Method for realizing data forwarding of multiple devices at remote station end

Abstract

The invention relates to the technical field of communication networks, in particular to a method for realizing data forwarding of multiple devices at a remote station end, which comprises the following steps: a data sending end classifies data packets and dynamically allocates bandwidth according to a classification result; the optical line terminal records source address information and forwarding rules of the classified data packets according to the classification result; and the data receiving end selects an optimal transmission path for the classified data packets based on the source address information and the forwarding rule through an improved Dijkstra algorithm. According to the invention, the data sending end classifies the data packets, dynamically distributes the bandwidth, and selects the optimal transmission path by improving the Dijkstra algorithm, so that the efficiency and reliability of data transmission are improved.

Description

Technical Field

[0001] The present application relates to the field of communication network technology, and in particular to a method for implementing data forwarding for multiple devices at a remote station. Background Art

[0002] In modern communication networks, efficient and reliable data transmission is the key to ensuring the normal operation of various businesses. With the rapid development of the Internet of Things, big data and cloud computing, the requirements for data transmission speed and quality are getting higher and higher, especially in the environment of multiple devices at the remote station, the efficiency and reliability of data transmission are particularly important.

[0003] At present, optical fiber communication networks and power microgrid communications have been widely used. Optical fiber communication networks have become the mainstream choice of modern communication networks due to their high bandwidth, low latency and high reliability. Power microgrid communications play an important role in smart grids, distributed energy management and automatic control. However, existing communication technologies still face many challenges in handling large-scale, multi-device data transmission. For example, traditional data forwarding methods are prone to network congestion and packet loss under high load conditions, affecting the efficiency and reliability of data transmission.

[0004] Traditional data forwarding methods often rely on fixed bandwidth allocation strategies and static path selection, and cannot dynamically adapt to changes in network topology and real-time network load conditions. Packet classification and bandwidth allocation are not accurate enough, and network congestion assessment and path selection lack intelligent means. In addition, when processing high-priority packets, existing methods are difficult to ensure that they can still obtain sufficient bandwidth when the network is congested, resulting in a decrease in the transmission quality of key services. These problems seriously affect the overall performance of the network and user experience. Summary of the invention

[0005] In view of this, it is necessary to provide a method for realizing data forwarding of multiple devices at a remote station end, so as to solve the problems in the prior art such as inaccurate bandwidth allocation, network congestion assessment and path selection, which lead to the degradation of transmission quality of key services and seriously affect the overall performance of the network and user experience.

[0006] A method for implementing data forwarding of multiple devices at a remote station, characterized by comprising:

[0007] The data transmitter classifies the data packets and dynamically allocates bandwidth based on the classification results;

[0008] The optical line terminal records the source address information and forwarding rules of the classified data packets according to the classification result;

[0009] The data receiving end selects the best transmission path for the classified data packet based on the source address information and the forwarding rule by improving the Dijkstra algorithm.

[0010] Preferably, the data transmitting end classifies the data packets and dynamically allocates bandwidth according to the classification results, including:

[0011] classifying the data packets using a Bayesian classifier;

[0012] Conducting credit evaluation on the classified data packets;

[0013] Based on the credit evaluation results, the optical network unit allocates the minimum guaranteed bandwidth to the classified data packets;

[0014] Use the weighted fair queue model to dynamically allocate the remaining bandwidth.

[0015] Preferably, the minimum guaranteed bandwidth satisfies:

[0016] B min =(T cycle -T g ×N)×R×W k / 8

[0017] Among them, T cycle Indicates the time it takes for all optical network units to poll once, T g represents the protection time, N represents the number of optical network units, R represents the transmission speed, and W k The weight for allocating bandwidth.

[0018] Preferably, the use of a weighted fair queue model to dynamically allocate bandwidth to the remaining bandwidth includes:

[0019] Calculate the weight settings of the weighted fair queue model to satisfy:

[0020]

[0021] Among them, B i is the bandwidth of the i-th priority request, B min is the minimum guaranteed bandwidth.

