Dynamic Task Offload System Adapting to the State of Network Resources in Mobile Edge Computing

In disaster situation, collecting information about victims and disaster area quickly is required. However, it is expected that the communication environment is not fully prepared because communication facilities may be damaged. Mobile Edge Computing (MEC) is attracting attention, in which servers are distributed to the vicinity of end users. Thus, dynamic task offloading adapting to changes of network resources is useful in unstable network. This paper proposes a dynamic task offload system based on the state of network. The proposed offload system consists of two mechanisms. First mechanism is Software Defined Mobile Edge Computing (SDMEC) architecture. Second mechanism is dynamic task placement based on the state of network resources. We formulated an optimization problem to minimize the total response time while application execution. In the proposed offload system, multiple tasks are placed dynamically depending on the load of servers and network resources to minimize response time of application. We implemented the prototype of the proposed offload system and evaluated its performance.