Fault-tolerant sensor integration using multiresolution decomposition

Signal integration is an important aspect of many physical applications. It is often necessary to limit the effects of noise when data from several sensors are integrated to provide a consolidated estimate of some physical quantity being measured. This paper proposes a method of applying the idea of multiresolution to the problem of efficient integration of abstract sensor estimates when the number of sensors is very large and a large number of sensor faults are tame. The idea essentially consists of constructing a simple function from the outputs of the sensors in a cluster and resolving this function at various successively finer scales of resolution to isolate the region over which the correct sensors lie. We develop an optimal O(NlogN) algorithm, where N is the total number of sensors, that implements this idea efficiently. This proposed application will result in speeding up computations involved in reducing the measure of the integrated output estimate by giving rise to an alternative method of narrowing down the region containing the correct value of the parameters being measured by the sensors.