[0022] Preferably, the optical line terminal records the source address information and forwarding rules of the classified data packets according to the classification result, including:

[0023] The optical line terminal receives the data frame of the classified data packet and records the source MAC address, port number and logical circuit identifier of the data frame;

[0024] The optical line terminal uses the MAC address learning process to associate the source MAC address with the corresponding physical port and store it in the MAC address linked list;

[0025] The optical line terminal establishes and records forwarding rules according to the classification result and the MAC address chain list.

[0026] Preferably, the data receiving end selects the best transmission path for the classified data packet based on the source address information and the forwarding rule by using an improved Dijkstra algorithm, including:

[0027] Set the initial distance and predecessor nodes of all nodes;

[0028] Select the node with the shortest distance as the current node, and update the shortest path estimate of the neighboring nodes and the predecessor node;

[0029] Compare the shortest path estimate of each node with the known path length to determine whether the path length and predecessor node need to be updated;

[0030] According to the comparison result, the shortest path estimation and the predecessor node of the node are determined as the optimal transmission path.

[0031] Preferably, the classified data packets are subjected to credit evaluation, and then the following steps are further included:

[0032] Calculate the credit of each link;

[0033] Prioritize the transmission of data packets on links with high credit.

[0034] Preferably, the classified data packets are subjected to credit evaluation, including:

[0035] Credit rating is assessed based on the classified data packet type, historical behavior, and priority.

[0036] Preferably, obtaining the improved Dijkstra model includes:

[0037] Initialization module, used to set the initial conditions of the improved Dijkstra model algorithm;

[0038] An iteration module is used to determine the node with the shortest distance as the current node, compare the shortest path estimates of the neighboring nodes of the current node, and update the shortest path estimates of the neighboring nodes and the predecessor node according to the comparison results;

[0039] a comparison module for comparing the shortest path estimate of each node with the known path length;

[0040] The node updating module is used to update the shortest path estimation and predecessor node of the node according to the result of the comparison module.

[0041] Preferably, the data transmitting end classifies the data packets and dynamically allocates bandwidth according to the classification results, further comprising:

[0042] The data sender uses a machine learning model to predict network traffic and adjusts the bandwidth allocation strategy based on the prediction results;

[0043] The data sender reserves bandwidth based on the network traffic prediction results;

[0044] The data sending end monitors the network traffic in real time and dynamically adjusts the reserved bandwidth.

[0045] Compared with the prior art, the beneficial effects of the present application are: the data transmission end classifies the data packets and dynamically allocates bandwidth, and the optimal transmission path is selected by improving the Dijkstra algorithm, thereby improving the efficiency and reliability of data transmission. Through the Bayesian classifier and credit evaluation, it is ensured that high-priority data packets can still obtain sufficient bandwidth when the network is congested. The optical line terminal records the source address information and establishes forwarding rules to ensure accurate forwarding of data packets. The improved Dijkstra algorithm selects the best transmission path to reduce network delays and congestion. These measures not only optimize the utilization of network resources, but also improve the user experience and the overall performance of the network. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a flow chart of a method for implementing data forwarding of multiple devices at a remote station provided in an embodiment of the present application.

[0047] Figure 2 It is a schematic diagram of the REPORT / GATE mechanism and waiting time in the method for realizing high-speed forwarding of multiple devices at a remote station provided in an embodiment of the present application.

[0048] Figure 3 It is a flow chart of a bandwidth allocation model in a method for implementing data forwarding for multiple devices at a remote station provided in an embodiment of the present application.

[0049] Figure 4 It is a schematic diagram of the structure of an optical splitter in a method for implementing data forwarding of multiple devices at a remote station provided in an embodiment of the present application.

