E-sports broadband access optimization method and system based on computing network fusion campus scene

Abstract

The invention relates to the field of network optimization and resource scheduling, and discloses a method and a system for realizing E-sports broadband access optimization based on a campus scene integrated with a computing network, and the method comprises the following steps: step 1, collecting real-time state information in a campus network, the state information comprising user bandwidth requirements, path time delay and link load data; step 2, constructing a multi-objective optimization model, wherein objectives of the model include minimizing user path time delay and deviation between bandwidth allocation and ideal requirements; step 3, solving the multi-objective optimization model by using a Lagrange multiplier method to obtain a bandwidth allocation result of the user; and step 4, constructing a differential game model based on the user utility function, and solving a resource allocation strategy between users by using a dynamic game method. By adopting a multi-objective optimization scheme based on the Lagrange multiplier method and the differential game model, the technical effect of dynamically allocating the bandwidth and the path in a multi-user scene is achieved.

Description

Technical Field

[0001] The present invention relates to the field of network optimization and resource scheduling, and specifically to a method and system for optimizing e-sports broadband access based on a computing-network integrated campus scenario. Background Art

[0002] With the rapid development of the e-sports industry, the demand for e-sports broadband access in campus scenarios has increased significantly. E-sports services usually place higher requirements on network bandwidth, latency, and stability. Especially in scenarios where multiple people interact in real time, low latency and high bandwidth guarantees become key factors affecting user experience. At the same time, there are many users in campus networks with diverse needs, and the network behaviors of different users have a direct impact on resource allocation and performance. Therefore, designing an efficient broadband access optimization method for e-sports services is an important research direction in current campus network scenarios.

[0003] Among the existing network optimization technologies, most methods use static resource allocation or single-objective optimization strategies, usually focusing only on optimizing bandwidth utilization or path latency. However, these methods often ignore the impact of competition between users on resource allocation and lack the ability to adapt to dynamic changes in user demand. In addition, traditional path selection and bandwidth allocation are mostly processed independently, and there is no effective synergy between the two. These technical limitations make it difficult for existing solutions to meet the complex and diverse network requirements in e-sports broadband access scenarios, especially during peak user demand periods, when network performance is prone to fluctuations.

[0004] In the prior art, due to the lack of a dynamic response mechanism, it is impossible to adjust the resource allocation strategy in time when the network status changes, resulting in the optimization effect of bandwidth allocation and path selection failing to take into account both fairness and timeliness. Especially in e-sports scenarios, this static strategy may cause resource scramble or path congestion among users, further reducing the user experience quality. Therefore, how to achieve efficient collaborative optimization of bandwidth and paths in a dynamic environment, and balance fairness and efficiency in resource competition, is a problem that has not been effectively solved in the prior art, and is also the core technical problem to be solved by the present invention. Summary of the invention

[0005] In view of the shortcomings of the existing technology, the present invention provides a method and system for optimizing e-sports broadband access based on a computing-network integrated campus scenario, which solves the problem that the existing technology lacks a dynamic response mechanism and cannot adjust the resource allocation strategy in time when the network status changes, resulting in the optimization effect of bandwidth allocation and path selection failing to take into account both fairness and timeliness.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a method for optimizing e-sports broadband access based on a computing-network-integrated campus scenario, the method comprising the following steps:

[0007] Step 1: Collect real-time status information in the campus network, the status information includes user bandwidth requirements, path delay and link load data;

[0008] Step 2: construct a multi-objective optimization model, wherein the objectives of the model include: minimizing the deviation between user path delay and bandwidth allocation and ideal demand;

[0009] Step 3: Use the Lagrange multiplier method to solve the multi-objective optimization model to obtain the bandwidth allocation result of the user;

[0010] Step 4: Construct a differential game model based on the user utility function, use the dynamic game method to solve the resource allocation strategy among users, and generate the user's path selection and bandwidth allocation optimization plan;

[0011] Step 5: Send the optimization calculation results to the network devices through the software-defined network controller to adjust the user's path and bandwidth allocation;

[0012] Step 6: Based on the feedback mechanism, collect the user's actual bandwidth allocation and path delay data, update the network status information and repeat the optimization process.

[0013] Preferably, the constraints of the multi-objective optimization model in step 2 include:

[0014] S1. The total bandwidth allocation of users shall not exceed the total bandwidth limit of the system;

[0015] S2. The bandwidth allocated to the user shall not be lower than the preset minimum bandwidth requirement;

[0016] S3. The delay of the user path shall not exceed the preset maximum delay threshold.

[0017] Preferably, the solution process of the Lagrange multiplier method in step 3 includes:

[0018] S1. Construct a Lagrangian function and use path delay, bandwidth allocation deviation and constraint conditions as a combination expression of the optimization target;

[0019] S2, iteratively optimizing user bandwidth allocation using a gradient descent method;

[0020] S3. Adjust the Lagrange multiplier until the optimization result meets the convergence condition.

[0021] Preferably, the user utility function of the differential game model in step 4 includes:

[0022] S1, negative effect of user path delay;

[0023] S2, the positive impact of users’ actual allocated bandwidth on utility;

[0024] S3. The negative impact of resource allocation competition among users on utility.

[0025] Preferably, the dynamic game solving process in step 4 includes:

[0026] S1. Define the dynamic change equation of user bandwidth allocation and path selection based on user utility function;

[0027] S2. Solve the game equilibrium point through numerical iteration and generate the user's optimal resource allocation strategy.

