Delay Optimization of Computation Offloading in Multi-Hop Ad Hoc Networks

In recent years, computation-intensive mobile applications are rapidly developed such as Augmented Reality(AR), which will become one of the most popular services. However, mobile devices are resource-limited in memory, bandwidth and process capacity. To relieve the computation burden of mobile devices, computation offloading appears. Offloading computation to remote cloud can take advantage of the rich resource of core network, but it also introduces long delay which has effects on user experience especially for delay-sensitive application. Computation offloading based on ad hoc network reduces the delay of computation offloading and makes good use of the computation resource of mobile devices. In this paper, the average delay minimization of component-based linear applications is investigated in multi-hop ad hoc networks. The optimization problem is formulated as a general integer linear programming with the constraints of processing capacity of nodes, which is a NP-hard problem. In order to get the optimized result, a delay-optimization based ant colony optimization algorithm (DoACO) is proposed. Simulations show that the DoACO has a better performance compared with traditional ant colony optimization algorithm(ACO), simulated annealing algorithm, random offloading algorithm and local execution algorithm.