[0050] Figure 5 The application embodiment provides a Dijkstra algorithm flow chart for implementing a high-speed forwarding method for multiple devices at a remote station. DETAILED DESCRIPTION

[0051] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0052] See also Figure 1-Figure 5 , Figure 1 is a flow chart of a method for implementing data forwarding of multiple devices at a remote station provided by an embodiment of the present application, Figure 2 It is a schematic diagram of the REPORT / GATE mechanism and waiting time in the method for realizing high-speed forwarding of multiple devices at a remote station provided in an embodiment of the present application; Figure 3 It is a flow chart of a bandwidth allocation model in a method for implementing data forwarding of multiple devices at a remote station provided in an embodiment of the present application; Figure 4 This is a schematic diagram of the structure of the optical splitter in the method for implementing data forwarding of multiple devices at a remote station provided in an embodiment of the present application. Figure 5 The application embodiment provides a Dijkstra algorithm flow chart for implementing a high-speed forwarding method for multiple devices at a remote station.

[0053] The method for implementing data forwarding of multiple devices at a remote station includes the following steps:

[0054] A method for implementing data forwarding of multiple devices at a remote station includes:

[0055] S101: The data transmitting end classifies the data packets and dynamically allocates bandwidth according to the classification results.

[0056] At the data sending end, the data packets are classified by the Bayesian classifier, and the credit is used to measure the network congestion. Especially when the network is congested, the Optical Network Unit (ONU) optimizes the remaining bandwidth through the weighted fair queue model after meeting the minimum guaranteed bandwidth requirements, thereby realizing dynamic bandwidth allocation.

[0057] Specifically, it may include:

[0058] Packet classification: At the data transmission end, the data packets are first classified by the Bayesian classifier. This process usually includes the following steps: Feature extraction: Extract features from the data packet, such as source address, destination address, packet size, timestamp, etc. When training the Bayesian classifier, use historical data to train the Bayesian classifier to enable it to recognize different types of data packets (for example, ordinary traffic, video traffic, voice traffic, etc.). Classification decision: Input the newly arrived data packet into the Bayesian classifier and determine the type of the data packet based on the trained model.

[0059] Furthermore, the data can be classified through a naive Bayes classifier with an AXI interface, including content type, text data, multimedia data, and control data; control data is used for control information for communication between network devices, wherein the control data includes device control instructions, protection control information, scheduling control data, and network optimization data; the scheduling control data includes historical data and real-time information.

[0060] Further, the Bayesian classifier includes a DSP data processor and an input interface, an output interface, an A / D conversion module, a D / A conversion module, an accelerator, a weight setting module, an attribute setting module and a data evaluation module connected to the Bayesian classifier including the DSP data processor. In the above embodiment, the input interface is where the Bayesian classifier receives data. It is a sensor, a network interface or other data source, which is used to convert raw data into a form for DSP processing. The output interface is where the Bayesian classifier sends processed data or classification results to other systems or devices. This includes a display device, a storage device or a control unit. The A / D conversion module converts analog signals (such as sound or images) into digital signals for DSP processing and analysis. The D / A conversion module converts digital signals back to analog signals and outputs the processed data to an analog device or system if necessary. An accelerator is a hardware or software component used to increase the speed of data processing. In the Bayesian classifier, this is a dedicated mathematical operation accelerator for quickly calculating Bayesian probabilities. The weight setting module is responsible for configuring the weight parameters in the Bayesian classifier. In Bayesian classification, weights are usually associated with the importance of features and are used to adjust classification decisions. The property setting module is used to define and configure data attributes such as feature name, data type, and value range. This is essential for training and tuning Bayesian classifiers. The data evaluation module is responsible for evaluating the characteristics of the input data and using Bayes' theorem to calculate the probability of each class. It also includes error rate evaluation and model validation functions.

