A wavelet-based multi-sensor data fusion algorithm

Absfrnd - This paper presents a wavelef transform-based data fusion algorithm for multi-sensor systems. Wfh fhis algorithm fhe optimum estimafe of a measurand can be obtained in terms of Minimum Mean Square Error. The variance of the optimum esfimate is not only smaller than that of each observalion sequence but also smaller than the arifhmefic average estimate. To implement this algorithm, fhe variance of each observalion sequence is estimafed using wavelef tronsform and fhe optimum weighfing factor to each observation is obtained accordingly. Since fhe variance of each observation sequence is esfimafed only from ifs most recent dafa of a predefermined lengfh, the algorithm is sew-adaptive. This algorithm is applicable to both stafic and dynamk sysfems including timeinvariant and lime-variant processes. The effeciiveness of the algorifhm is denwnsfraled using apiecewise-smoofh signal and a time-varyingflow signol.