[0028] Preferably, the optimization result in step 5 is sent to the network device through the software defined network controller, including the following operations:

[0029] S1, adjust user path selection and generate the optimal routing strategy based on the remaining bandwidth and load of the link;

[0030] S2. Configure the service priority queues in the network device to allocate higher bandwidth resources to high-priority users.

[0031] Preferably, the feedback data in step six is ​​analyzed by a time series prediction model to predict changes in network status in the next time period, and the time series prediction model includes an ARIMA model or an LSTM model.

[0032] The present invention also provides a system for optimizing e-sports broadband access based on a computing-network integration campus scenario, the system comprising:

[0033] The network status collection module is used to collect user bandwidth requirements, path delay and link load data;

[0034] The optimization calculation module is used to build a multi-objective optimization model based on network status data and solve the user bandwidth allocation through the Lagrange multiplier method;

[0035] The differential game module is used to solve the dynamic game of resource allocation among users based on the user utility function and generate user path selection and bandwidth allocation strategies;

[0036] The resource scheduling module is used to send the optimization calculation results to the network devices through the software-defined network controller to adjust the user's path selection and bandwidth allocation;

[0037] The feedback monitoring module is used to collect users' actual bandwidth allocation and path delay data, and update network status information to support optimization iteration.

[0038] Preferably, the network status collection module includes a distributed network monitoring node and a data aggregation unit, which is used to collect and summarize the user's network status information in real time.

[0039] Preferably, the optimization calculation module solves the multi-objective optimization model based on the KKT condition, and calculates the optimization result of the differential game module in combination with the Nash equilibrium theory.

[0040] The present invention provides a method and system for optimizing e-sports broadband access based on a computing-network-integrated campus scenario. It has the following beneficial effects:

[0041] 1. The present invention adopts a multi-objective optimization solution based on the Lagrange multiplier method and differential game model, achieving the technical effect of dynamically allocating bandwidth and paths in multi-user scenarios. Compared with the technical solutions of single-objective optimization or static allocation in the prior art, it solves the problem of not being able to balance latency minimization and bandwidth fair allocation, thereby improving the user experience and resource utilization of e-sports broadband access.

[0042] 2. The present invention realizes real-time monitoring and dynamic adjustment of network status by introducing a design combining SDN controller and feedback mechanism, and achieves the effect of maintaining the stability of system resource allocation when the network status changes. Compared with the traditional allocation method that relies on fixed strategies, the present invention solves the problem of network performance fluctuation caused by lack of dynamic response capability, and significantly enhances the robustness and adaptability of the system.

[0043] 3. The present invention achieves a dynamic balance under user resource competition by constructing a utility function that includes the effects of user latency, bandwidth allocation, and resource competition, combined with dynamic game theory. Compared with the allocation method in the prior art that ignores the competitive relationship between users, it solves the problem of unfair allocation caused by resource preemption, especially in the e-sports scenario where bandwidth resources are tight.

[0044] 4. The present invention adopts the collaborative working mechanism of the path optimization module and the bandwidth scheduling module, and realizes the joint optimization effect of bandwidth allocation and path selection. Compared with the method of independently processing path planning and bandwidth allocation in the prior art, the problem of low resource allocation efficiency caused by the inability to coordinate optimization between the two is solved, thereby significantly improving the overall service quality and user satisfaction in a complex network environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 is a flow chart of the method of the present invention;

[0046] Figure 2 It is a system flow chart of the present invention. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the specification 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.

[0048] Please see attached Figure 1 The embodiment of the present invention provides a method for optimizing e-sports broadband access based on a computing-network integration campus scenario, and the method includes the following steps:

[0049] Step 1: Collect real-time status information in the campus network, including user bandwidth requirements, path delay, and link load data;

[0050] Step 2: Build a multi-objective optimization model. The objectives of the model include: minimizing the deviation between user path delay and bandwidth allocation and ideal demand;

[0051] Step 3: Use the Lagrange multiplier method to solve the multi-objective optimization model to obtain the bandwidth allocation result of the user;

[0052] Step 4: Construct a differential game model based on the user utility function, use the dynamic game method to solve the resource allocation strategy among users, and generate the user's path selection and bandwidth allocation optimization plan;

[0053] Step 5: Send the optimization calculation results to the network devices through the software-defined network controller to adjust the user's path and bandwidth allocation;

[0054] Step 6: Based on the feedback mechanism, collect the user's actual bandwidth allocation and path delay data, update the network status information and repeat the optimization process.

[0055] The constraints of the multi-objective optimization model in step 2 include:

[0056] S1. The total bandwidth allocation of users shall not exceed the total bandwidth limit of the system;

[0057] S2. The bandwidth allocated to the user shall not be lower than the preset minimum bandwidth requirement;

[0058] S3. The delay of the user path shall not exceed the preset maximum delay threshold.

[0059] The solution process of the Lagrange multiplier method in step 3 includes:

[0060] S1. Construct a Lagrangian function and use path delay, bandwidth allocation deviation and constraint conditions as a combination expression of the optimization target;

[0061] S2, iteratively optimizing user bandwidth allocation using a gradient descent method;

[0062] S3. Adjust the Lagrange multiplier until the optimization result meets the convergence condition.

[0063] The user utility function of the differential game model in step 4 includes:

[0064] S1, negative effect of user path delay;

[0065] S2, the positive impact of users’ actual allocated bandwidth on utility;

[0066] S3. The negative impact of resource allocation competition among users on utility.

