Multi-user Stochastic Game for Utility Optimization in Mobile Ad Hoc Cloud

Through offloading task to mobile ad hoc cloud (MAHC), mobile devices can execute computation-intensive tasks faster and consume less energy without infrastructures. However, when multiple resource demanders (RDs) offload tasks to resource providers (RPs) in the MAHC, the disorderly competition among RDs can cause inhomogeneous task distributions on the RPs. This can reduce the service efficiency of the MAHC and lead to a low utility of each RD. Accordingly, we propose a stochastic game approach to solve this problem. Firstly, the utility of each RD is examined as the combination of task execution time and monetary cost. Then, we model the competition for each RD pursuing its maximum utility as a static noncooperative game. After that, we transform the single-shot game in one time slot into a stochastic game in an infinite time horizon to obtain the optimal strategy of each RD. Finally, we propose the backward iteration algorithm to reduce the computational complexity for reaching the e-Nash equilibrium of the game. Numerical results show various performances of this stochastic game. Compared with the strategy obtained from the static game, the equilibrium strategy derived from the stochastic game can effectively improve the utility of each RD.