ParGen: A Parallel Method for Partitioning Data Stream Applications in Mobile Edge Computing

The emergence of mobile edge computing enables many new mobile applications to run with relatively low costs by offloading the modules to nearby edge clouds. The edge cloud always and has limited resources and serves multiple users in the proximity. As a result, it is important to partition the computations between the mobile devices and the edge cloud, and meanwhile allocate the resources for multiple users with the aim of maximizing the average performance of the users. Existing optimization methods are usually costly in time especially when the number of users is large. In this paper, we propose a parallel method for achieving high efficiency as well as good performance in multi-user mobile edge computation partitioning. Our proposed method divides the users into small groups, and performs the genetic algorithm for the groups in parallel to find the in-group optimal solutions. By iteratively allocating the resources among the groups, the method converges to a near-optimal solution in global. Through extensive simulations, we show that our parallel method significantly reduces the execution time while guaranteeing competitive performance in average throughput compared with the benchmark algorithm.