Service channel time slot allocation method

Abstract

The invention relates to the technical field of communication networks, and discloses a service channel time slot allocation method, which comprises the following steps of: 1, acquiring network data in real time from network equipment, the network data comprising bandwidth utilization rate, delay, packet loss rate and service flow type; 2, predicting future network flow fluctuation through a long-short-term memory network model based on historical flow data and real-time network data; and step 3, dynamically adjusting the allocation of time slot resources according to the network flow fluctuation prediction result, and preferentially allocating the time slot resources for high-priority services. According to the method, the dynamic adjustment of the time slot resources is realized by combining the flow prediction of the LSTM model and the Q-learning reinforcement learning algorithm, so that the system can flexibly allocate the resources according to the actual network load condition, the resource waste is reduced, the utilization efficiency of the bandwidth is improved, and the utilization rate of the overall network resources is remarkably improved especially under the condition of high network load.

Description

Technical Field

[0001] The invention relates to the technical field of communication networks, and in particular to a service channel time slot allocation method. Background Art

[0002] With the continuous advancement of informatization, the number of users and types of services in communication networks have shown explosive growth. In particular, high-priority services such as video streaming, voice calls, and telemedicine have put forward higher requirements on network bandwidth, latency, and transmission stability. In this context, network operators must ensure that time slot resources in service channels can be efficiently allocated to meet users' demand for high-quality network services. At the same time, the volatility and complexity of network traffic also pose new challenges to the allocation of time slot resources. Traditional static allocation methods are difficult to adapt to the real-time changes in network traffic and the needs of diversified services.

[0003] Existing time slot resource allocation methods usually use static rules or preset strategies to allocate bandwidth resources for different services. These methods rely on fixed-ratio resource allocation, fail to fully consider the dynamic changes of network traffic, and lack the ability to predict future traffic fluctuations. When network traffic changes suddenly or service requirements are adjusted, traditional methods are often difficult to respond in a timely manner, resulting in increased delays for high-priority services due to insufficient resources, and low-priority services may occupy too many resources, affecting the overall network service quality. In addition, real-time feedback data is rarely used in existing technologies to adjust allocation strategies, and the flexibility and adaptability of resource allocation are weak.

[0004] However, the main problem with the existing technology is that it cannot dynamically adjust time slot resource allocation according to the real-time network status, resulting in insufficient resources for high-priority services when the network load increases or burst traffic occurs, and the service quality decreases. Faced with a complex and changing network environment, traditional methods are difficult to flexibly adjust resource allocation according to actual needs, and cannot effectively guarantee low latency and stable transmission of high-priority services. Therefore, there is an urgent need for a solution that can dynamically adjust time slot resource allocation to adapt to real-time changes in network traffic, ensure priority protection of key services, and improve the resource utilization efficiency of the overall network. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a service channel time slot allocation method, which solves the problem that time slot resource allocation cannot be dynamically adjusted according to the real-time network status, resulting in insufficient resources for high-priority services and reduced service quality when the network load increases or burst traffic occurs.

[0006] To achieve the above purpose, the present invention is implemented by the following technical scheme: a service channel time slot allocation method, comprising the following steps:

[0007] Step 1: Collect network data from network devices in real time, the network data including bandwidth utilization, delay, packet loss rate and service traffic type;

[0008] Step 2: Predict future network traffic fluctuations through a long short-term memory network model based on historical traffic data and real-time network data;

[0009] Step 3: According to the network traffic fluctuation prediction result, dynamically adjust the allocation of time slot resources, and give priority to allocating time slot resources to high priority services;

[0010] Step 4: Use the Q-learning reinforcement learning algorithm to optimize the time slot allocation strategy based on real-time feedback of network performance data;

[0011] Step 5: Based on the feedback mechanism, dynamically adjust the time slot resources allocated to different service channels.

[0012] Preferably, the network data collection step includes:

[0013] Using SNMP protocol to communicate with network devices, including base stations, routers and switches;

[0014] Obtaining real-time performance data of network devices through the configuration management information base, the performance data including bandwidth utilization, delay and packet loss rate;

[0015] The traffic in-depth analysis tool is used to classify the business traffic, and the business traffic types include multimedia video streaming with high bandwidth requirements, voice calls with low latency requirements, and data transmission that is not sensitive to latency.

[0016] Preferably, the training step of the long short-term memory network model includes:

[0017] Use the historical traffic data of the past hour as the input of the model, the historical data includes the bandwidth occupancy, delay fluctuation and packet loss rate of different business traffic;

[0018] Calibrate the model based on the current real-time collected network data, wherein the real-time data includes the number of user accesses, service traffic type and bandwidth utilization;

[0019] The model's hyperparameters, including learning rate, number of hidden layer units, and number of training iterations, were adjusted by cross-validation to optimize the model's prediction accuracy.

[0020] Preferably, the network traffic fluctuation prediction step based on the long short-term memory network model includes:

[0021] The model outputs the network traffic forecast results for the next 30 minutes, including the expected peak and valley traffic periods;

[0022] The model weights and parameters are continuously modified through a feedback control mechanism, and the modification is based on the error value between the actual network load data and the predicted results.

[0023] Preferably, the state space of the Q-learning reinforcement learning algorithm includes:

[0024] Current network load level, bandwidth utilization, service priority, time slot resource utilization, and real-time feedback of network delay and packet loss rate;

[0025] The reinforcement learning algorithm optimizes the time slot allocation strategy by updating the Q value, and the Q value is updated based on the feedback information obtained after allocating the time slot resources.

[0026] Preferably, the real-time feedback adjustment of the time slot resource allocation strategy includes:

[0027] When the network load increases, dynamically reduce the time slot resources allocated to low-priority service channels;

[0028] Automatically increase time slot allocation for high priority services based on real-time network feedback data, including current delay and packet loss rate;

[0029] When the time slot resources of the low-priority service channel are not fully used, the unused time slot resources are automatically recovered and reallocated to the current high-load service.

