We describe a sparse signal detection algorithm that is being developed in the context of an on-going research project in spectrum sensing for cognitive radio. The objective is to detect the support of a sparse signal vector, whose exact number of nonzero components is unknown, using a sequence of noisy linear measurements of the signal vector. We use an adaptive tree-structured algorithm, wherein the signal vector is recursively subdivided and standard detection techniques are used to determine the presence or absence of signal components in smaller and smaller subvectors. Detailed description and performance evaluation will be discussed, as well as comparison with traditional techniques. The motivation behind this research can be found, for example, in the dramatically increasing need for communication services. The algorithm might be applied in a scenario in which a cognitive radio needs to scan a sparse wideband spectrum in order to determine which sub-band is suitable for opportunistic transmission.
October 18 @ 12:30
12:30 pm (1h)
Discovery Building, Orchard View Room