Multi-sensor quickest detection by exploiting radio correlation

In this paper, we present a novel quickest detection scheme to sequentially detect the emergence of an event with the help of multiple sensors. In the proposed scheme, we exploit the fact that the observations of the spatially proximal sensor nodes are highly correlated due to correlated shadowing effects. However, we will see that when it comes to infer the spatial correlation, the estimate of the spatially structured covariance matrix is not feasible as the proposed quickest detection scheme is a recursive method and operates with single sample. Hence, we propose to model the spatial covariance structure by using the a-priori information about the locations of the sensors. Moreover, the proposed scheme also takes into account a scenario where only a subset of sensors are affected by the event's signal. Therefore, it takes into account both the exploitation of the spatial structure and the selection of the subset of sensors in the process of detection. Both analytical and numerical results are developed for the mean detection delay, showing important advantages.