[0030] Preferably, the feedback data collection frequency of the time slot resource allocation is once per second, and the feedback data includes:

[0031] The time slot resource utilization rate of each service channel, the actual bandwidth utilization rate, the transmission delay and packet loss rate of the current network;

[0032] The feedback mechanism determines whether the service quality of each business channel meets the preset standard. If not, the system will automatically adjust the allocated time slot resources.

[0033] Preferably, the system automatically adjusts the time slot allocation strategy according to different network environments, including:

[0034] In mobile communication networks, the time slot resources allocated to each base station are dynamically adjusted based on the user's mobility and base station load;

[0035] In satellite communication networks, strict flow control and time slot reservation strategies are used to ensure that key services such as navigation and telemedicine have priority in obtaining time slot resources;

[0036] In a fiber-optic communication network, time slot resources are dynamically allocated based on the number of concurrent requests and bandwidth requirements of users.

[0037] Preferably, the feedback mechanism further comprises:

[0038] Compare the traffic fluctuations output by the prediction model with the actual network status. When the error exceeds the preset threshold, the system automatically adjusts the weights and parameters of the long short-term memory network model.

[0039] By iteratively optimizing the accuracy of the prediction model, we can ensure that the prediction results of future traffic fluctuations are closer to the actual network status.

[0040] Preferably, the time slot resource allocation is performed according to service priority, and the priority strategy includes:

[0041] High-priority service channels include real-time video calls, telemedicine, and navigation services, while low-priority service channels include file transfers and general data browsing;

[0042] During high-load periods, time slot resources are allocated preferentially to high-priority services to ensure that their delays and service quality are not affected, while reclaiming time slot resources that are not fully used by low-priority services.

[0043] The present invention provides a service channel time slot allocation method, which has the following beneficial effects:

[0044] 1. The present invention realizes the dynamic adjustment of time slot resources by combining the traffic prediction of LSTM model and Q-learning reinforcement learning algorithm. This method enables the system to flexibly allocate resources according to the actual network load, reduces resource waste, and improves the utilization efficiency of bandwidth, especially under high network load conditions, significantly improving the utilization rate of overall network resources.

[0045] 2. The present invention effectively reduces network delay and packet loss rate by intelligently predicting network traffic fluctuations and allocating resources to high-priority services first. High-priority services (such as video calls, telemedicine, etc.) can still maintain low delay and stable transmission quality during network congestion, ensuring efficient execution of key tasks and improving user experience.

[0046] 3. The present invention can allocate resources in advance for network peak periods through traffic prediction based on historical data, ensuring that the system can still maintain efficient operation under high load conditions. In particular, for high-priority services such as video streaming and voice calls that are sensitive to delays, the system will reserve sufficient time slot resources for them, effectively avoiding network congestion during peak periods.

[0047] 4. The present invention uses Q-learning algorithm to self-optimize resource allocation strategy according to real-time feedback data, which significantly improves the flexibility of resource allocation. The system can adapt to different network environments (such as mobile communication, satellite communication, optical fiber communication, etc.), perform optimal resource allocation according to the characteristics of different networks, and effectively adapt to the dynamic changes of network status.

[0048] 5. The present invention improves user experience while reducing network delay and improving bandwidth utilization through intelligent time slot allocation and optimization algorithm. High-priority services are given priority, network services are more stable, and user experience is particularly improved during peak hours or sudden load conditions, greatly improving system reliability.

[0049] 6. The present invention can monitor the network status in real time and dynamically adjust the time slot resource allocation, avoiding the occurrence of excessive network congestion. Through the feedback mechanism and adaptive optimization, the system can make timely adjustments, effectively reducing the frequency of congestion events, maintaining the stability of the network, and improving the overall network performance.

[0050] 7. The present invention ensures the priority transmission and service quality of key services (such as telemedicine and real-time navigation) by reserving sufficient time slot resources for high-priority services, especially during periods of high load and network congestion. Compared with traditional solutions, the guarantee rate of key services is significantly improved, ensuring that the system can meet business needs in key scenarios.

[0051] 8. The present invention optimizes resource management under different networks by analyzing the characteristics of different types of network environments (such as mobile, satellite, and optical fiber) and combining adaptive time slot resource allocation strategies. The system can perform personalized time slot allocation according to the specific needs of each network, ensuring that all types of services can obtain the best resource support in heterogeneous network environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0053] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] Embodiment 1:

[0055] Please refer to the attached Figure 1 , an embodiment of the present invention provides a service channel time slot allocation method, comprising the following steps:

[0056] Step 1: Network data collection

[0057] In this embodiment, network data collection is the basis for service channel time slot allocation. The system collects network status data in real time through communication with multiple network devices (such as base stations, routers, switches) to ensure accurate transmission of service flow information. The specific implementation is as follows:

[0058] In this embodiment, the system communicates with network devices by using the SNMP protocol, periodically sending query requests to obtain real-time performance data. The SNMP protocol obtains device status by reading the device's management information base (MIB), including:

[0059] Bandwidth utilization: used to determine the current network resource load and help the system understand the bandwidth usage of each service channel.

[0060] Latency: measures the transmission time of network data packets in the network. It is mainly used to evaluate the real-time performance of the network and ensure the service quality of high-priority business traffic (such as video calls and telemedicine) when high bandwidth requirements are met.

[0061] Packet loss rate: used to evaluate the stability of network transmission. When the packet loss rate increases, it may mean that the network transmission quality has deteriorated and the time slot resources need to be dynamically adjusted.

[0062] In this embodiment, the system uses a traffic depth analysis tool to classify traffic. The main goal of traffic classification is to prioritize each service according to different types of service requirements (such as video streaming, voice calls, and data transmission). For example, the system can analyze the protocol header information of each data packet to identify whether it is a video stream with high bandwidth requirements or a voice stream that is sensitive to delay.