[0067] The dynamic game solution process in step 4 includes:

[0068] S1. Define the dynamic change equation of user bandwidth allocation and path selection based on user utility function;

[0069] S2. Solve the game equilibrium point through numerical iteration and generate the user's optimal resource allocation strategy.

[0070] The optimization results in step 5 are sent to the network devices through the software-defined network controller, including the following operations:

[0071] S1, adjust user path selection and generate the optimal routing strategy based on the remaining bandwidth and load of the link;

[0072] S2. Configure the service priority queues in the network device to allocate higher bandwidth resources to high-priority users.

[0073] In step six, the feedback data is analyzed by a time series prediction model to predict changes in network status in the next period. The time series prediction model includes an ARIMA model or an LSTM model.

[0074] Please see attached Figure 2 The present invention also provides an e-sports broadband access optimization system based on a computing-network integration campus scenario, the system comprising:

[0075] The network status collection module is used to collect user bandwidth requirements, path delay and link load data;

[0076] The optimization calculation module is used to build a multi-objective optimization model based on network status data and solve the user bandwidth allocation through the Lagrange multiplier method;

[0077] The differential game module is used to solve the dynamic game of resource allocation among users based on the user utility function and generate user path selection and bandwidth allocation strategies;

[0078] The resource scheduling module is used to send the optimization calculation results to the network devices through the software-defined network controller to adjust the user's path selection and bandwidth allocation;

[0079] The feedback monitoring module is used to collect users' actual bandwidth allocation and path delay data, and update network status information to support optimization iteration.

[0080] The network status collection module includes distributed network monitoring nodes and data aggregation units, which are used to collect and summarize users' network status information in real time.

[0081] The optimization calculation module solves the multi-objective optimization model based on the KKT condition, and calculates the optimization results of the differential game module in combination with the Nash equilibrium theory.

[0082] In the technical implementation process of the present invention, the collection of network status information is the basis of the entire method. In order to ensure the accuracy of the optimization calculation, it is necessary to obtain the dynamic operation status of the campus network through real-time monitoring means, including the user's bandwidth demand, path delay and link load. This information directly affects the subsequent optimization model construction and the effectiveness of the optimization results.

[0083] Generally, the user demand and link status of campus networks will change over time, especially during peak hours or in case of burst traffic, the network status will be more complicated. Therefore, the collection of status information must be real-time and accurate, and at the same time, it is necessary to have a certain predictive ability to cope with possible changes in the future. As an option, the present invention realizes the real-time collection and processing of network status by combining distributed monitoring nodes with a centralized data processing platform.

[0084] In this embodiment, distributed monitoring devices are deployed at each major node in the campus network to collect status information. These monitoring devices can be switches and routers that support data collection functions, or they can be specially deployed probe devices. Specifically, the collected information includes the following:

[0085] In some embodiments, the collection of user bandwidth requirements can be achieved by analyzing the data packets sent by the user device. Generally, network devices can identify the types of requests and traffic requirements sent by users through deep packet inspection (DPI) technology. For example, when a user accesses a cloud e-sports platform, the DPI module extracts traffic characteristics, including the size, frequency, and bandwidth usage of game data packets.

[0086] As a possible implementation, the user's ideal bandwidth requirement is It can be calculated by the following formula:

[0087]

[0088] in:

[0089] represents the ideal bandwidth requirement of user i;

[0090] R i : The amount of data requested by the user (unit: bytes);

[0091] P i : User interaction frequency (unit: times / second);

[0092] T i : Expected response time of the user (in seconds).

[0093] Specifically, in e-sports scenarios, users send high-frequency traffic requests and have strict requirements on response time. The above formula can accurately calculate the user's ideal bandwidth requirements.

[0094] Path delay

[0095] The collection of path delay is crucial for the optimization of e-sports services. In some embodiments, the path delay can be measured by periodically sending probe packets. Generally, network devices calculate the path delay d by the round-trip time of the probe packet. i As an alternative, the following formula can be used:

[0096]

[0097] d i : represents the path delay of user i;

[0098] t r : The timestamp returned by the probe packet;

[0099] t s : The timestamp when the probe packet is sent.

[0100] In a possible implementation, the probe packet can be sent from the user terminal to the target server, and all switches and routers in the path record the forwarding time. This can further accurately measure the delay distribution in the path and identify high-delay nodes.

[0101] Link load is an important reference indicator for network optimization and is used to measure the usage of links. In some embodiments, link load L k It can be calculated in real time through the traffic statistics module of the network device. The specific formula is:

[0102]

[0103] in:

[0104] Lk : represents the load rate of link k;

[0105] The current used bandwidth of link k (unit: bps);

[0106] The total bandwidth capacity of link k (in bps).

[0107] In a possible implementation, link load data is reported to a central control platform in real time through distributed monitoring nodes. Specifically, when the link load approaches 100%, the system will issue an alarm to trigger resource reallocation.

[0108] In this embodiment, after the collected real-time status information is aggregated to the central control platform, it needs to be further processed and predicted. Specifically, the system uses a time series prediction model (such as LSTM or ARIMA model) to predict future changes in bandwidth demand, latency, and load. As a possible implementation method, the prediction formula can be expressed as:

[0109] X t+1 =f(X t ,X t-1 ,…,X t-n )

[0110] in:

[0111] X t+1 : Indicates the state variables at the next moment (such as bandwidth requirements);

[0112] X t ,X t-1 ,…,X t-n : Represents the state variables of the current and past n moments;

[0113] f is the mapping function of the time series forecasting model.

