Channel aware encryption and decision fusion for wireless sensor networks

In this paper, we study a secure transmission scheme in a distributed detection scenario. We consider passive eavesdropping where an enemy fusion center (EFC) located in a vicinity of an ally fusion center (AFC) tries to eavesdrop on wireless communications between sensors and the AFC. To ensure confidentiality in the transmitted data from the sensors, we design a transmission strategy for the sensors, which makes the AFC properly detect an unknown target while the EFC is totally ignorant of it. Two groups of sensors are activated and encrypt their measurements in different ways by exploiting a random channel gain of wireless fading channel as a seed for encryption. To detect the target, the AFC performs LLR based fusion rules based on the channel statistics instead of instantaneous channel state information. In addition, at high signal-to-noise ratio, we show that our fusion rule at the AFC is simplified to a modified Chair-Varshney fusion rule that only requires local detection probabilities to decide the target state. Our performance evaluations show that detection performance at the AFC exponentially decays with growing the number of sensors, whereas information-theoretic perfect secrecy is achieved at the EFC.