iStep: A Step-Aware Sampling Approach for Diffusion Profiling in Mobile Sensor Networks

We investigate the mission-critical diffusion profiling problem in mobile sensor networks, where the energy cost and the time delay are constrained. Previous studies fail to solve this problem well enough due to the extremely large searching spaces for sensor measurements and the dynamic evolutions during the diffusion process. We propose a step-aware spatiotemporal sampling approach, which is called iStep, achieving near-optimal profiling precision in terms of the overall profiling process. Leveraging the upper bound of the deviation value in sampling position during each iteration, we build a comprehensive model between sampling interval and the current state parameters, including the distances between the mobile sensors and the pollution source, as well as the confidence interval radius of the source. Building upon this model, we propose an algorithm on computing the step-aware sampling interval. The computational complexity of iStep is $O(NS^{2})$, which is in the same order of the radial approach, where $N$ is the number of sensors, and $S$ is the total number of movement steps in each iteration. Extensive simulation results show that, compared with the radial approach, iStep can improve the average precision by up to 56.8% and the worst precision by up to 59.1%. Moreover, it performs closely to the brute-force approach, which potentially verifies the near-optimal character of iStep.