Improving Localization Accuracy Using Spatio-Temporal Information of Objects

Improvised explosive devices (IEDs) are an increasingly serious military threat as is witnessed in Iraq and Afghanistan. To combat the IED emplacement, it is important to have persistent surveillance over time. Due to the low cost and capabilities of sensors, wireless sensor networks (WSNs) have tremendous potential for military and civilian surveillance. In this paper, we explore methods to improve an important aspect of surveillance: localization accuracy. Though there are many localization algorithms in the literature, all of them try to improve the accuracy from the side of sensor networks. In this paper, we tackle this problem from a new angle, that is, we look at the spatio-temporal relationships of objects we track, which, as far as we know, unprecedented in this attempt. We first develop algorithms that use spatial and temporal relationships of objects separately and then design ones that combine them. Experimental results show that all our proposed algorithms can improve localization accuracy, especially those combined ones. Moreover, since our methods use features related to objects themselves and not the underlying localization mechanism, they can be built on any localization algorithm to improve accuracy.