[0063] In this embodiment, the collected data is stored and managed through a NoSQL database (such as MongoDB). MongoDB has high concurrent access capabilities to ensure efficient real-time data storage. The system collects data from multiple network devices every 1 second to ensure that the collected network status information can accurately reflect the current operation status of the network and provide support for subsequent steps.

[0064] Accurate collection of network data is the basis for time slot allocation. Without the support of comprehensive and real-time network data, subsequent traffic prediction and time slot resource allocation will not be carried out correctly. By collecting bandwidth utilization, delay and packet loss rate, the system has a clear understanding of the network operation status.

[0065] Step 2: Network traffic fluctuation prediction

[0066] In this embodiment, based on historical and real-time network data, the system uses the long short-term memory network (LSTM) model to predict traffic fluctuations. The LSTM model is good at processing time series data and can predict future traffic trends based on past traffic patterns. The specific implementation is as follows:

[0067] In this embodiment, the input data of the LSTM model comes from the historical network data collected in the past hour. Each input data point includes information such as past bandwidth utilization, latency, and packet loss rate, and the system refreshes this data every minute. The LSTM model learns the changing pattern of network traffic through these data points, especially the time and regularity of traffic peaks and troughs.

[0068] In this embodiment, the system calibrates the LSTM model through real-time network data. Real-time data not only includes the current bandwidth usage, but also includes the number of user accesses, the types of different service flows and their bandwidth usage. Real-time calibration ensures that the model predicts future network loads more accurately and avoids excessive prediction deviations.

[0069] In this embodiment, the system uses cross-validation to adjust the hyperparameters of the LSTM model. These hyperparameters include learning rate, number of hidden layer neurons, and number of training iterations. By optimizing these parameters, the LSTM model can improve prediction accuracy during the training process and ensure the stability and accuracy of the prediction results.

[0070] In this embodiment, the output of the LSTM model is the network traffic prediction results for the next 30 minutes. These prediction results include the expected peak and valley traffic periods, and the system will reserve time slot resources for high-load periods in advance based on these results. In order to further optimize the prediction accuracy, the system will adjust the model parameters based on the actual network feedback to ensure that the prediction results are close to the actual situation.

[0071] Traffic fluctuation prediction provides a basis for time slot resource allocation. By predicting future traffic, the system can prepare for resource allocation in advance to ensure that during peak network traffic periods, the system has reserved enough time slot resources to ensure the transmission quality of high-priority services.

[0072] Step 3: Dynamic allocation of time slot resources

[0073] In this embodiment, the system dynamically adjusts the time slot resource allocation based on the traffic fluctuation prediction results. High-priority services (such as video calls, telemedicine, etc.) are given priority in obtaining time slot resources during resource allocation to ensure their quality. The specific implementation is as follows:

[0074] In this embodiment, the system allocates different service flows to channels of different priorities and sets different time slot resource allocation ratios for each priority. Services with high real-time requirements such as video streaming and telemedicine will be allocated more time slot resources, while low-priority services such as file transfer will be dynamically allocated based on the remaining resources.

[0075] In this embodiment, the system reserves sufficient time slot resources according to the traffic peak predicted by the LSTM model. For example, when it is predicted that the video call traffic will increase significantly in the next 10 minutes, the system will allocate sufficient time slot resources for the video stream channel in advance during this period to avoid increased delay or increased packet loss rate due to insufficient resources.

[0076] In this embodiment, low priority services (such as non-real-time data transmission) will be allocated based on the current remaining resources after the high priority services are allocated. If the network load is low, low priority services can obtain more time slot resources, and when the load is high, their resources will be dynamically compressed.

[0077] In this embodiment, the allocation strategy of time slot resources is determined by the priority mechanism and the prediction results of the LSTM model. For example, file transfer services are low-priority services. When the network load increases, the system automatically reduces the time slot resources for such services and ensures that the time slots of high-priority service traffic are not affected.

[0078] The allocation of time slot resources directly depends on the prediction results of the LSTM model. The system reasonably allocates resources according to future traffic demand to ensure that high-priority services can maintain high-quality services during peak traffic periods, while low-priority services can be appropriately adjusted when resources are limited.

[0079] Step 4: Optimization strategy of Q-learning reinforcement learning algorithm

[0080] In this embodiment, the system continuously optimizes the time slot resource allocation strategy through the Q-learning reinforcement learning algorithm. Q-learning gradually converges to the optimal allocation plan by learning the feedback data under different allocation strategies. The specific implementation is as follows:

[0081] In this embodiment, the Q-learning algorithm regards each time slot resource allocation as a learning process and records the feedback data after each allocation. These feedback data include network delay, packet loss rate, bandwidth utilization, etc. By analyzing these feedback data, the system can evaluate the effect of the current allocation strategy.

[0082] In this embodiment, the state space of the Q-learning algorithm includes the current load of the network, the priority of each service, the time slot resource usage, and the feedback network performance data. After each time slot resource allocation, the system updates the Q value according to the feedback data. By continuously adjusting and updating the Q value, the system can find the optimal resource allocation strategy.

[0083] In this embodiment, when the network load fluctuates greatly, the Q-learning algorithm can dynamically adjust the time slot resource allocation. For example, when the delay of the video call service suddenly increases, the system will increase the time slot resources allocated to the video channel and reduce the resource allocation of low priority services according to the learning results of the Q-learning algorithm.

[0084] In this embodiment, the convergence process of the Q-learning algorithm depends on a large amount of historical allocation results and feedback data. As the system running time increases, the algorithm can gradually converge to the optimal time slot allocation strategy, thereby ensuring efficient and stable resource allocation of the system under different network loads.

[0085] The Q-learning algorithm continuously optimizes the time slot resource allocation strategy through a feedback mechanism and can make adjustments based on the dynamic changes of the network. The algorithm complements the prediction function of the LSTM model to ensure that reasonable resource allocation can be maintained when the network fluctuates suddenly.