[0114] In a specific implementation, the system performs a prediction calculation every 1 second and adjusts subsequent optimization parameters according to the prediction results.

[0115] Through the above steps, the real-time status information in the campus network is completely collected and processed, providing sufficient data support for subsequent optimization calculations. The collection method of the present invention realizes comprehensive perception and dynamic prediction of network status by combining distributed monitoring and centralized prediction.

[0116] In some possible extended embodiments, the acquisition module of the present invention can be further extended to support anomaly detection functions. For example, by analyzing abnormal changes in path delay and link load, the system can detect possible network congestion or failures in advance and trigger corresponding protection mechanisms. This expansion capability can significantly improve the robustness and reliability of the system.

[0117]

[0118] in:

[0119] B i The bandwidth allocated to the i-th user, in bps, is an optimization variable;

[0120] d i is the path delay of the i-th user, in seconds, and is an optimization variable;

[0121] is the ideal bandwidth requirement of the ith user, in bps, which is a known quantity;

[0122] α is the weight coefficient of path delay in the optimization objective, indicating the importance of path delay;

[0123] β is the weight coefficient of bandwidth deviation in the optimization objective, indicating the priority of bandwidth allocation fairness;

[0124] N is the total number of users in the network.

[0125] In a possible implementation, the values ​​of α and β can be adjusted according to the priority of the actual scenario. For example, in an e-sports competition, the weight of α can be increased to prioritize reducing path delay, while in a general user scenario, β can be appropriately increased to make the allocation more fair.

[0126] The present invention also sets multiple constraints for the optimization target to ensure the feasibility of the optimization result in practical applications. Generally, the constraints include the following:

[0127] The total bandwidth constraint limits the total bandwidth allocated to all users to not exceed the total bandwidth of the system. Its expression is:

[0128]

[0129] in:

[0130] B total Indicates the total bandwidth resources available to the system, in bps, which is a constant.

[0131] B i Defined as the bandwidth allocated to the ith user, in bps.

[0132] The minimum bandwidth requirement constraint is used to ensure that each user is allocated at least the basic bandwidth to meet the most basic service quality requirements. Its expression is:

[0133]

[0134] in:

[0135] B i Defined as the bandwidth allocated to the ith user, in bps.

[0136] B min : Defined as the minimum bandwidth requirement of a single user, in bps, which is the lower limit preset by the system.

[0137] The path delay constraint is used to limit the maximum path delay of the user to ensure the low delay requirement in the e-sports scenario. Its expression is:

[0138]

[0139] in:

[0140] d i It is defined as the path delay of the ith user in seconds.

[0141] d max : Defined as the maximum path delay acceptable to users, in seconds, which is the upper limit preset by the system.

[0142] Link load constraints:

[0143] Link load constraints are used to limit the bandwidth usage of a single link in the network to prevent individual links from affecting overall performance due to overload. As an option, the constraint can be expressed as:

[0144]

[0145] in:

[0146] L k is the load rate of link k;

[0147] is the current bandwidth usage of link k, in bps;

[0148] is the total bandwidth capacity of link k, in bps;

[0149] L max is the maximum allowable load rate of the link.

[0150] In some embodiments, L can be dynamically adjusted based on historical data. maxFor example, during peak hours, this value can be appropriately increased to improve the overall utilization of the system.

[0151] In order to enhance the adaptability of the model, the present invention also introduces a predictive parameter adjustment mechanism. Specifically, in the optimization model, the time series prediction results can be used to adjust the and d i Dynamic adjustments are made. For example, the user's ideal bandwidth requirement It can be predicted by the following formula:

[0152]

[0153] in:

[0154] The ideal bandwidth requirement for the next predicted period;

[0155] is the ideal bandwidth requirement for the current period;

[0156] f is the prediction function, which can be implemented using a linear regression model or a deep learning model.

[0157] In this embodiment, after the optimization model is constructed, it will provide clear objective functions and constraints for the optimization solution in the subsequent steps. This model based on multi-objective optimization can effectively adapt to different user scenarios and network conditions, and improve the efficiency and fairness of resource allocation. By combining real-time data with predicted data, the optimization model of the present invention has significant advantages in dynamics and robustness.

[0158] After the multi-objective optimization model is constructed, the next task is to solve the optimization model. The present invention introduces the Lagrange multiplier method to organically combine the optimization objectives with the constraints to form an optimization equation set that can be solved by mathematical methods. In general, the Lagrange multiplier method is applicable to multi-constraint and multi-variable optimization problems, especially to scenarios with clear constraints such as resource allocation and path optimization.

[0159] As an option, the present invention adopts an improved iterative optimization algorithm, by deriving the gradient of the Lagrangian function, gradually adjusting the variable value and the multiplier coefficient until the convergence requirements of the optimization objective and the constraint conditions are met. Specifically, the solution process of this embodiment includes the construction of the Lagrangian function, the solution of the partial derivatives, and the implementation of the iterative convergence mechanism.

