Jamming In Underwater Sensor Networks As A Bayesian Zero-Sum Game With Position Uncertainty

We investigate a jamming problem in an underwater acoustic sensor network, where nodes try to communicate in spite of an adversary that is attempting to block their communications. We take into account that the attenuation of underwater acoustic channels is strongly dependent on the communication distance and the signal frequency. We frame the problem in a game theoretic setup, as a Bayesian zero-sum game where the sensor network acts as the maximizer of the transmission capacity, while the jammer is the minimizer. In particular, we are interested in evaluating the effect of the nodes' position on the resulting equilibrium. The Bayesian character comes into play to represent the uncertainty on the position information of the nodes. Our evaluations show that for many network configurations, the equilibrium strategy of the jammer is pure. Thus, the transmitters can act as though the jammer only causes a higher level of interference. This allows us to identify positions where the damage caused by a jammer is easier to quantify, but the jammer itself is harder to detect.