Localizing transceiver-free objects: the rf-based approaches

Traditional radio-based localization technologies all require the target object to carry a transmitter (e.g., active RFID), a receiver (e.g., 802.11x detector), or a transceiver (e.g., sensor node). In practice, however, such requirements can not be satisfied in many applications, such as security and surveillance, intrusion detection, outdoor asset protection, and location-aware applications. In this dissertation, I propose a new localization scheme called transceiver-free localization. The basic idea of transceiver-free localization is to utilize the change of wireless signals of different wireless links to locate the target object. I prove that the object detection behavior of each wireless link can be described by two models. They are called Signal-Dynamic (S-D) Model and Transmitter-Receiver (T-R) model. The former one is a deterministic model and the later one is a probabilistic model. T-R model presents many unique features and new requirements. Although it is derived from transceiver-free object localization, it presents promising generality which enable it be applied in a much broader scope of application, calling a revisit for most of coverage problems. Moreover, in order to serve different localization requirements and address the problem in centralized or distributed environments, I propose five localization algorithms called Midpoint, Intersection, Best-cover, Dynamic Clustering and RASS. The former 3 algorithms are based on centralized environment. Dynamic Clustering and RASS are able to locate multiple objects. I prove that RASS guaranteed tracking accuracy is bounded by only about 0.26s without sacrificing the accuracy and scalability. Experimental results show that these algorithms can have remarkable high accuracy up to 0.85m. At Last, our transceiver-free localization approaches can also be utilized to improve the accuracy of traditional transceiver-based approaches. Cocktail is a hybrid approach by using WSN and RFID technologies. Experiment results show that it can improve the accuracy of traditional pure RFID system by 75% in a large indoor area. Keywords: wireless sensor networks, transceiver-free, RSSI, dynamic clustering, SVM, multi-channel, signal dynamic, T-R model, object detection probability