[0160] First, a Lagrangian function is constructed for the multi-objective optimization model. In order to combine the objective functions of path delay and bandwidth allocation deviation with the constraints, the present invention introduces Lagrangian multipliers λ and μi to express the optimization problem in the following form:

[0161]

[0162] in:

[0163] is the Lagrangian function;

[0164] B i The bandwidth allocation variable for the i-th user, in bps;

[0165] d i is the path delay variable of the i-th user, in seconds;

[0166] is the ideal bandwidth requirement of the ith user, in bps;

[0167] B total is the total available bandwidth resource of the system, in bps;

[0168] B min The minimum bandwidth requirement for a single user, in bps;

[0169] λ is the Lagrange multiplier of the total bandwidth constraint;

[0170] μ i is the Lagrange multiplier of the minimum bandwidth requirement constraint, which indicates the constraint strength corresponding to each user;

[0171] α and β are the weight coefficients of path delay and bandwidth deviation targets, respectively, which determine their importance in the optimization target;

[0172] N is the total number of users.

[0173] Secondly, by solving the partial derivatives of the Lagrangian function, the necessary conditions for the optimization problem are obtained. In this embodiment, the solution of the partial derivatives mainly includes the following parts:

[0174] Secondly, by solving the partial derivatives of the Lagrangian function, the necessary conditions for the optimization problem are obtained. In this embodiment, the solution of the partial derivatives mainly includes the following parts:

[0175]

[0176] sgn(x) is the sign function, which takes the value 1 when x>0, -1 when x<0, and 0 when x=0.

[0177] Find the partial derivative with respect to λ:

[0178]

[0179] For μ i Find partial derivatives:

[0180]

[0181] By solving the above partial derivatives, the necessary conditions for the optimization problem can be obtained, that is, the solution that satisfies the optimization objective when the partial derivative is zero.

[0182] In a possible implementation, the partial derivatives are numerically solved using a gradient descent method.

[0183] Specifically, the update formula of the gradient descent method is:

[0184]

[0185] in:

[0186] represents the bandwidth allocation value of the i-th user at the t-th iteration;

[0187] λ (t) represents the Lagrange multiplier at the tth iteration;

[0188] represents the minimum bandwidth constraint multiplier of the i-th user at the t-th iteration;

[0189] η is the learning rate, which represents the update step size at each iteration.

[0190] Through multiple iterations, when and When the absolute values ​​of are all less than the preset threshold, the algorithm stops and the final bandwidth allocation result is obtained.

[0191] In addition, the present invention ensures that the optimization process can be completed within a limited number of iterations through a dynamic convergence determination mechanism.

[0192] In some embodiments, the convergence condition may be defined as:

[0193]

[0194] in:

[0195] ∈ is the convergence threshold, usually a small positive number (such as 10 -6 ).

[0196] When the convergence condition is met, the iteration stops and the optimization results are output.

[0197] In this embodiment, the result of the differential game model is the optimal bandwidth allocation and path selection strategy for the user.

[0198] These results will be used in subsequent steps and sent to network devices by the software-defined network controller (SDN) for implementation. By constructing and solving the differential game model, the present invention can effectively solve the resource competition problem between users, achieve a dynamic balance between bandwidth allocation and path optimization, and provide an efficient and fair solution for e-sports broadband access.

[0199] After completing the preliminary solution of bandwidth allocation and path optimization, it is further necessary to solve the fairness and efficiency of resource allocation among users. The present invention constructs a user utility function and introduces a differential game model to achieve dynamic balance and optimization in the bandwidth competition among users. In general, users' resource requirements and network behaviors are not only driven by their own needs, but also affected by the behaviors of other users. Therefore, the modeling method based on game theory can effectively solve the problem of resource competition, especially in the optimization of e-sports broadband access, where the bandwidth competition between different users has a significant impact.

[0200] As an option, the present invention quantifies the demand and competition degree of each user in resource allocation by designing a user utility function. Then, the differential game model is used to simulate the dynamic evolution of user behavior, and the balance point of resource allocation between users is found through iterative solution, thereby obtaining a joint solution for bandwidth allocation and path optimization.

[0201] In this embodiment, the specific definition of the user utility function is as follows:

[0202] The utility function is used to describe the resource allocation effect of users. Its goal is to maximize the overall satisfaction of each user by dynamically adjusting the user bandwidth and path selection strategy. The user utility function is defined as:

[0203]

[0204] in:

[0205] u i (B i ,d i ): utility value of the i-th user;

[0206] B i : The bandwidth allocation value of the i-th user, in bps;

[0207] d i : path delay of the ith user, in seconds;

[0208] B j : The bandwidth allocation value of the jth user, in bps;

[0209] γ i : The coefficient of user i’s sensitivity to path delay, which is used to indicate the negative impact of delay on utility;

[0210] δ i : The sensitivity coefficient of user i to bandwidth demand, which is used to indicate the positive impact of bandwidth allocation on utility;

[0211] η: Resource competition weight coefficient, used to describe the impact of other users’ bandwidth allocation on the utility of user i.

[0212] Generally speaking, path delay has a negative impact on user utility, so it is presented as a negative value in the utility function. Bandwidth allocation has a positive contribution to user utility, so it is presented as a positive value. However, bandwidth competition between users has a negative impact, which is quantified by the last term.

[0213] In one possible implementation, the user's behavior is modeled as a dynamic game process.

[0214] Specifically, the evolution of user behavior is regarded as a dynamic adjustment process, whose goal is to gradually adjust its own strategy (bandwidth allocation and path selection) to maximize its own utility value u i (B i ,d i ). This process can be expressed by the following differential equation:

[0215]

[0216] in:

[0217] Bandwidth allocation change rate of the i-th user;

[0218] The path delay variation rate of the i-th user;

[0219] The partial derivative of user i’s utility function with respect to path delay;

[0220] κ: speed factor of user policy adjustment.

