Dynamic subnets for sensor networks [navy radar tracking sensor network example]

A sensor´s capability to distribute data in a wireless network is limited by the available bandwidth (Hz) and/or allocated spectrum and the sensor´s communication spectral efficiency (bits/sec/Hz). It is communication constraints that prevent sensors in a network from passing arbitrarily large volumes of data to one another, potentially hampering even modest scalability. To address scalability limitations in such sensor networks, we have developed an algorithm for dynamically creating subnets comprised of small groups of sensors that share information. This algorithm chooses the sensors that comprise a subnet, based on the quality of the mutual information that each sensor´s data imparts weighed against the attendant cost of communications; we call this algorithm dynamic subnets for sensor networks (DSSNET). This paper formalizes the algorithms that constitute DSSNET, compares DSSNET´s performance to other robust and popular data compression techniques, and describes DSSNET´s application in a representative navy littoral combat networked sensor (LCNS) scenario.