Probability grid: a location estimation scheme for wireless sensor networks

Location information is of paramount importance for wireless sensor networks (WSN). The accuracy of the collected data can significantly be affected by an imprecise positioning of the event of interest. Despite the importance of location information, real system implementations that do not use specialized hardware for localization purposes have not been successful. In this paper, we propose a location estimation scheme that uses a probabilistic approach for estimating the location of a node in a sensor network. Our localization scheme makes use of additional knowledge of topology deployment. We assume a sensor network is deployed in a controlled manner, where the goal of the deployment is to form a grid topology. We evaluate our localization scheme through simulations, showing localization errors as low as 3% of radio range. We outperform similar localization schemes by obtaining 50% less error in localization. We also evaluate our localization solution and the DV-hop scheme in a real implementation, obtaining an average error in location of 79% of radio range, outperforming DV-hop by approximately 40%. We analyze the significant differences in performance between simulations and a real implementation and stress the importance of further evaluations of real implementations. The result is an effective and realistic protocol that works in an actual implementation, under certain assumptions, because it exploits deployment information.