Using fuzzy logic for localization in mobile sensor networks: simulations and experiments

Localization is an issue of vital importance for the functioning of autonomous mobile sensor networks. Location information, allows a mobile sensor node to navigate complex environments and perform local tasks successfully. In mobile sensor networks, this information facilitates important functions like topology control, collision avoidance and development and security of routing protocols. This issue can be broadly divided into the problems of global position estimation, and once that is achieved, of local position tracking. To tackle these, mainly two distinct methods have been used in the past. One is the use of specialized hardware and another is the use of probabilistic estimation methods. This paper proposes the use of fuzzy logic to tackle this problem. Fuzzy logic allows us to do away with strict probabilistic rules and to set up heuristic fuzzy rules. It also reduces computation time. A grid-based map is used to describe the environment of the mobile node and its confidence in its position at each grid-point is determined using sensor measurements. In case the node is receiving information from multiple sensors, this paper demonstrates using simulations, the robustness of the scheme to inaccurate sensor information or node confidence within practical limits. This paper also applies the fuzzy rules to track the node's position as it moves. In order to reduce computational cost, this paper proposes limiting the computation of confidences to significant grid-points only. An experiment is conducted using a DIRRS range sensor which demonstrates how the sensor measurements are used for global position estimation and local position tracking