Destination Driven Computation Offloading in Internet of Things

Abstract

Computation offloading can be widely used in Internet of Things (IoT), since most IoT devices do not have enough resource for complex computing. According to the target of computing results, computation offloading can be divided into source driven pattern and destination driven pattern. Currently, most studies focus on the former, where the results should be returned to the task source after computing. Another widely existing pattern, i.e., destination driven computation offloading, where the results should be transmitted to some other node instead of the task source, is rarely noticed. In this paper we propose a scheme of Destination Driven Step-by-step Computation Offloading along the Route (DDSCOR), which optimizes the delay from the source node releasing the task to the results arriving the destination node. Specifically, the proposed scheme combines the data transmission and computation offloading together, dividing the computing task into several procedures and offloading them to appropriate nodes along the data transmission route. Simulation results show that the proposed scheme can approximately achieve the optimal offloading- transmission combined route, and tremendously reduce the overall latency in resource limited IoT.