Protection Strategies For Mobile Edge Computing Network

As an important part of 5G mobile network architecture,mobile edge computing can effectively overcome the disadvantages of high latency and extra network load in traditional cloud computing networks.Moreover,survivability is an important aspect for the quality of service(QoS)of MEC networks.Focusing on the survivability of MEC networks,we propose two efficient protection strategies to protect against the failures of access point and MEC server,which are referred to as "1+1 dedicated backup resource protection" and"shared backup resource protection."In the 1+1 dedicated backup resource protection strategy,we choose MEC servers that can meet the latency requirement of a MEC service to reserve protection resources,and link-disjoint paths to reserve protection bandwidth,respectively.Considering different requirements on latency and bandwidth,we classify user MEC services into two different types,i.e.,services supported by distributed protection resources(i.e.,distributed protection)and services only supported by centralized protection resources(i.e.,centralized protection).Based on these two types of services,we develop Integer Linear Programming(ILP)models and propose efficient heuristic algorithms to balance loads in MEC servers and minimize the total protection bandwidth.Simulation results show that the proposed strategy can effectively balance the server loads and minimize network protection bandwidth in MEC networks.For the case of distributed protection,the proposed heuristic based on the Barrel Effect can achieve performance close to its corresponding ILP model and significantly better than the heuristic based on average load distribution,which demonstrates the efficiency of the heuristic.For the case of centralized protection,the proposed heuristic based on minimum load allocation achieves performance close to its corresponding ILP model and significantly better than the heuristic based on shortest path allocation.In the 1+1 dedicated backup resource protection strategy,dedicated computing and bandwidth resources are reserved for service protection.When there is no failure in the MEC network,these resources would remain idle,which leads to low network utilization.To reduce these idle backup resources while providing 100%protection in the MEC network,we also propose a shared backup resource protection strategy,in which protection MEC resources and protection bandwidth can be shared among multiple users that do not share any common MEC servers or communication links.Based on the service types of distributed protection and centralized protection,we develop their ILP models and also propose their corresponding heuristic algorithms to balance MEC server loads and minimize the total protection bandwidth.Simulation results show that compared with the 1+1 dedicated backup resource protection strategy,the shared backup protection strategy can achieve higher utilization of protection resources.Moreover,the proposed heuristic based on the Barrel Effect and minimum load allocation can efficiently achieve performance close to their respective ILP models,which verifies the efficiency of both heuristics.