IMPROVED GLRT BASED ON THE EXPLOITATION OF SPATIAL CORRELATION BETWEEN NEIGHBORING SENSORS

On the basis of the use of the generalized likelihood ratio test (GLRT) and model order selection techniques, an improved event detector is proposed herein by incorporating the spatial correlation existing in measurements coming from neighbor-ing wireless sensors. Such a spatial dependence is proposed to be extracted through the use of the so-called signature vec-tors. These vectors incorporate the a-priori knowledge of the effect that the event will have on the sensor field. For the proposed detector, analytical and empirical results are pro-vided, showing a significant gain in performance compared to traditional approaches. 1