[0061] The OLT obtains bandwidth demand information, traffic container status, priority, service level identification, and error and status indications by receiving the DBRu report carried by the ONU in the upstream frame. After the OLT receives the DBRu report, if the queue waiting frame data is continuously high and the bandwidth is granted, it indicates that the network bandwidth is insufficient or the traffic needs to be optimized. When network congestion occurs, the network administrator grants bandwidth according to the current network load and the preset bandwidth allocation strategy to dynamically adjust the bandwidth. The hardware that can be implemented may include: Bayesian classifier: used to solve the problem of data classification. Given data and related probability information, the Bayesian classifier calculates the left and right classification decisions through the Bayesian theorem to minimize the classification error. Digital signal processor (DSP): For statistical and probability calculation problems that require complex calculations, DSP is needed to accelerate processing. The transmission data statistics of data under high load are used to calculate the average bandwidth. Counter: used to record frame data.

[0062] Credit measurement: After classifying the data packets, the system will measure their credit according to the type of data packets or other factors. Credit is based on the following factors: Data packet type: Different types of data packets have different requirements for network performance. Historical behavior: The credit is evaluated based on the historical behavior of the data packet (such as delay, packet loss rate, etc.). Priority: Credit is assigned based on the priority of the data packet. The credit measurement formula is as follows:

[0063] C=μ(B g -B×n),

[0064] Among them, C represents credit, which reflects the eligibility of a data packet or data flow to obtain bandwidth resources. C is a constant and a variable proportional to the requested bandwidth. It is a proportional factor, a constant and a variable proportional to the requested bandwidth. Its function is to adjust the weight in the credit calculation to make bandwidth allocation more flexible. g represents granted bandwidth, which refers to the amount of bandwidth allocated by the network for a specific data packet or data stream. B represents average bandwidth, which refers to the average amount of bandwidth available on the network over a period of time. This value is used to measure the overall load of the network. n represents the number of frames waiting in the queue, which refers to the number of frames waiting to be transmitted in the queue. This value is used to measure the degree of network congestion. μ represents the scaling factor, which is an adjustment factor used to adjust the credit based on the queue length. When the queue length increases, the scaling factor increases, thereby reducing the credit, because network congestion affects the transmission quality of the data packet. Through such a credit measurement formula, the network allocates bandwidth resources more intelligently to ensure that high-priority or important data packets can obtain the necessary bandwidth support even in congested conditions.

[0065] Network congestion detection: Through credit evaluation, the system detects network congestion. If network congestion is detected, the following measures are taken: Congestion indicators: Monitor network performance indicators such as bandwidth utilization, latency, packet loss rate, etc. Threshold setting: Set the congestion threshold. When the indicator exceeds the threshold, congestion processing is triggered.

[0066] Minimum guaranteed bandwidth: When network congestion occurs, the ONU (Optical Network Unit) will ensure that a minimum guaranteed bandwidth is provided for each data flow. This is usually achieved in the following ways: Traffic shaping: Limit the sending rate of each data flow to ensure that it does not exceed the minimum guaranteed bandwidth. Priority queue: Provide priority bandwidth for high-priority data flows. The minimum guaranteed bandwidth meets the following requirements:

[0067] B min =(T cycle -T g ×N)×R×W k / 8

[0068] Among them, T cycle Indicates the time it takes for all optical network units to poll once, T g represents the protection time, N represents the number of optical network units, R represents the transmission speed, and W k The weight for allocating bandwidth.

[0069] Minimum guaranteed bandwidth is the minimum amount of bandwidth guaranteed for each ONU, ensuring that each ONU can get enough bandwidth to transmit data even when the network load is high. The total time required for all ONUs to be polled once in a polling cycle. In the polling protocol, the network controller will query the status or request of each ONU in turn. The protection time is the reserved time used to handle burst traffic or ensure that critical data in the network is transmitted in time to prevent network congestion. The total number of optical network units in the network. Each ONU needs to be allocated a certain amount of bandwidth. In order to ensure that each ONU can get the minimum guaranteed bandwidth, the network needs to reserve a total amount of bandwidth. The product of (the time for all OUNs to be polled once) and (the protection time) gives a time window in which the network must provide bandwidth to all ONUs. The total time in this time window is divided by the product of (the number of ONUs) and (the transmission network speed) to get the total bandwidth requirement of all ONUs in the network. Finally, by multiplying the weight of the allocated bandwidth, the bandwidth allocation of each ONU is adjusted to adapt to different needs. The application of this formula helps to ensure that each ONU in the network can obtain basic bandwidth guarantees in multicast or broadcast scenarios, thereby improving the fairness and service quality of the network. In this embodiment, it is necessary to ensure that each ONU should meet the minimum required resources. In the network bandwidth allocation during network allocation, this means that each user can obtain at least the required bandwidth to ensure basic communication quality. In further specific implementation cases, the time for ONU to poll once, the protection time, and the hardware environment for allocating bandwidth weights to implement this formula may include:

