Generalized Nash Equilibrium Game for Radio and Computing Resource Allocation in Co-located MEC

Abstract

The tower sharing approach has been widely used by Mobile Network Operators (MNOs) to save their Capital Expenditure (CAPEX) by sharing the physical infrastructure hosted by a third party tower provider. In addition, multiple Computing Resource Providers (CRP) are deploying their servers at towers by cooperating with tower providers to grant low latency, real time services to users. Thus, the resource allocation has become a challenging issue where users of different MNOs need to share the computing resources provided by CRPs. In this paper, the joint allocation of uplink, downlink and computing resources is considered to minimize the end to end latency of users where the offloading process is modeled as a network of queues. Since the resource allocation of MNOs and the CRP are coupled with each other, we formulate it as a Generalized Nash Equilibrium Problem (GNEP). We propose a penalty based algorithm to solve the formulated GNEP with an effective initialization approach to improve the performance of the algorithm. Then, we perform the simulation to analyze the performance of the algorithm.