization parameter set as

[0090] in, is the synchronization parameter vector of node i;

[0091] Use gRPC bidirectional stream to push batches to the scheduling controller, which maintains the length The ring cache ensures that the most recent valid data can be rolled back when a node is temporarily offline.

[0092] All nodes control the clock error within , ensuring that the sampling windows are aligned.

[0093] For example, taking a four-node cluster N = {n1,n2,n3,n4} as an example, during a monitoring period, the following values ​​are obtained:

[0094] Table 1. Statistics of four-node synchronization parameters

[0095] node τ / ms U (%) L (%) Q <![CDATA[n1]]> 18.7 58.2 1.9 0.87 <![CDATA[n2]]> 27.4 42.5 0.8 0.91 <![CDATA[n3]]> 33.6 71.3 3.4 0.79 <![CDATA[n4]]> 21.1 65.5 2.1 0.82

[0096] Scheduling controller immediate storage vector ={18.7 ms, 0.582, 0.019, 0.87}, and triggers the dynamic feedback algorithm of step S2.

[0097] It should be noted that the embodiment of the present invention provides millisecond-level, complete and noise-resistant network status quantification data for subsequent weight calculations through the implementation of the above-mentioned process steps. Compared with the existing technology that only relies on a single indicator, it improves the adaptive ability of the dynamic load balancing strategy to transient network fluctuations.

[0098] Calculate the initial load weight and make dynamic corrections:

[0099] In an optional embodiment, the scheduling controller establishes a load history database for each communication node in the network, periodically records the node's CPU usage, memory usage, I / O queue length, and link bandwidth usage, and saves the data for the most recent monitoring cycles at a preset depth;

[0100] The historical database is processed, and the load records in the last m monitoring cycles are assigned decreasing weights in order of recent to distant. The weighted moving average is calculated to represent the change trend of node load over time.

[0101] Based on the system upper limit of each indicator, the real-time load vector in the current cycle is normalized so that the results of each dimension fall into the interval [0, 1], facilitating horizontal comparison of data from different resource dimensions.

[0102] Then, the normalized real-time load, weighted moving average, and node channel quality score are linearly superimposed according to the weight ratio, where:

[0103] Normalized real-time load reflects the instantaneous resource usage of the node;

[0104] The load change trend reflects the growth or decay inertia of the node load;

[0105] The node channel quality score reflects the link level availability;

[0106] If the comprehensive load score value is larger, it means that the overall node load is heavier and the link status is worse;

[0107] The initial weight vector is obtained by taking the reciprocal of the comprehensive load score to avoid high-load nodes from receiving too high a weight. The reciprocal value of each node is then normalized proportionally so that the sum of the initial load weights of all nodes is always 1.

[0108] For example, the initial weight generation formula is:

[0109] Initial weight generation

[0110]

[0111] in, For the comprehensive load rating, Countdown for the rating, is the initial load weight;

[0112] It should be further explained that after each monitoring window ends, the actual traffic ratio received by each node in the previous window is counted and compared with its initial load weight. If there is a positive or negative deviation between the two, the initial weight is fine-tuned with a limited amplitude according to the preset convergence coefficient to eliminate the accumulated error and maintain system convergence.

[0113] Each transmission request entering the scheduling queue is marked with a real-time level, ranging from non-real-time to extremely real-time. The controller calculates the average real-time level of all requests in the same node queue and uses this to obtain a priority coefficient from a table lookup. The higher the real-time level, the larger the corresponding coefficient, ensuring scheduling priority for delay-sensitive services.

[0114] Multiply the feedback-corrected node weight by the corresponding priority coefficient to obtain the weight value of the integrated load link service demand three-dimensional information;

[0115] To prevent a single node from experiencing abnormally low weight and starvation, the controller sets a safety lower limit for the corrected weights of all nodes. If a node's weight falls below this lower limit, it is directly raised to the safety threshold, and the difference in weight is proportionally deducted from the higher-weight node. The weight vector is then normalized again to ensure that the sum of the elements of the final weight vector remains 1.

[0116] The processed weight vector is sent to the scheduling execution module in real time for subsequent steps to generate a dynamic load distribution strategy and drive the weighted round-robin algorithm to complete data flow scheduling.

[0117] As an example, set a real-time level for each transmission request , where 1 is non-real-time and 5 is extremely real-time;

[0118] node The average real-time performance level in the current scheduling window is:

[0119]

[0120] in, is the load history trend vector, is the real-time level of the k-th request;

[0121] Corresponding coefficient ;

[0122] The weight after priority weighting correction is ;

[0123] in, is the feedback correction weight, is the node priority coefficient;

[0124] Output the final weight vector:

[0125]

[0126] And immediately provide step S4 with the generation of a dynamic load distribution strategy.

[0127] Preferably, this step ensures that the scheduling system can not only quickly absorb transient changes in the network but also differentiatedly guarantee high real-time services through a two-level weight generation mechanism coupled with four factors of historical trend real-time indicators link quality priority, thereby improving the load balancing accuracy and system throughput stability in large-scale distributed environments.