[0070] Memory (RAM): Memory is used to store current data packets, historical data packets, and temporary data during the calculation process. Sufficient memory capacity ensures that the calculation of the weight function will not be affected by insufficient data.

[0071] Memory (ROM / Flash): The memory is used to store the operating system, firmware, configuration files, and algorithm code of the weight function.

[0072] Network interface: The network interface is used to receive and send data packets, including Ethernet, fiber optic interface, etc.

[0073] Timer: The timer / counter is used to record the time, the difference between polling and protection.

[0074] SNMP interface: The SNMP interface is used to communicate with ONU devices to obtain status information or configuration parameters of network devices, and also to receive trap messages from the agent. These messages are usually notifications about abnormalities or important events of network devices.

[0075] Weighted Fair Queuing (WFQ) Model: After the minimum guaranteed bandwidth is met, the ONU uses the weighted fair queuing model to optimize the allocation of the remaining bandwidth. The characteristics of the WFQ model include: Traffic classification: Divide the traffic into different categories, each with different weights. Bandwidth allocation: Dynamically allocate bandwidth based on the weight and traffic demand of each category. Fairness: Ensure that each category receives fair bandwidth allocation over a long period of time. Calculate the weight settings of the weighted fair queuing model to meet:

[0076]

[0077] Among them, B i is the bandwidth of the i-th priority request, B min is the minimum guaranteed bandwidth.

[0078] For heavy load environments, the weight coefficient satisfies:

[0079] H=B min ×α 1(n) +B e ×α 2(n) ×(H k +M k +L k ) / B t +a

[0080] Where H represents the bandwidth of high priority requests, and B min represents the minimum guaranteed bandwidth, the R k Represents the sum of high, medium and low request bandwidths, which is the sum of high priority, medium priority and low priority request bandwidths. t Represents the sum of bandwidth, which is the sum of all available bandwidth in the network. a The exact length of the high priority request arriving during the last round of waiting time under the medium priority. This refers to the bandwidth demand of the high priority request arriving during the waiting time of the medium priority request. α 1(n) Indicates that B is under level n business min The weight of the n-level service (medium priority or low priority) is used to reflect the importance of services of different priorities in bandwidth allocation. 2(0)Indicates the weight of the remaining bandwidth under level 0 service. This refers to the weight coefficient of the remaining bandwidth under level 0 service (which is the lowest priority or normal service). In a heavy load environment, the weight coefficient of each priority request will be used to determine the bandwidth allocation ratio. The weight coefficient first considers the ratio of the bandwidth of high-priority requests to the minimum guaranteed bandwidth to ensure that high-priority requests receive basic bandwidth guarantees. The ratio of the sum of the bandwidth of all priority requests to the sum of the total bandwidth is considered, which reflects the relative importance of requests of different priorities in the total bandwidth. This weighted fair queue model helps to ensure that high-priority services get appropriate bandwidth when the network is congested, while taking into account the needs of services of different priorities and the overall load of the network.

[0081] Dynamic bandwidth allocation: Through the WFQ model, the following dynamic bandwidth allocation is achieved: real-time monitoring, real-time monitoring of network traffic and bandwidth usage; adaptive adjustment, dynamically adjusting the bandwidth allocation of each category according to the real-time monitoring results; optimized performance, through dynamic bandwidth allocation, optimizing network performance and reducing congestion and latency.

