RGDAG: A Framework for Joint Optimization of Edge Server, User Request, and Application Placement in MEC

With the rapid development of wireless technology, traditional cloud computing technologies are no longer sufficient to meet current computing demands. Mobile Edge Computing (MEC) has emerged as a new computing paradigm. MEC enables service providers to deploy IoT applications on edge servers, thereby delivering low-latency services to users.However, due to the limited capacity of MEC to host applications at the network edge, service providers face new challenges in how to fully utilize scarce edge resources to maximize system performance.Addressing this challenge requires careful study of application placement and service offloading to coordinate different MEC nodes while also reasonably resolving routing request and user mobility issues. However, existing research discusses application deployment, task unloading, and routing matching separately, which does not actually correspond to the real situation. Based on this situation, this paper proposes a dynamic optimization algorithm Random Approximate Games and Dynamic Approximate Matching Games(RGDAG) under long-term budget constraints, which decomposes the long-term problem into a series of one-time subproblems, each of which is a two-stage problem. Among these, the Nash Approximation Edge Caching (NAEC) is used to address the application deployment subproblem, where an approximate method is employed to determine the deployment decisions for applications. Subsequently, these decisions are adjusted using Nash equilibrium. To address the user routing decision subproblem, Dynamic Preference Network Matching (DPNM) and Game Theory Matching Balancers (GTMB) are utilized to achieve a balanced connection between servers and users.This paper conducts rigorous theoretical analysis and extensive simulation experiments to validate the effectiveness of the proposed method.The results indicate that the proposed method is superior to the compared algorithms in terms of effectiveness.