[0221] Through the above differential equations, each user will gradually adjust the bandwidth allocation and path selection strategy until the entire system reaches a dynamic balance.

[0222] In this embodiment, the equilibrium point of the differential game model is obtained by numerical calculation.

[0223] In some embodiments, the game equilibrium point is the Nash equilibrium point, which means that under this state, any single user's strategy adjustment will not further improve its own utility value. The equilibrium condition is:

[0224]

[0225] When the game equilibrium point is reached, the bandwidth allocation and path selection of user i satisfy:

[0226]

[0227] in:

[0228] The optimal bandwidth allocation value for user i at the equilibrium point;

[0229] The optimal path selection value of user i at the equilibrium point.

[0230] By solving the above optimization equations, we can obtain the bandwidth allocation and path selection strategy for each user.

[0231] In some embodiments, the present invention discretizes the iterative process of the differential game model.

[0232] In order to facilitate numerical calculations, discrete time steps are used to iterate the dynamic evolution of user strategies. The discretization update formula is:

[0233]

[0234] in:

[0235] and Bandwidth allocation and path selection value of user i at the tth iteration;

[0236] Δt: time step, in seconds;

[0237] and are the partial derivatives of the user utility function at the tth iteration respectively.

[0238] Through multiple iterations, the user strategy will gradually converge to the equilibrium point.

[0239] The differential game model result in this embodiment is used for dynamic adjustment of bandwidth allocation.

[0240] After obtaining the equilibrium result of the differential game model, the final user bandwidth allocation and path selection strategy will be used as the input of the software defined network controller (SDN) in the subsequent steps to guide the specific network scheduling operation. Through the differential game model, the present invention can achieve fair bandwidth allocation and balanced path selection among users, effectively solving the competition problem among users.

[0241] This method is particularly suitable for campus e-sports scenarios where user demands change rapidly and bandwidth resources are limited, and provides strong support for achieving low-latency and high-fairness broadband access optimization.

[0242] After solving the differential game model and obtaining the user's optimal bandwidth allocation and path selection strategy, these optimization results need to be delivered to the network device to perform the actual network resource adjustment operation. The present invention implements the distribution of optimization results through a software-defined network (SDN) controller, and dynamically adjusts the routing strategy and bandwidth allocation scheme in the network by using its global network view and centralized scheduling capabilities. In general, the SDN controller communicates with network devices such as switches and routers through the southbound interface to achieve refined control of traffic forwarding paths and service queues.

[0243] As an option, the present invention combines the path optimization module with the bandwidth scheduling module to design a dynamic, hierarchical resource allocation strategy that can not only quickly respond to changes in user needs, but also achieve reasonable resource reallocation when the network status fluctuates.

[0244] Generally, the goal of path optimization is to select a forwarding path with the lowest latency and balanced link load for the user. The present invention integrates a path optimization module in the SDN controller, recalculates the routing path based on the results of the differential game model, and generates an optimized routing table.

[0245] Specifically, the present invention combines link load data and path delay data to calculate the comprehensive performance index of each path through the following formula:

[0246]

[0247] in:

[0248] P k : Comprehensive performance index of path k, the smaller the value, the better the path;

[0249] d e : The delay of link e in path k, in seconds;

[0250] L e : Load rate of link e (ratio of used bandwidth to total bandwidth, unitless);

[0251] L th : Link load threshold, usually set to 0.80.80.8 or 0.90.90.9, to avoid high-load links.

[0252] As an option, the present invention uses the Dijkstra algorithm combined with the above path performance index to generate the optimal path from the source node to the target node. In order to ensure the stability of path switching, a path switching threshold can be set. current Significantly higher than the alternative path performance index P new The switch is performed only when:

[0253] P current -P new >ΔP

[0254] in:

[0255] ΔP: path switching threshold, usually a small positive number (such as 0.05).

[0256] In a possible implementation, after generating a new routing table, the path optimization module updates the routing table to the network device through the southbound interface (such as the OpenFlow protocol) of the SDN controller to ensure that the traffic is forwarded along the optimized path.

[0257] The present invention realizes dynamic adjustment of user bandwidth allocation through a bandwidth scheduling module. Specifically, the bandwidth scheduling module adjusts the bandwidth allocation value B of each user according to the optimization result. i , by configuring the priority queue or rate limiting rules of the network device, ensure that the actual bandwidth obtained by the user is consistent with the optimized value.

[0258] Generally speaking, the implementation of the bandwidth scheduling module includes the following:

[0259] Configure a priority queue (PQ) for each user to ensure that high-priority users (such as e-sports users) are allocated higher service bandwidth.

[0260] Set the maximum available bandwidth value B for each user max,i To prevent some users from taking up too much resources. The bandwidth limit rule can be determined by the following formula:

[0261]

[0262] in:

[0263] B max,i : The maximum available bandwidth of user i, in bps;

[0264] B i : The optimized bandwidth allocation value of user i, in bps;

[0265] B total : The total available bandwidth of the system, in bps;

[0266] B j : The optimized bandwidth allocation value for other users, in bps.

[0267] In some embodiments, the present invention further designs a dynamic bandwidth allocation strategy. Specifically, when the user's actual bandwidth usage value Below the optimal value B iWhen the system is in a state of low bandwidth, it will temporarily allocate excess bandwidth to other users; when user demand increases, bandwidth will be recycled first to meet the optimal allocation result.

[0268] In order to cope with the real-time changes in user demand or network status, the present invention introduces a feedback mechanism and designs a dynamic adjustment strategy based on actual data. This mechanism monitors the user's actual bandwidth usage and path delay data in real time, and dynamically adjusts the network configuration when deviations from the optimization results are found.

