Modeling Chernoff information loss in cooperative sensor networks

We investigate the Chernoff information loss (CIL) characteristics in wireless sensor networks (WSNs) as the sensor attempts to send a 1-bit information about the presence or absence of a phenomenon-of-interest (POI) to a remote data sink through either the single-hop (SH) or multi-hop (MH) wireless connectivity. Firstly, we describe the sensing and communication processes by using the cascaded discrete memoryless channels (DMCs), which can be represented as an equivalent sensing channel (ESC). We then define a typical ESC and derive its Chernoff information. Both the decode-and-forward (DF) and amplify-and-forward (AF) protocols are studied. Subsequently, we consider a random-walk formulation of the MH relay path, and derive the relationship between the average distance traveled by a virtual POI-marked photon and the number of hops. Finally, the CIL model can be obtained by applying the preceding results, which provide critical insight into the problems of when and how to relay in WSNs. Eventually, this would lead to context and channel aware sensing and communication in future cognitive sensor networks.