TABLE OF CONTENTS So far, we have focused on the problem of node localization, which is that of determining the location of a single unknown node given a number of nearby references. A broader problem in sensor systems is that of network localization, where several unknown nodes have to be localized in a network with a few reference nodes. While the network localization problem can rarely be neatly decomposed into a number of separate node localization problems (since there may be unknown nodes that have no reference nodes within range), the node localization and ranging techniques described above do often form an integral component of solutions to network localization.
The performance of network localization depends very much on the resources and information available within the network. Several scenarios are possible: for instance there may be no reference nodes at all, so that perhaps only relative coordinates can be determined for the unknown nodes; if present, the number/density of reference nodes may vary (generally the more reference nodes there are, the lower the network localization error); there may be just a single mobile reference. Information about which nodes are within range of each other may be available; or inter-node distance estimates may be available; inter-node angle information may be available.
Some network localization approaches are centralized, in which all the available information about known nodes, and the inter-node distances or other internode relationships are provided to a central node, where the solution is computed. Such...
