MERIT: MEsh of RF sensors for Indoor Tracking

Traditionally, the problem of indoor tracking is solved by using some non-RF ranging technologies (such as infrared or ultrasound). The problem with these non-RF ranging techniques is that they do not work very well when the tracking device is buried in users' wallets or bags. Therefore, there has been considerable interest in using pure RF techniques. However, most the existing techniques in this space requires an expensive site survey or a floor-plan for reasonable accuracies. In this paper, we describe the {em MERIT} system that we designed, implemented, and evaluated. MERIT is significantly different from previous indoor tracking systems. It is pure RF-based, but it requires neither a site survey nor a floor-plan. It can even be deployed by the end users themselves. An obvious problem to pure RF-based systems, however, is whether such a system is accurate enough for practical purposes. Specifically, RF signal can traverse through walls and partitions, and there is the phenomenon of indoor multipath interference, which makes correlating signal strength and distance difficult. Under these two conditions, can a RF-based system accurately disambiguate locations in two neighboring rooms? As a first step to address the problem, we proposed two simple techniques: {em spatial diversity} and {em RF reflector}, and showed that they are effective. In our evaluation MERIT achieved accuracy of 98.8%. MERIT was first preceived for a telecommunication application: a service of intelligent telephone call routing, but it can also be used for other location-aware services.