[0269] Specifically, the execution process of the dynamic adjustment mechanism includes the following steps:

[0270] Collect users' actual bandwidth usage data and path delay data and the optimization result B i and d i Make a comparison.

[0271] When the deviation exceeds the preset threshold ∈ B or ∈ d When , the adjustment operation is triggered. The adjustment amount is determined by the following formula:

[0272]

[0273] in:

[0274] ΔB i : The bandwidth adjustment amount of user i, in bps;

[0275] Δd i : The path delay adjustment of user i, in seconds;

[0276] η and κ: adjustment rate factors, used to control the adjustment amplitude.

[0277] The dynamically adjusted results are sent to network devices in real time through the SDN controller, thus ensuring the consistency of resource allocation and optimization results.

[0278] In this embodiment, the path optimization and bandwidth scheduling modules complement each other and jointly achieve efficient execution of the optimization results.

[0279] The optimization results are sent down through the SDN controller, the path adjustment ensures that user traffic passes through the path with the lowest latency, and the bandwidth allocation module ensures that users get fair resource allocation. The dynamic adjustment mechanism further improves the system's adaptability in complex network environments.

[0280] Through the above method, the present invention can provide low-latency, high-reliability network services in campus e-sports broadband access scenarios, and maintain stable resource allocation efficiency and user experience under network load fluctuations.

[0281] After the optimization results are sent and executed, the present invention further introduces a feedback mechanism to monitor the actual bandwidth allocation and path delay of users in real time, analyze the network operation status, and dynamically adjust the network optimization model based on the feedback data. In general, due to the dynamic characteristics of network status and user needs, it is difficult to adapt to complex scene changes by relying solely on a single optimization. Therefore, the present invention designs a closed-loop feedback control process to achieve continuous optimization of network status through a feedback mechanism.

[0282] As an option, the feedback mechanism compares the collected real-time data with the optimization results, identifies possible deviations and triggers corresponding adjustment actions. By updating the network status information and re-executing the optimization steps, the robustness and dynamic adaptability of the system can be effectively guaranteed.

[0283] Generally speaking, the user's actual bandwidth usage and path delay data are the key input data of the feedback mechanism. To ensure the accuracy and real-time nature of the data, the present invention collects the user's actual network status information through distributed monitoring nodes. Specifically, the collected data includes:

[0284] Actual bandwidth allocation value Get real-time bandwidth usage of each user through the traffic monitoring module of the switch or router.

[0285] Actual path delay value The round-trip delay of user traffic on the current path is measured through the delay statistics function of the probe package or the network device.

[0286] As a possible implementation method, the SDN controller periodically sends statistical requests to network devices to collect the above data and store it in a central database for subsequent analysis.

[0287] Deviation analysis and trigger mechanism:

[0288] In order to evaluate the gap between the actual operating status and the optimization result, the present invention designs a trigger mechanism based on deviation analysis. Specifically, the deviation analysis includes the following steps:

[0289]

[0290] in:

[0291] ΔB i Indicates the bandwidth deviation of user i, in bps;

[0292] B i is the bandwidth allocation value of user i in the optimization result, in bps;

[0293] The actual bandwidth usage of user i, in bps.

[0294] Delay deviation calculation:

[0295]

[0296] in:

[0297] Δd i It represents the path delay deviation of user i, in seconds;

[0298] d i is the path delay value of user i in the optimization result, in seconds;

[0299] is the actual path delay value of user i, in seconds.

[0300] As an option, when the bandwidth deviation ΔB i Or delay deviation Δd i Exceeding the preset threshold ∈ B or ∈ d When , the feedback adjustment process is triggered. In general, the bandwidth deviation threshold ∈ B and delay deviation threshold ∈ d The values ​​of can be dynamically set according to actual scene requirements, for example, set to optimized values ​​of 10% and 20% respectively.

[0301] In this embodiment, the feedback adjustment process triggered by deviation analysis includes the following contents:

[0302] When the bandwidth deviation ΔB i When the threshold is exceeded, the bandwidth allocation policy for the user is readjusted. The adjustment amount is calculated using the following formula:

[0303]

[0304] in:

[0305] The new bandwidth allocation value for user i, in bps;

[0306] η is the bandwidth adjustment coefficient, which ranges from 0<η≤1 and is used to control the adjustment amplitude.

[0307] Path Adjustment:

[0308] When the delay deviation Δd i When the threshold is exceeded, the user's path selection is re-evaluated and the routing table is updated based on the current network status. The new path selection is determined by the following formula:

[0309]

[0310] The parameter definitions are the same as those in the path optimization module, and the criteria for optimal path remain unchanged.

[0311] Status information update:

[0312] The adjusted bandwidth allocation and path selection values ​​will be used as the input state information for the next round of optimization, thus realizing feedback closed-loop control.

[0313] Dynamic prediction and model iteration:

[0314] In order to further improve the adaptability of the feedback mechanism, the present invention predicts the future network status through a time series prediction model, and updates the optimization model in combination with the prediction results.

[0315] As a possible implementation method, the ARIMA model or LSTM model is used to predict user bandwidth requirements and path delay. The prediction formula is as follows:

[0316]

[0317] in:

[0318] is the predicted bandwidth requirement of the i-th user at the next moment;

[0319] f is the mapping function of the time series model.