[0082] Satisfy each group's request bandwidth H k +M k +L k The sum is less than the sum of the bandwidths B t When the condition is met, the bandwidth allocated to the ONU is the requested bandwidth, that is, B k =H k +M k +L k

[0083] The remaining bandwidth definition formula is as follows:

[0084]

[0085] Among them, B e Remaining bandwidth is the amount of unallocated bandwidth in the network after all requested bandwidth has been considered. k Indicates the bandwidth of high priority requests, the total amount of bandwidth available in the network, including the sum of the bandwidth of all links, interfaces, or bands. k Indicates the bandwidth of medium priority requests. The bandwidth required by requests marked as medium priority. The bandwidth required by requests marked as high priority. k Indicates the bandwidth required by requests marked as low priority. k Indicates bandwidth; refers to the bandwidth requirement of a single request, or the sum of the bandwidth of all requests in a priority class.

[0086] The remaining bandwidth weight setting satisfies:

[0087]

[0088] M represents the bandwidth of medium priority requests, L represents the bandwidth of medium and low priority requests, and the B min represents the minimum guaranteed bandwidth, the R k Represents the sum of high, medium and low request bandwidths. t represents the sum of bandwidth, α 1(n) Indicates that B is under level n business min The weight, α 2(n) Indicates the weight of the remaining bandwidth for level n services.

[0089] In a specific embodiment, when the sum of all requested bandwidths is less than or equal to the sum of the bandwidths, it means that the available bandwidth in the network is sufficient to meet all requests and does not exceed the total bandwidth capacity of the network. The bandwidth allocated to the ONUs (or other network devices) is directly equal to their requested bandwidth because network resources are sufficient. The defined remaining bandwidth refers to the amount of bandwidth that is not allocated when network resources are sufficient. If the sum of all requested bandwidths exceeds the sum of the bandwidths, then the network resources are insufficient to meet all requests. At this time, bandwidth needs to be allocated based on priority or other policies, and the remaining bandwidth will be zero or a negative value, indicating insufficient network resources. Through this formula, network administrators understand how much bandwidth is left unallocated in the network when all requested bandwidths are met, which is very helpful for optimizing network resources and making future bandwidth allocation decisions.

[0090] The above process involves packet classification, credit evaluation, congestion detection, provision of minimum guaranteed bandwidth, implementation of the WFQ model, and dynamic allocation of bandwidth, aiming to improve network performance and user experience.

[0091] S102: The optical line terminal records the source address information and forwarding rules of the classified data packets according to the classification result.

[0092] The optical line terminal learns the source MAC address according to the data frame, connects the MAC address with the physical port of the optical network unit, and stores it in the MAC address linked list, thereby realizing the function of recording the MAC address, port number and logical circuit identifier;

[0093] At the data transmission end, the Optical Line Terminal (OLT) records the source address information, learns the source MAC address, connects the MAC address to the physical port of the optical network unit, and stores it in the MAC address list to achieve the function of recording the MAC address, port number and logical circuit identifier:

[0094] The optical line terminal first receives data frames from the optical network. Data frames usually contain information such as source MAC address, destination MAC address, data payload, etc. Then the source address information is recorded. The OLT records the source MAC address of each data frame in order to track the source of the data packet. Learn the source MAC address. The OLT uses a process called "MAC address learning" to identify and store the MAC addresses of devices connected to its physical ports. When the OLT receives a data frame, it checks the source MAC address and associates it with the corresponding physical port. Establish a MAC address linked list. The OLT maintains a MAC address linked list that maps each MAC address to its corresponding physical port. The linked list also includes a logical circuit identifier, which is an identifier used inside the OLT to identify different service flows or virtual circuits. Connect the MAC address with the physical port. Once the association between the MAC address and the physical port is established, the OLT ensures that the data frame is forwarded to the correct physical port. This is usually achieved by using a lookup table (such as content address memory, CAM) inside the switch. Then store the information. The OLT stores the following information in the MAC address linked list: MAC address: source MAC address of the data frame. Physical port: the physical port where the data frame arrives. Logical circuit identifier: If applicable, the logical circuit identifier is used to identify a specific service flow or virtual circuit. As the network topology changes or devices move, the OLT needs to update the MAC address linked list to reflect these changes. The OLT uses the following mechanism to maintain the MAC address linked list. If no activity is detected for a certain MAC address for a period of time, the OLT will remove it from the linked list. When a new MAC address or device movement is detected, the OLT updates the linked list. By recording the MAC address, port number, and logical circuit identifier, the OLT: filters data frames and only allows the correct data frames to pass through the corresponding physical port. Implement VLAN (Virtual Local Area Network) to route data frames to the correct VLAN based on the logical circuit identifier. Traffic monitoring is used to monitor the traffic on each port for network management and optimization. This process is the basis of network switching and routing, ensuring that data is transmitted efficiently and accurately to the correct destination.