[0128] Generate dynamic load distribution strategy to complete dynamic load balancing of transmission requests:

[0129] After receiving the modified weight vector from the above step, the scheduling controller first accumulates the weights of all nodes to obtain the total weight of the current window;

[0130] The controller uses the total weight as the divisor and scales the modified weight of each node proportionally so that the sum of all node weights after scaling is equal to 1. Through this normalization process, the weight vectors between different scheduling cycles are ensured to remain consistent, which is convenient for direct reference by subsequent scheduling algorithms.

[0131] In an optional embodiment, generating the schedule includes:

[0132] The final normalized weights are read in batches at a preset update period T (e.g., 1s);

[0133] To improve the polling granularity, set an amplification factor. Multiply the node weight by the amplification factor and round it up to get the number of slots for the node in the scheduling table.

[0134] The node identifiers are repeatedly written into the scheduling table in order of the node weights, so that the number of times the identifiers appear is proportional to the weights;

[0135] The scheduling table is stored in the form of a circular queue to facilitate free reading and writing of locks in high-concurrency scenarios.

[0136] Furthermore, the controller compares the relative change rate of each node weight in two adjacent update cycles. When the change rate exceeds a threshold θ (such as 20%), it automatically calls the exponential smoothing algorithm to soften the node weight and then replaces the original entry with the corrected value to suppress scheduling jitter caused by drastic fluctuations in weight.

[0137] In an optional implementation, a burst traffic fallback mechanism is provided to implement real-time updating of the scheduling table, wherein:

[0138] Monitor the message queue length of each node;

[0139] When the queue length exceeds twice the node's historical average, a rollback is triggered;

[0140] Multiply the overloaded node weight by the decay factor;

[0141] Distribute the attenuated weights to online idle nodes and standby nodes in proportion;

[0142] When the queue length falls below the threshold, the original weight of the node is restored and the scheduling table is updated in real time.

[0143] In an optional implementation, the scheduling execution module builds a node polling queue according to the latest scheduling table order, and allocates the nodes strictly according to the number of slots occupied by the node weight during the scheduling traversal, ensuring that the traffic distribution ratio in the long period is highly consistent with the node weight;

[0144] Each time a terminal transmission request is received, the node identifier is sequentially taken from the head of the circular queue and the request is forwarded to the corresponding node;

[0145] After the request is distributed, the node ID is moved to the end of the queue to implement the "retrieval" closed loop;

[0146] When the new weight vector arrives, the scheduling execution module immediately reconstructs the polling queue. At the same time, after each distribution is completed, the current window weight is fine-tuned according to the slight changes in the node's real-time load (such as CPU or queue length) to ensure continuous smoothness in the short term.

[0147] Exemplarily, the management of the standby node resource pool includes:

[0148] Pre-maintain several communication nodes in standby state and continuously monitor their resource usage;

[0149] When the online node load exceeds the set threshold (e.g. CPU 80%), the scheduling controller automatically activates the standby node;

[0150] Dynamically insert the standby node into the polling queue based on its real-time weight to quickly distribute traffic;

[0151] When the load of the hot node drops below the threshold, the controller re-evaluates the importance of the backup node and lowers its priority according to the load-performance-cost combination index to achieve resource elastic management.

[0152] It should be noted that the actual allocation ratio generated by weighted round-robin scheduling will be fed back to step S3 at the end of each monitoring cycle, forming a closed-loop control chain of "weight allocation feedback and re-correction". This link automatically stabilizes within the millisecond time scale, ensuring that both system throughput and business real-time performance are met during long-term operation.

[0153] As an example, assume that a communication network contains three nodes: A, B, and C. During a monitoring period, the revised weights of the nodes are:

[0154] Node A: 0.30, Node B: 0.50, Node C: 0.20;

[0155] After normalization, the final weights of each node are A: 0.30, B: 0.50, C: 0.20;

[0156] When the amplification factor is set to 100, the number of slots is obtained, where 30 are for A nodes, 50 are for B nodes, and 20 are for C nodes;

[0157] Construct a dynamic scheduling table with a length of 100 slots, in which node B appears 50 times, node A 30 times, and node C 20 times;

[0158] When each transmission request arrives, it is distributed in sequence according to the polling sequence to achieve refined load balancing;

[0159] At a certain moment, when the queue length of node B reaches three times the historical average, the sudden backoff rule is automatically triggered, the weight of node B is reduced by half, and the attenuation amount is redistributed to nodes A and C. The updated scheduling table takes effect immediately, thereby effectively alleviating instantaneous congestion.

[0160] Preferably, through the example illustration of the above steps, the present invention realizes real-time and accurate monitoring of load, dynamic and rapid adjustment, and efficient and stable traffic distribution during network operation, thereby improving the utilization rate of network transmission resources and the accuracy and timeliness of load balancing.

