一种分布式博弈驱动的边缘缓存和资源优化方法OA

This work addresses the inefficiency of resource allocation in cell-free massive multiple-input multiple-output caching networks,where users compete for access point resources.A distributed resource allocation optimization method based on a buy-sell game is proposed.First,the interaction characteristics between access points and user equipment under asymmetric information in edge caching networks are analyzed.It is shown that traditional unilateral decision-making mechanisms fail to accurately capture dynamic interaction scenarios.On this basis,access points and user equipment are abstracted as resource sellers and buyers,respectively.A content pricing mechanism is introduced to establish a mathematical model rooted in the buy-sell game.Considering users'concurrent requests for popular content and power resources,as well as the limited nature of access point resources,optimization problems are formulated to maximize both buyer utility and seller utility,respectively.To tackle the mixed-integer nonlinear programming nature of these problems and the exponentially growing strategy space,a distributed optimization method based on the elite genetic algorithm is developed.This method alternately optimizes the strategies of buyers and sellers to achieve effective resource allocation under incomplete information conditions.Simulation results demonstrate that,compared with benchmark algorithms including the most popular content caching,least recently used,and random caching,the proposed elite genetic algorithm delivers significant advantages in buyer utility,seller utility,system power consumption,and cache hit rate.Notably,it maintains stable performance gains across different access point cache capacities and content popularity parameters,confirming the effectiveness of integrating the buy-sell game mechanism with elite genetic optimization for resource management in cell-free massive multiple-input multiple-output caching networks.