Decentralized detection in sensor networks

Abstract

In this paper, we investigate a binary decentralized detection problem in which a network of wireless sensors provides relevant information about the state of nature to a fusion center. Each sensor transmits its data over a multiple access channel. Upon reception of the information, the fusion center attempts to accurately reconstruct the state of nature. We consider the scenario where the sensor network is constrained by the capacity of the wireless channel over which the sensors are transmitting, and we study the structure of an optimal sensor configuration. For the problem of detecting deterministic signals in additive Gaussian noise, we show that having a set of identical binary sensors is asymptotically optimal, as the number of observations per sensor goes to infinity. Thus, the gain offered by having more sensors exceeds the benefits of getting detailed information from each sensor. A thorough analysis of the Gaussian case is presented along with some extensions to other observation distributions.

Asymptotic Results for Decentralized Detection in Power Constrained Wireless Sensor Networks

“…From this observation, we obtain a result analogous to our previous result on capacity constrained sensor networks [11].…”

“…Since the conditions of Theorem 3 hold for Gaussian observations and exponential observations (see [11]), having identical binary sensor nodes is asymptotically optimal in these two situations. Note that the results presented in this paper assume only one observation per sensor node and are valid as long as the wireless sensing system is large enough.…”

Fuzzy-set-based decision making through energy-aware and utility agents within wireless sensor networks

“…Thus, the data gathering of a WSN is separated into the following processes: sampling, local decision making, data collection and belief generation. Data fusion is generally categorized into three levels, namely raw data fusion, feature fusion and decision fusion (Dasarthy 1994;Chamberland and Veeravalli 2003):…”

“…Centralized decision-making is usually adopted for the WSNs where all information passes up the hierarchy to the BS network (Niu et al 2004;Chamberland and Veeravalli 2003), or even further up the hierarchy to a fixed workstation. A key limitation of such decision-making is that it can incur a significant penalty cost in terms of time and power to propagate the fused data back to the BSs.…”

Sparse multidimensional scaling for blind tracking in sensor networks

“…The detection problem has also been addressed for under communication constraints, where the sensor transmitting the information needs to send an optimal summary of the gathered information to the fusion center [7]. In the context of anomaly detection in internet data, approximate density of incoming traffic is constructed for each location.…”