[0161] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A dynamic load balancing transmission method based on synchronous adjustment, characterized in that: include: Acquire the synchronization parameter set of each communication node in real time, including the current network delay, bandwidth utilization, packet loss rate, and channel quality score; Based on the synchronization parameter set, calculating the initial load weight of each communication node through a dynamic feedback algorithm; Calculate the initial load weight, including: Establish a load history database to periodically record the load conditions of each node; Use weighted moving average algorithm to analyze historical load data and determine the load change trend of each node; Calculate the initial load weight of the node based on the node's real-time load and historical load change trends; Introducing a priority coefficient to dynamically modify the initial load weight according to the real-time level of the transmission request, wherein the higher the real-time level, the greater the priority coefficient; Generate the final load weight through normalization so that the sum of the load weights of all nodes is 1, and generate a dynamic load distribution strategy based on the final load weight; According to the dynamic load distribution strategy, a weighted round-robin algorithm is adopted to dynamically distribute the transmission requests of the terminal devices to each communication node according to the load weight ratio of each node, thereby completing the dynamic load balancing of the transmission requests.

2. The dynamic load balancing transmission method based on synchronous adjustment according to claim 1, characterized in that: The synchronization parameter set includes the current network delay, bandwidth utilization, packet loss rate and channel quality score, where: Each communication node periodically sends a probe data packet and records the round-trip response time of the data packet to calculate the network delay; Monitor the data throughput of each communication node in real time and calculate the bandwidth utilization rate based on the maximum available bandwidth of the node; Count the ratio of the number of node data packets sent to the number of successfully received data packets per unit time to determine the real-time packet loss rate; The signal-to-noise ratio and signal-to-interference ratio between each communication node and the terminal device are measured, and a weighted sum is performed based on the signal-to-noise ratio and the signal-to-interference ratio. The predicted value is output in combination with the multipath fading prediction model to obtain a real-time dynamic channel quality score.

3. The dynamic load balancing transmission method based on synchronous adjustment according to claim 2, characterized in that: The initial load weight is calculated based on the current load of each communication node, the historical load change trend and the channel quality score, wherein the historical load change trend is obtained by periodically collecting node load data and analyzing and calculating using a weighted moving average algorithm.

4. The dynamic load balancing transmission method based on synchronous adjustment according to claim 1, characterized in that: The normalization process includes: Superimpose the corrected weights of each node and calculate the total weight; Based on the total weight, calculate the weight ratio of each node so that the sum of all adjusted node weights is equal to 1.

5. The dynamic load balancing transmission method based on synchronous adjustment according to claim 1, characterized in that: The weighted round-robin algorithm comprises: Establish a node polling queue and distribute transmission requests to corresponding nodes in accordance with the node weight ratio; After each transmission request allocation is completed, the node polling queue is reordered according to the real-time updated node load weight.

6. The dynamic load balancing transmission method based on synchronous adjustment according to claim 5, characterized in that: Also includes: Pre-configure the standby node resource pool and monitor the resource usage of the standby nodes in real time; When the node load exceeds the set threshold and the standby node activation mechanism is triggered, the priority order of the standby node in the queue is dynamically adjusted based on the load weight calculated in real time; When the load decreases, the priority ranking of the standby nodes is re-evaluated and updated dynamically.

7. The dynamic load balancing transmission method based on synchronous adjustment according to claim 1, characterized in that: The dynamic load distribution strategy includes a scheduling table generation rule, an adaptive weight smoothing rule, and a burst traffic fallback rule, wherein the scheduling table generation rule includes: Collect the final load weights of all communication nodes at a preset update period T; Set the amplification factor, multiply the final load weight of each communication node by the amplification factor and round up to get the corresponding number of slots ; Repeat the node identification in the schedule table times to construct a polling sequence so that the number of times a node identifier appears in the schedule is proportional to its load weight; The scheduling table is stored in a circular queue manner, and when a terminal transmission request is received, the node identifier is taken out in queue order, and after one scheduling is completed, the node identifier is moved to the end of the queue.

8. The dynamic load balancing transmission method based on synchronous adjustment according to claim 7, characterized in that: The adaptive weight smoothing rule includes: When the final load weight change rate of any node in two adjacent update cycles is greater than the threshold θ, the weight of the node is corrected using the exponential smoothing algorithm; Using the revised weights Replace the original weight update schedule to suppress scheduling jitter caused by drastic changes in weights.

9. The dynamic load balancing transmission method based on synchronous adjustment according to claim 7, characterized in that: The burst traffic fallback rules include: Monitor the queue length of each communication node in real time. When the queue length exceeds twice the node's historical average queue length, multiply the node's allocation weight by a decay factor γ, where 0 < γ < 1. Redistribute the attenuated weight proportionally to online nodes and standby nodes with lower loads; When the queue length of a node drops below 2 times the historical average queue length, the original allocation weight of the node is restored and the scheduling table is updated synchronously.

Images