[0320] Prediction results and It will be used as the input parameter of the optimization model to dynamically update the objective function and constraints.

[0321] In this embodiment, the ultimate goal of the feedback mechanism is to achieve system stability and global optimality through multiple optimization iterations. After each optimization is completed, the system will readjust the optimization target based on the latest feedback data and repeat the optimization process of steps 2 to 5 until the network state converges or meets the preset performance indicators.

[0322] As an option, the termination condition of the optimization process can be defined as one of the following two:

[0323] The bandwidth deviation and delay deviation of all users are less than the threshold, that is:

[0324] max(ΔB i ,Δd i )<∈

[0325] The rate of change of the objective function value of the optimization model is less than the set value, that is:

[0326]

[0327] Where δ is the optimization convergence threshold.

[0328] The feedback mechanism in this embodiment ensures the robustness and stability of the system through real-time data collection, deviation analysis and dynamic adjustment.

[0329] Through the above feedback control process, the present invention can always keep resource allocation highly matched with user needs in a complex and changeable network environment, thereby significantly improving the overall performance of e-sports broadband access.

[0330] 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 method for optimizing e-sports broadband access based on a computing-network integration campus scenario, characterized in that: The method comprises the following steps: Step 1: Collect real-time status information in the campus network, the status information includes user bandwidth requirements, path delay and link load data; Step 2: construct a multi-objective optimization model, wherein the objectives of the model include: minimizing the deviation between user path delay and bandwidth allocation and ideal demand; Step 3: Use the Lagrange multiplier method to solve the multi-objective optimization model to obtain the bandwidth allocation result of the user; Step 4: Construct a differential game model based on the user utility function, use the dynamic game method to solve the resource allocation strategy among users, and generate the user's path selection and bandwidth allocation optimization plan; Step 5: Send the optimization calculation results to the network devices through the software-defined network controller to adjust the user's path and bandwidth allocation; Step 6: Based on the feedback mechanism, collect the user's actual bandwidth allocation and path delay data, update the network status information and repeat the optimization process.

2. The method for optimizing e-sports broadband access based on the computing-network integration campus scenario according to claim 1 is characterized in that: The constraints of the multi-objective optimization model in step 2 include: S1. The total bandwidth allocation of users shall not exceed the total bandwidth limit of the system; S2. The bandwidth allocated to the user shall not be lower than the preset minimum bandwidth requirement; S3. The delay of the user path shall not exceed the preset maximum delay threshold.

3. The method for optimizing e-sports broadband access based on the computing-network integration campus scenario according to claim 1 is characterized in that: The solution process of the Lagrange multiplier method in step 3 includes: S1. Construct a Lagrangian function and use path delay, bandwidth allocation deviation and constraint conditions as a combination expression of the optimization target; S2, iteratively optimizing user bandwidth allocation using a gradient descent method; S3. Adjust the Lagrange multiplier until the optimization result meets the convergence condition.

4. The method for optimizing e-sports broadband access based on the computing-network integration campus scenario according to claim 1 is characterized in that: The user utility function of the differential game model in step 4 includes: S1, negative effect of user path delay; S2, the positive impact of users’ actual allocated bandwidth on utility; S3. The negative impact of resource allocation competition among users on utility.

5. The method for optimizing e-sports broadband access based on the computing-network integration campus scenario according to claim 1 is characterized in that: The dynamic game solving process in step 4 includes: S1. Define the dynamic change equation of user bandwidth allocation and path selection based on user utility function; S2. Solve the game equilibrium point through numerical iteration and generate the user's optimal resource allocation strategy.

6. The method for optimizing e-sports broadband access based on the computing-network integration campus scenario according to claim 1 is characterized in that: The optimization result in step 5 is sent to the network device through the software defined network controller, including the following operations: S1, adjust user path selection and generate the optimal routing strategy based on the remaining bandwidth and load of the link; S2. Configure the service priority queues in the network device to allocate higher bandwidth resources to high-priority users.

7. The method for optimizing e-sports broadband access based on the computing-network integration campus scenario according to claim 1 is characterized in that: The feedback data in step six is ​​analyzed by a time series prediction model to predict changes in network status in the next period of time. The time series prediction model includes an ARI MA model or an LSTM model.

8. An e-sports broadband access optimization system based on a computing-network integration campus scenario, using the e-sports broadband access optimization method based on a computing-network integration campus scenario as described in any one of claims 1-7, characterized in that: The system comprises: The network status collection module is used to collect user bandwidth requirements, path delay and link load data; The optimization calculation module is used to build a multi-objective optimization model based on network status data and solve the user bandwidth allocation through the Lagrange multiplier method; The differential game module is used to solve the dynamic game of resource allocation among users based on the user utility function and generate user path selection and bandwidth allocation strategies; The resource scheduling module is used to send the optimization calculation results to the network devices through the software-defined network controller to adjust the user's path selection and bandwidth allocation; The feedback monitoring module is used to collect users' actual bandwidth allocation and path delay data, and update network status information to support optimization iteration.

9. The system for optimizing e-sports broadband access based on computing-network integration campus scenarios according to claim 8 is characterized in that: The network status collection module includes a distributed network monitoring node and a data aggregation unit, which is used to collect and summarize the user's network status information in real time.

10. The system for optimizing e-sports broadband access based on computing-network integration campus scenarios according to claim 8 is characterized in that: The optimization calculation module solves the multi-objective optimization model based on the KKT condition, and calculates the optimization result of the differential game module in combination with the Nash equilibrium theory.

Images