Efficient Task Allocation for Mobile Crowd Sensing Based on Evolutionary Computing
Xi Tao, Wei Song · 2018
Mobile crowd sensing (MCS) offers a promising paradigm for big data collection in a large scale. It leverages the power of mobile smart devices, and shows various advantages over traditional sensing networks, such as high energy efficiency, cost-effectiveness, and flexibility. A key problem in MCS is to efficiently allocate distributed tasks to mobile users (MUs) while addressing various constraints, e.g., in terms of the quality of sensed data and collection cost. In this paper, we further take into account the clustering effect of sensing tasks and propose an efficient approach to solve the NP-hard task allocation problem. In our solution, a variant genetic algorithm (GA) is utilized to maximize the task complete ratio and balance the sensed data among tasks while respecting the MUs' constraints. Considering the unique characteristics of the task allocation problem, we divide the crossover and mutation process of the GA into two steps. First, an available subset of tasks is selected for each MU. Then, a feasible travelling path is designed over this subset of tasks in the second step. The simulation results show that the proposed GA-based solution significantly outperforms the baseline solution in terms of task complete ratio and data balance.