ABSTRACT
The simplest quantity that one can measure with a microphone is signal strength without any reference to neighboring microphones. The chapter provides an estimation of errors in a source location, in particular when one is unable to correctly estimate the noise power. It considers an alternate approach where one can estimate the direction cosines of the transmitter measured from a reference sensor. It explores how frequency difference of arrival (FDoA) data can be used for localization. In the presence of random position errors of the anchor nodes, DoAs, even though error-free, will not merge into a single point representing the transmitter but within a blurred patch. One can improve the accuracy and bias of location estimates by invoking the concept of total least squares (TLS). The basic idea in compressive sensing is combining random sampling with reconstruction so that the original signal being sampled is recovered under a condition that signals are sparse, which we shall elaborate on later.