[0086] Step 5: Feedback mechanism and dynamic adjustment

[0087] In this embodiment, the feedback mechanism monitors network performance data in real time and dynamically adjusts the time slot allocation strategy based on the feedback to form a closed-loop optimization system. The specific implementation is as follows:

[0088] In this embodiment, the system collects feedback data in real time by monitoring the time slot resource usage of each service channel, including bandwidth utilization, network delay, packet loss rate, etc. The system collects feedback data once a second to ensure that the data can reflect changes in network status in real time.

[0089] In this embodiment, when the feedback data indicates that the delay or packet loss rate of a service channel has increased, the system will automatically reclaim underused low-priority time slot resources and reallocate them to high-priority service channels. For example, when the delay of a video call channel increases, the system will reclaim unused time slot resources from the file transfer channel to ensure the quality of the video channel.

[0090] In this embodiment, feedback data is not only used for short-term resource adjustment, but also for optimizing the parameters of the LSTM model and the learning process of the Q-learning algorithm. The system fine-tunes the prediction parameters of the LSTM model through long-term accumulated feedback data, so that it can better adapt to future network traffic fluctuations.

[0091] In this embodiment, the feedback adjustment mechanism of the system is an adaptive closed-loop system, which continuously optimizes the time slot allocation strategy based on feedback data, and makes adjustments through the Q-learning algorithm and LSTM model to ensure that the system's resource allocation remains efficient and stable under different load conditions.

[0092] The feedback mechanism ensures that the system can respond to the dynamic changes of the network in real time and continuously optimize the allocation strategy through feedback data. By combining with the LSTM model and Q-learning algorithm, the system forms a complete closed loop to ensure the rational and efficient allocation of resources in a complex network environment.

[0093] Embodiment 2:

[0094] The second embodiment aims to optimize the prediction and allocation of time slot resources by combining the historical behavior patterns of users, so that the system can reserve resources for high-priority services in advance when resources are tight and reduce network congestion. This embodiment can effectively improve the system's response speed to changes in user network requirements and improve the utilization efficiency of time slot resources. Specifically, it includes the following steps:

[0095] Step 1: Data collection and classification of user behavior patterns

[0096] In this embodiment, the system not only collects network performance data, but also introduces data monitoring of user behavior patterns. The specific collected user behavior data includes user online time, service type preference, network access frequency, etc.

[0097] In this embodiment, the user behavior pattern continuously collects the user's online history through the system. For example, the system will record the user's network activities at different times of the day, including the type of service accessed (such as video calls, web browsing, file transfers, etc.), the device used (such as mobile phones, computers, etc.), the type of business traffic, and other information. These data are broken down into behavioral features and associated with network performance data (such as latency, packet loss rate).

[0098] In this embodiment, the analysis of user behavior patterns is classified through machine learning algorithms. The system classifies users into different categories such as "high-frequency video call users" and "low-frequency file transfer users" by comparing the Internet time and frequency of different users. The classified user behavior patterns will be associated with the business traffic data to form a specific behavior model.

[0099] Step 2: Traffic prediction based on user behavior patterns

[0100] This embodiment predicts network traffic based on the collected user behavior data and the LSTM model. With the intervention of user behavior data, the LSTM model can more accurately predict future network load changes and allocate resources in advance for peak hours.

[0101] In this embodiment, the system uses the user behavior pattern as an input layer of the LSTM model, and further improves the model's prediction accuracy for traffic fluctuations by combining the user's historical behavior data. For example, if a user made a video call from 10 to 11 a.m. every day in the past week, the user is very likely to make similar operations during the same time period in the future. Therefore, the system will estimate the future traffic load in advance based on this rule.

[0102] In this embodiment, the system processes the user's behavior pattern in segments, and different user groups are classified according to their Internet habits and peak and off-peak periods. For users who use the network frequently, the system reserves more time slot resources during these periods, and for users who use less frequently, their time slot resource allocation is reduced accordingly.

[0103] Step 3: Dynamic time slot resource allocation

[0104] Based on the traffic prediction results of user behavior patterns, the system can dynamically adjust the time slot resource allocation in different time periods. Specifically, the system will reserve sufficient time slot resources in advance for users who use the network frequently, ensuring that the needs of these users are met first during periods of high network load.

[0105] In this embodiment, when the system predicts that a user will conduct a high-bandwidth service (such as a video call or online game) within a specific time period, the system will reserve sufficient time slot resources for the user in advance. In this case, even if other users in the network also use the network during this period, the service quality of the user will not be affected.

[0106] In this embodiment, when the user behavior pattern shows that a user surfs the Internet during a period of low network load, the system can reduce the time slot resource allocation to the user. For example, if a user usually downloads files late at night, the system can allocate fewer resources to the user during this period because the available bandwidth in the network is relatively sufficient at this time and no additional resources need to be reserved.

[0107] In this embodiment, the allocation of time slot resources is not only based on the current real-time traffic data, but also based on the prediction of future traffic peaks in combination with user behavior patterns. In this way, the system can dynamically adjust the resource allocation strategy in different time periods. For example, during the morning rush hour on weekdays, the system allocates more resources to users who often conduct video conferences, while during non-working hours, the system can allocate more resources to other users.

[0108] Step 4: Resource optimization combined with Q-learning

[0109] In order to further improve the efficiency of resource allocation, this embodiment introduces the Q-learning reinforcement learning algorithm to optimize the allocation strategy of time slot resources by continuously learning user behavior patterns and network feedback.

[0110] In this embodiment, the system uses the Q-learning algorithm to optimize the allocation of time slot resources. The Q-learning algorithm dynamically adjusts the allocation of time slot resources by analyzing the network usage of different users in different time periods. For example, for high-priority users, when the system finds that their behavior patterns are stable and their network requirements are high, the system will automatically increase the time slot resources of the user; on the contrary, for low-priority users, the Q-learning algorithm will reduce unnecessary time slot allocation to avoid resource waste.