[0095] S103: The data receiving end selects an optimal transmission path for the classified data packet based on the source address information and the forwarding rule by using an improved Dijkstra algorithm.

[0096] The above improved Dijkstra algorithm sets a set G to record the vertices of the shortest path that has been found. Initially, the source point v0 is put into G. Each time a new vertex v is added to the set G i , the current shortest path length value from the source point v0 to the vertex in the set VG must be modified; the initialization module creates all ONUs as a set G, and the set G is used to store all ONUs participating in the communication, so that the initialization set G = {0};

[0097] The iteration module introduces dist[] to record the shortest path length from the initialization node to other nodes. If there is a direct path from a node v0 to another node vi, the weight dist[i] between the two nodes is recorded, and the iterator is used to implement the loop traversal of the nodes in the set G.

[0098] The update module uses the formula Determine whether the node meets the requirements, if so, update the ONU node. dist (u) represents the currently known shortest path length from the source node ONU to the secondary optical network unit u, τ(v0,v i ), means from v0 to v i The delay, w(v0,v i ) means from v0 to v i The bandwidth of α is the adjustment factor. Represents the shortest path.

[0099] Finally, the node with the shortest distance is selected as the current node, and the shortest path estimate and predecessor node of the neighboring node are updated; the shortest path estimate of each node is compared with the known path length to determine whether the path length and predecessor node need to be updated; based on the comparison results, the shortest path estimate and predecessor node of the node are determined as the optimal transmission path.

[0100] Specifically, the initialization module is responsible for setting the initial conditions of the algorithm, which usually includes the initial distance of each node (usually set to infinity, except for the source node, whose distance is 0) and the predecessor node (usually set to empty). The output of the initialization module provides initial data to the iteration module to start calculating the shortest path. The comparison module is responsible for comparing the shortest path estimate of the current node with the known path length. Compare the shortest path estimate of each node and determine whether the path length and predecessor node need to be updated. The output of the comparison module provides the comparison result to the node update module. The node update module updates the shortest path estimate and predecessor node of the node according to the result of the comparison module. The output of the node update module is usually not directly output to other modules, but reflects the result of path selection by updating the internal data structure. The initialization module sets the initial distance and predecessor node of all nodes. The iteration module selects a node with the smallest distance as the current node. The comparison module compares the shortest path estimate of the neighbor nodes of the current node. The node update module updates the shortest path estimate and predecessor node of the neighbor nodes according to the comparison result. Repeat the above steps until the shortest paths of all nodes are determined or a certain termination condition is reached.

[0101] In summary, the data forwarding method for multiple devices at a remote station provided by the present application includes the data sending end classifying data packets and dynamically allocating bandwidth, the optical line terminal recording the source address information and forwarding rules of the classified data packets, and the data receiving end selecting the best transmission path by improving the Dijkstra algorithm. These steps together improve the efficiency and reliability of data transmission. Through the Bayesian classifier and credit evaluation, it is ensured that high-priority data packets can still obtain sufficient bandwidth when the network is congested. The optical line terminal records the source address information and establishes forwarding rules to ensure accurate forwarding of data packets. The improved Dijkstra algorithm selects the best transmission path to reduce network delays and congestion. These measures not only optimize the utilization of network resources, but also improve the user experience and the overall performance of the network.

