Service Migrations in Multi-Access Edge Computing Using Adaptive Migration Window

5G networks support a Multi-access Edge Computing (MEC) infrastructure to host many heterogeneous applications at the network's edge. MEC infrastructure helps 5G to achieve the high-bandwidth and low-latency targets demanded by modern applications. Applications hosted in MEC have Service Level Agreements (SLAs) that guarantee service quality to the users. Service migrations are done in MEC to prevent SLA violations caused by application performance or overload conditions. While several existing research on MEC study the impact of service migrations on user mobility, there is a paucity of studies on service migrations performed during application overload conditions. This paper proposes a novel Adaptive Service Migration Window (ASMWA) algorithm to initiate service migrations in MEC using the application's resource utilisation, user QoS and service migration duration. We have compared the performance of ASMWA with the state-of-the-art Exponential Weighted Moving Average (EWMA3) and static threshold algorithms. Our simulations show that the ASMWA performs better when compared to EWMA3 and static threshold algorithms in minimising SLA violations and QoS degradation during service migrations in MEC.