[0111] In this embodiment, the system uses the Q-learning algorithm to learn network feedback after each allocation, especially indicators such as bandwidth utilization, delay, and packet loss rate, and gradually optimizes the time slot allocation strategy in each time period. For example, if a user frequently initiates high bandwidth demands in a certain time period, and the network load in this time period is high, the Q-learning algorithm will adjust the strategy to ensure that the user's key business obtains more time slot resources in the next allocation.

[0112] Step 5: Feedback mechanism and continuous optimization

[0113] This embodiment relies on a real-time feedback mechanism to continuously optimize the allocation of time slot resources. Through long-term monitoring and learning of user behavior patterns and network feedback, the system can continuously adjust the allocation strategy to ensure efficient use of resources.

[0114] In this embodiment, the system monitors the network usage of each user in real time and adjusts resource allocation according to changes in user behavior patterns. For example, if a user's behavior pattern changes after a period of time, the system will relearn and adapt to the new behavior pattern to ensure that new network needs are met.

[0115] In this embodiment, the feedback data after each time slot allocation, including the user's network service quality (such as delay, packet loss rate), will be input into the Q-learning algorithm as reference data for optimizing the next resource allocation. For example, if the network service quality of a user continues to decline, the system will automatically allocate more time slot resources to the user until its service quality returns to the expected level.

[0116] By combining the analysis of user behavior patterns and the Q-learning reinforcement learning algorithm, this embodiment can predict the network needs of users in different time periods in advance and dynamically adjust the allocation of time slot resources to ensure that network resources are reasonably allocated and meet the needs of high-priority users. At the same time, the system can continuously optimize resource allocation strategies based on real-time feedback data to avoid resource waste and ensure that when the network load is peak, the system can allocate sufficient resources for key services in advance to ensure efficient operation.

[0117] Embodiment three:

[0118] The core of the third embodiment is to enable the system to flexibly adjust the allocation of time slot resources in different types of network environments (such as mobile communication networks, satellite communication networks, and optical fiber communication networks) through an adaptive time slot resource allocation strategy, thereby ensuring efficient and stable resource management in heterogeneous network environments.

[0119] Step 1: Data collection in multiple network environments

[0120] In this embodiment, the system needs to collect real-time performance data under different network environments to ensure that a reasonable time slot resource allocation strategy is made according to the characteristics of each network. The specific collected data includes network bandwidth utilization, delay, packet loss rate and characteristic parameters of network types.

[0121] In this embodiment, the system first communicates with various network devices (base stations, satellite terminals, optical fiber nodes, etc.) through the SNMP protocol to obtain real-time network status data. The collected data includes key performance indicators such as bandwidth utilization, time slot resource utilization, packet loss rate and network delay in different network environments.

[0122] In this embodiment, the system configures different collection frequencies and data types according to the characteristics of different networks. For example, in an optical fiber communication network, due to the low transmission delay and large bandwidth, the system collects data every 10 seconds; while in a mobile communication network, the user location and base station load fluctuate greatly, and the system collects data at a higher frequency (once per second) to adjust the resource allocation strategy in real time.

[0123] In this embodiment, the system associates the collected performance data with the user's traffic demand and historical behavior data to generate a time slot resource allocation table for each network environment, ensuring that resource usage in different network environments is monitored and adjusted in real time.

[0124] Step 2: Characteristic analysis and classification of different network environments

[0125] In this embodiment, the system establishes resource allocation strategies for different network environments by analyzing and classifying the characteristics of different types of networks. For example, the resources of mobile communication networks fluctuate greatly, while the bandwidth of satellite communication networks is relatively limited, so the system needs to optimize the resource allocation plan according to the characteristics of each network.

[0126] In this embodiment, the system first classifies the network environment. The system classifies the network into three categories based on its physical characteristics: mobile communication networks (such as 4G and 5G networks), satellite communication networks, and optical fiber communication networks. The system establishes different resource allocation strategy models for each network.

[0127] In this embodiment, for the mobile communication network, the system will pay special attention to the load of the base station, the mobility of the user, and the frequent base station switching. In this network environment, the bandwidth utilization and delay fluctuate greatly. The system dynamically adjusts the time slot resource allocation of each user by monitoring the user's access status and the base station load in real time to ensure network load balance. For example, when the load of a base station is too large, the system will transfer part of the time slot resources from other base stations, or dynamically adjust the resource allocation of low-priority services.

[0128] In this embodiment, for the satellite communication network, the system pays more attention to the efficient use of bandwidth and the guarantee of key services. Since the bandwidth resources of satellite communication are limited, the system will give priority to allocating time slot resources to high-priority services (such as navigation, telemedicine, etc.) and strictly control the use of time slot resources for low-priority services. For example, in satellite communication, the resources for video streaming and file transmission will be limited to ensure the real-time transmission of navigation data.

[0129] In this embodiment, for the optical fiber communication network, due to its fast transmission speed, low latency and large bandwidth, the system can ensure efficient use of the network by adjusting the time slot resources of concurrent users. In the optical fiber communication environment, the system can flexibly adjust the time slot allocation strategy for each user to allow more concurrent access, thereby reducing network congestion.

[0130] Step 3: Adaptive time slot resource allocation strategy

[0131] In this embodiment, the system adaptively adjusts the allocation strategy of time slot resources based on the analyzed and classified network characteristics. The system adaptively adjusts the time slot resources according to the real-time status in different network environments to ensure the stability and efficiency of the network.

[0132] In this embodiment, the system first initializes the time slot resource allocation strategy based on the collected network performance data. In the optical fiber communication network, the system will automatically allocate more time slot resources to services with high bandwidth requirements (such as video streaming or file transfer), while in the satellite communication network, the system will strictly control the allocation of time slot resources to ensure that the bandwidth of key services such as navigation and telemedicine is given priority.

[0133] In this embodiment, the adaptive allocation strategy relies on a real-time feedback mechanism. For example, when the system detects that a base station is overloaded, the system will automatically reduce the time slot resource allocation for low-priority services (such as file transfer) to ensure that the service quality of high-priority services (such as video calls or online games) is not affected. At the same time, the system will automatically recycle unused time slot resources for use by other users or network devices.

