Algorithms and Models for Positioning and Scheduling in Wireless Sensor Networks

Doctoral thesis, 2008

This thesis considers two problems related to wireless sensor networks. The first (and main) considered problem is the inference of sensor node positions based on transmissions of RF signals between sensor nodes and/or between sensor nodes and fixed reference nodes. We study the Cramér-Rao lower bound on positioning errors in asynchronous wireless sensor networks, and propose positioning algorithms tailored for implementation in these types of networks. In addition to positioning algorithms, we also study algorithms for the estimation of distance between network transceivers based on transmitted wide-band RF signals, and consider the interaction between the ranging and the positioning algorithm. In the algorithm design, we aim for low complexity and robustness against the most common types of error sources, including errors caused by blocked (non-line-of-sight) RF channels, and/or multipath propagation. On a side-track, we study the feasibility of characterizing the surrounding environment in which a wide-band wireless sensor network is deployed. This characterization is done in terms of the approximate location of reflective objects that generate significant multi-path components and/or amplify-and-forward relays. The second problem we consider is a scheduling problem that appears not only in wireless sensor networks, but also in other wireless networks. We propose a relatively simple model for packet-loss in a Rayleigh fading environment, and use this model in an attempt to schedule transmissions in the network so as to minimize the average probability of packet-loss. Since wireless sensor nodes often only have a limited energy source, and packet retransmissions consume energy, this problem is especially important in the context of wireless sensor networks.