Distributed Detection with Non-identical Wireless Sensors for Industrial Applications

There has been very little exploration when it comes to design distributed detection techniques and data fusion rules with non-identical sensors. This concept can be utilized in many possible applications within industrial automation, surveillance and safety. Here, for a global event, some of the sensors/detectors in the network can observe the full set of the hypotheses, whereas the remaining sensors infer more than one hypotheses as a single hypothesis. The local decisions are sent to the decision fusion center (DFC) over a multiple access wireless channel. In this paper, a fusion rule based on minimization of variance of the local mis-detection is proposed. The presence of sensors with limited detection capabilities is found to have a positive impact on the overall system performance, both in terms of probability of detection and transmit power consumption. Additionally, when the DFC is equipped with a large antenna array, the overall transmit power consumption can be reduced without sacrificing the detection performance.