[0134] In this embodiment, the system will also adaptively adjust the priority of each service according to the dynamic changes in the network load. For example, when the number of concurrent users in the optical fiber communication network increases, the system will reduce the time slot resource allocation for some non-critical services. In the mobile communication network, when the user switches base stations or the network load increases dramatically, the system will increase the time slot resources for video calls or voice calls in real time to ensure low latency and stable service quality.

[0135] Step 4: Combined with Q-learning algorithm optimization

[0136] In this embodiment, the system continuously optimizes the time slot resource allocation strategy through the Q-learning reinforcement learning algorithm. The Q-learning algorithm gradually finds the optimal time slot resource allocation solution by learning feedback data under different network environments.

[0137] In this embodiment, the Q-learning algorithm uses the characteristics of each network environment as state space input and determines the allocation of time slot resources by analyzing the current bandwidth utilization, delay, packet loss rate and other indicators. After each allocation, the system updates the Q value according to the network feedback data and dynamically adjusts the allocation strategy. For example, in satellite communications, the Q-learning algorithm will automatically reduce the allocation of time slot resources for non-critical services to ensure the real-time nature of navigation data.

[0138] In this embodiment, the Q-learning algorithm continuously optimizes the allocation strategy by learning historical allocation results and real-time feedback. In a mobile communication network, the algorithm analyzes the load of different base stations and user behavior patterns, adjusts the dynamic allocation of time slot resources, and ensures that when users switch base stations, the smooth transition of resource allocation does not affect the service quality.

[0139] In this embodiment, when the system detects sudden changes in the network environment (such as base station congestion, insufficient satellite bandwidth, etc.), the Q-learning algorithm will quickly learn the feedback of the current environment and make corresponding adjustments. For example, when the network load of a base station suddenly increases, the system will reduce the resources of low-priority services in real time to ensure the service quality of high-priority services.

[0140] Step 5: Feedback mechanism and continuous optimization

[0141] In this embodiment, the feedback mechanism monitors the network status in real time to ensure that the system can continuously optimize the time slot resource allocation strategy according to changes in different network environments.

[0142] In this embodiment, the system collects feedback data in real time, including bandwidth utilization, time slot resource utilization, delay and packet loss rate, etc. The feedback data is used by the system to adjust the allocation strategy of the current time slot resources and is used as input to retrain the Q-learning algorithm to further optimize the future allocation strategy.

[0143] In this embodiment, the feedback mechanism is used not only for real-time adjustment but also for long-term optimization. For example, the system will accumulate feedback data in various network environments, evaluate long-term network performance, and gradually optimize the overall strategy for time slot resource allocation to ensure that resources can be allocated more efficiently in the same or similar environments in the future.

[0144] In this embodiment, the system will update the network priority strategy based on the feedback data to ensure that key services are always given priority in different network environments. For example, in a satellite communication network, the system continuously optimizes the time slot allocation strategy for navigation services through feedback data to ensure that its real-time performance is not affected by other low-priority services.

[0145] Through this embodiment, the system can adaptively adjust the time slot resource allocation strategy according to different types of network environments, ensure efficient resource management in heterogeneous networks, and improve the overall network service quality and resource utilization efficiency. Especially when the network load fluctuates greatly, the adaptive allocation strategy is combined with the Q-learning reinforcement learning algorithm to ensure that the system can flexibly respond to changes in the network environment and provide priority resource guarantees for key services.

[0146] This adaptive time slot resource allocation method enables the system to have greater flexibility and adaptability, able to cope with complex and changing network environments, and ensure the efficiency and stability of network services.

[0147] Table 1

[0148]

[0149] Experimental data comparison

[0150] Average network delay: In Example 1 (standard time slot allocation method), the average network delay is relatively high because there is no optimization for user behavior or network environment. In Example 2, the network delay is significantly reduced by introducing user behavior pattern prediction. In Example 3, due to the combination of adaptive allocation strategies for multiple network environments, further optimization is achieved under different network environments, with the lowest delay.

[0151] Packet loss rate: Embodiments 2 and 3 both reduce the packet loss rate by optimizing the allocation of time slot resources. Embodiment 3 in particular uses adaptive network characteristics to adjust the resource allocation strategy, thereby reducing the packet loss rate to 0.4%.

[0152] Bandwidth utilization: Example 2 optimizes user behavior patterns to achieve a bandwidth utilization of 85%; and the adaptive allocation strategy of Example 3 further improves the bandwidth utilization to 90%.

[0153] Time slot resource allocation response time: In a multi-network environment, through real-time feedback and adaptive adjustment, the time slot resource allocation response time is the fastest, which is 60 milliseconds, significantly better than other embodiments.

[0154] Resource utilization during peak hours: Example 3 significantly improves resource utilization during peak hours by predicting traffic fluctuations in different network environments and optimizing allocation for key services, reaching 85%.

[0155] User satisfaction score: Example 3 has significantly improved key business assurance and response speed, and therefore has the highest user satisfaction score of 4.5 points.

[0156] High-priority service delay: The key service delay of the third embodiment is the lowest, which can ensure the smoothness and stability of high-priority services such as video calls and telemedicine.

[0157] Key business guarantee rate: The key business guarantee rate of Example 3 is 98%, which is significantly improved compared with Example 1 (90%) and Example 2 (95%).

[0158] Network congestion: The adaptive allocation strategy of the third embodiment significantly reduces the frequency of network congestion, with only one congestion event per hour, while the number of congestion events in the first embodiment is relatively large.

[0159] Resource allocation flexibility score: Since Example 3 can flexibly allocate time slot resources in a variety of network environments, it has the highest flexibility score of 4.8 points.

[0160] Through the comparison of experimental data, the key performance indicators of Example 2 and Example 3 are significantly better than Example 1. Among them, Example 3 significantly improves the system's response speed, bandwidth utilization, and high-priority service guarantee capabilities by combining adaptive time slot resource allocation and Q-learning algorithm in multiple network environments, and is particularly outstanding in reducing network delay and packet loss rate.