Claims

1. A method for realizing high-speed data forwarding of multiple devices at a remote station, characterized in that: include: The data transmitter classifies the data packets and dynamically allocates bandwidth based on the classification results; The optical line terminal records the source address information and forwarding rules of the classified data packets according to the classification results; the data receiving end selects the best transmission path for the classified data packets based on the source address information and the forwarding rules by improving the Dijkstra algorithm.

2. The method according to claim 1, characterized in that The data transmitting end classifies the data packets and dynamically allocates bandwidth according to the classification results, including: classifying the data packets using a Bayesian classifier; Conducting credit evaluation on the classified data packets; Based on the credit evaluation results, the optical network unit allocates the minimum guaranteed bandwidth to the classified data packets; Use the weighted fair queue model to dynamically allocate the remaining bandwidth.

3. The method according to claim 2, characterized in that The minimum guaranteed bandwidth satisfies: B min =(T cycle -T g ×N)×R×W k / 8 Among them, T cycle Indicates the time for all optical network units to poll once, T g represents the protection time, N represents the number of optical network units, R represents the transmission speed, and W k The weight for allocating bandwidth.

4. The method according to claim 3, characterized in that The method of using a weighted fair queue model to dynamically allocate the remaining bandwidth includes: Calculate the weight settings of the weighted fair queue model to satisfy: Among them, B i is the bandwidth of the i-th priority request, B min is the minimum guaranteed bandwidth.

5. The method according to claim 1, characterized in that: The optical line terminal records the source address information and forwarding rules of the classified data packets according to the classification result, including: The optical line terminal receives the data frame of the classified data packet and records the source MAC address, port number and logical circuit identifier of the data frame; The optical line terminal uses the MAC address learning process to associate the source MAC address with the corresponding physical port and store it in the MAC address linked list; The optical line terminal establishes and records forwarding rules according to the classification result and the MAC address chain list.

6. The method according to claim 1, characterized in that The data receiving end selects an optimal transmission path for the classified data packet based on the source address information and the forwarding rule by improving the Dijkstra algorithm, including: Set the initial distance and predecessor nodes of all nodes; Select the node with the shortest distance as the current node, and update the shortest path estimate of the neighboring nodes and the predecessor node; Compare the shortest path estimate of each node with the known path length to determine whether the path length and predecessor node need to be updated; According to the comparison result, the shortest path estimation and the predecessor node of the node are determined as the optimal transmission path.

7. The method according to claim 2, characterized in that: The classified data packets are evaluated for creditworthiness, followed by: Calculate the credit of each link; Prioritize the transmission of data packets on links with high credit.

8. The method according to claim 2, characterized in that: Conduct credit assessment on classified data packets, including: Credit rating is assessed based on the classified data packet type, historical behavior, and priority.

9. The method according to claim 1, characterized in that: The improved Dijkstra model includes: an initialization module for setting the initial conditions of the improved Dijkstra model algorithm; an iteration module for determining the node with the smallest distance as the current node, comparing the shortest path estimates of the neighboring nodes of the current node, and updating the shortest path estimates of the neighboring nodes and the predecessor nodes according to the comparison results; a comparison module for comparing the shortest path estimate of each node with the known path length; The node updating module is used to update the shortest path estimation and predecessor node of the node according to the result of the comparison module.

10. The method according to claim 1, characterized in that The data transmitting end classifies the data packets and dynamically allocates bandwidth according to the classification results, and also includes: The data sender uses a machine learning model to predict network traffic and adjusts the bandwidth allocation strategy based on the prediction results; The data sender reserves bandwidth based on the network traffic prediction results; The data sending end monitors the network traffic in real time and dynamically adjusts the reserved bandwidth.

Images