To offload or not to offload: An efficient code partition algorithm for mobile cloud computing

A new class of cognition augmenting applications such as face recognition or natural language processing is emerging for mobile devices. This kind of applications is computation and power intensive and a cloud infrastructure would provide a great potential to facilitate the code execution. Since these applications usually consist of many composable components, finding the optimal execution layout is difficult in real time. In this paper, we propose an efficient code partition algorithm for mobile code offloading. Our algorithm is based on the observation that when a method is offloaded, the subsequent invocations will be offloaded with a high chance. Unlike the current approach which makes an individual decision for each component, our algorithm finds the offloading and integrating points on a sequence of calls by depth-first search and a linear time searching scheme. Experimental results show that, compared with the 0-1 Integer Linear Programming solver, our algorithm runs 2 orders of magnitude faster with more than 90% partition accuracy.