[0161] Comparative Example 1:

[0162] The following is a traditional comparison. This traditional solution represents a more basic method for allocating service channel time slots. It lacks intelligent prediction and dynamic adjustment mechanisms and mainly relies on static rules for time slot allocation. Specifically, it includes the following steps:

[0163] Step 1: Static rule network data collection

[0164] In the traditional comparison example, the system collects data from network devices through static rules. The specific data items collected include bandwidth utilization, network delay, and packet loss rate, but the solution does not use real-time monitoring or intelligent data analysis tools.

[0165] In this traditional comparison example, network data collection relies only on simple SNMP protocol queries. The system collects network performance data from base stations, routers, and switches at fixed time intervals (e.g., every 5 minutes). This data does not include real-time behavior monitoring, so it cannot reflect sudden changes in network status in a timely manner.

[0166] In this traditional comparison example, traffic classification is also performed using static rules. For example, the system simply distinguishes service types based on port numbers or protocol types (such as HTTP, FTP), and allocates corresponding resources based on fixed priorities. This method does not have dynamic optimization for specific service needs, so it cannot accurately respond to service traffic fluctuations.

[0167] Step 2: Static time slot resource allocation without intelligent prediction

[0168] In the traditional comparison example, the system does not use any traffic prediction model, and the time slot resource allocation is based on static rules. Even if the network demand changes significantly, the system will not make predictions or make adjustments in advance.

[0169] In this traditional comparison example, the time slot resource allocation for each service is based on a pre-set fixed ratio. For example, the system reserves 30% of the time slot resources for video streaming services, allocates 20% of the time slot resources for voice services, and only 10% for file transfer services. These ratios are determined at system startup and are not dynamically adjusted during the entire operation process.

[0170] In this traditional comparison example, no prediction model (such as LSTM) is used to predict future traffic fluctuations. Therefore, the system cannot provide early warning and adjustments based on changes in network traffic or fluctuations in business demand. This will result in the system being unable to effectively handle sudden resource shortages during network peaks or when bursts of traffic occur.

[0171] Step 3: Static rule time slot resource allocation

[0172] In this traditional solution, the rules for time slot resource allocation are static and do not take into account real-time data feedback or dynamic changes.

[0173] In this traditional comparison example, the allocation of time slot resources is based only on the initial preset rules. For example, the video streaming service is always allocated 30% of the resources, regardless of whether there are more high-priority service demands in the current network. Low-priority services (such as file transfer) cannot get more time slot resources, even if the network load is low.

[0174] In this traditional comparison example, when the network load increases, the system has no dynamic adjustment mechanism, so high-priority services may experience increased delays or increased packet loss due to insufficient time slot resources. At the same time, low-priority services cannot obtain excess resource allocation based on the actual load of the current network.

[0175] Step 4: Time slot resource allocation without optimization algorithm

[0176] In the traditional comparison example, the system does not use any optimization algorithm (such as Q-learning), and the time slot allocation strategy completely relies on the fixed rules set initially, without any learning and improvement mechanism.

[0177] In this traditional comparison example, the system only allocates resources based on the initial priority rules and cannot dynamically adjust according to changes in business needs and network status feedback. For example, the system will not adjust time slot resources due to increased latency of certain services, nor will it optimize resource allocation due to frequent network congestion.

[0178] In this traditional comparison example, due to the lack of any learning mechanism, the system's time slot allocation strategy becomes rigid when the network environment changes and cannot adapt to the dynamic changes of different loads or business requirements. This leads to poor system resource utilization and user experience, especially under high load or burst traffic, the allocation efficiency of time slot resources is inefficient.

[0179] Step 5: Resource allocation without feedback mechanism

[0180] In this traditional solution, the system does not have a real-time feedback mechanism. After the allocated time slot resources are initially set, the system will not adjust them according to subsequent network status or service demand feedback.

[0181] In this traditional comparison example, the system does not monitor the usage of time slot resources, nor does it collect any real-time feedback data (such as bandwidth utilization, delay, and packet loss rate). Therefore, the system cannot evaluate the effect of resource allocation, nor can it make improvements to the existing allocation plan.

[0182] In this traditional comparison example, the allocation strategy of time slot resources remains unchanged, and the system will not respond even if the network load changes significantly (such as network congestion and increased packet loss). This allocation method without feedback mechanism is very inefficient in a dynamic network environment.

[0183]

[0184] Data analysis and summary

[0185] Average network delay: The traditional comparative example uses static rules for time slot resource allocation, which cannot be adjusted according to changes in network requirements, resulting in high average network delay. In contrast, the first embodiment significantly reduces delay through intelligent dynamic allocation.

[0186] Average packet loss rate: Since the traditional solution cannot respond to changes in network load in time, the packet loss rate is high (1.5%). However, the first embodiment uses the LSTM model and Q-learning algorithm to dynamically adjust the time slot allocation, reducing the packet loss rate to 0.8%.

[0187] Bandwidth utilization: The bandwidth utilization of the traditional comparative example is only 60%, while the first embodiment improves the bandwidth utilization to 70% by predicting future traffic and dynamically allocating resources.

[0188] Time slot resource allocation response time: The response time of the conventional comparative example is relatively long (150 milliseconds) because there is no real-time adjustment mechanism. However, the first embodiment can respond faster according to the network status, with a response time of 100 milliseconds.

[0189] Peak resource utilization: Traditional solutions cannot effectively manage resources during network peak periods, resulting in low resource utilization during peak periods (55%). Embodiment 1 allocates resources in advance through intelligent prediction, increasing resource utilization during peak periods to 65%.

[0190] User satisfaction score: Due to the high latency and packet loss caused by the traditional solution, the user experience is poor, with a score of only 2.8 points. Implementation Example 1 improves the user experience by optimizing the allocation strategy, with a score of 3.8 points.

[0191] High priority service delay: Traditional solutions cannot prioritize low delay for high priority services, resulting in high delay (60 milliseconds). Embodiment 1 can dynamically allocate more resources for high priority services, reducing the delay to 35 milliseconds.

[0192] Key business guarantee rate: In Example 1, the resource guarantee rate for high-priority business is 90%, which is significantly improved compared to 75% of the traditional solution, ensuring the service quality of key business under high load conditions.

[0193] Network congestion: As the traditional solution cannot be adjusted dynamically, the frequency of network congestion events is high (8 times / hour). Embodiment 1 reduces the number of congestion events to 5 times / hour through real-time monitoring and feedback mechanism.

[0194] Resource allocation flexibility score: The static allocation strategy of the traditional solution lacks flexibility and is scored 2.5 points. However, the first embodiment improves flexibility through an intelligent allocation strategy and is scored 3.5 points.

[0195] Through data comparison, it can be seen that Example 1 is significantly better than the traditional comparison examples in multiple key indicators, especially in terms of network delay, bandwidth utilization, packet loss rate, key business guarantee and user satisfaction. Traditional solutions show great limitations when facing dynamic network environments, while Example 1 uses intelligent prediction and dynamic adjustment mechanisms to better adapt to changes in network load and improve resource utilization efficiency and service quality.

[0196] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A service channel time slot allocation method, characterized in that: The following steps are involved: Step 1: Collect network data from network devices in real time, the network data including bandwidth utilization, delay, packet loss rate and service traffic type; Step 2: Predict future network traffic fluctuations through a long short-term memory network model based on historical traffic data and real-time network data; Step 3: According to the network traffic fluctuation prediction result, dynamically adjust the allocation of time slot resources, and give priority to allocating time slot resources to high priority services; Step 4: Use the Q-learning reinforcement learning algorithm to optimize the time slot allocation strategy based on real-time feedback of network performance data; Step 5: Based on the feedback mechanism, dynamically adjust the time slot resources allocated to different service channels.

2. The service channel time slot allocation method according to claim 1, characterized in that: The network data collection step comprises: Using SNMP protocol to communicate with network devices, including base stations, routers and switches; Obtaining real-time performance data of network devices through the configuration management information base, the performance data including bandwidth utilization, delay and packet loss rate; The traffic in-depth analysis tool is used to classify the business traffic, and the business traffic types include multimedia video streaming with high bandwidth requirements, voice calls with low latency requirements, and data transmission that is not sensitive to latency.

3. The service channel time slot allocation method according to claim 1, characterized in that: The training steps of the long short-term memory network model include: Use the historical traffic data of the past hour as the input of the model, the historical data includes the bandwidth occupancy, delay fluctuation and packet loss rate of different business traffic; Calibrate the model based on the current real-time collected network data, wherein the real-time data includes the number of user accesses, service traffic type and bandwidth utilization; The model's hyperparameters, including learning rate, number of hidden layer units, and number of training iterations, were adjusted by cross-validation to optimize the model's prediction accuracy.

4. The service channel time slot allocation method according to claim 1, characterized in that: The network traffic fluctuation prediction step based on the long short-term memory network model includes: The model outputs the network traffic forecast results for the next 30 minutes, including the expected peak and valley traffic periods; The model weights and parameters are continuously modified through a feedback control mechanism, and the modification is based on the error value between the actual network load data and the predicted results.

5. The service channel time slot allocation method according to claim 1, characterized in that: The state space of the Q-learning reinforcement learning algorithm includes: Current network load level, bandwidth utilization, service priority, time slot resource utilization, and real-time feedback of network delay and packet loss rate; The reinforcement learning algorithm optimizes the time slot allocation strategy by updating the Q value, and the Q value is updated based on the feedback information obtained after allocating the time slot resources.

6. The service channel time slot allocation method according to claim 1, characterized in that: The real-time feedback adjustment of the time slot resource allocation strategy includes: When the network load increases, dynamically reduce the time slot resources allocated to low-priority service channels; Automatically increase time slot allocation for high priority services based on real-time network feedback data, including current delay and packet loss rate; When the time slot resources of the low-priority service channel are not fully used, the unused time slot resources are automatically recovered and reallocated to the current high-load service.

7. The service channel time slot allocation method according to claim 1, characterized in that: The feedback data collection frequency of the time slot resource allocation is once per second, and the feedback data includes: The time slot resource utilization rate of each service channel, the actual bandwidth utilization rate, the transmission delay and packet loss rate of the current network; The feedback mechanism determines whether the service quality of each business channel meets the preset standard. If not, the system will automatically adjust the allocated time slot resources.

8. The service channel time slot allocation method according to claim 1, characterized in that: The system automatically adjusts the time slot allocation strategy according to different network environments, including: In mobile communication networks, the time slot resources allocated to each base station are dynamically adjusted based on the user's mobility and base station load; In satellite communication networks, strict flow control and time slot reservation strategies are used to ensure that key services such as navigation and telemedicine have priority in obtaining time slot resources; In a fiber-optic communication network, time slot resources are dynamically allocated based on the number of concurrent requests and bandwidth requirements of users.

9. The service channel time slot allocation method according to claim 1, characterized in that: The feedback mechanism further comprises: Compare the traffic fluctuations output by the prediction model with the actual network status. When the error exceeds the preset threshold, the system automatically adjusts the weights and parameters of the long short-term memory network model. By iteratively optimizing the accuracy of the prediction model, we can ensure that the prediction results of future traffic fluctuations are closer to the actual network status.

10. The service channel time slot allocation method according to claim 1, characterized in that: The time slot resource allocation is performed according to the service priority, and the priority strategy includes: High-priority service channels include real-time video calls, telemedicine, and navigation services, while low-priority service channels include file transfers and general data browsing; During high-load periods, time slot resources are allocated preferentially to high-priority services to ensure that their delays and service quality are not affected, while reclaiming time slot resources that are not fully used by low-priority services.

Images