Connectivity in wireless networks with dynamic key compromise and recovery

This paper considers the problem of key management in wireless networks. In particular, we investigate the effect of dynamic key compromise and recovery on connectivity in large networks. A queuing model with a finite buffer is used to model the dynamics of key compromise. The exact distribution of the fraction of keys compromised is obtained. The result of the queuing analysis is used to determine the probability of outage, where an outage occurs whenever instantaneous end-to-end connectivity, in percolation sense, is not present. Numerical results show that in order to obtain a low outage probability, it is critical that key compromises are detected accurately, and that the average key recovery rate has a weak influence on the outage probability. Thus, for the same average key recovery rate the system must be designed to have a high key recovery probability rather than a large number of key recoveries per unit time with a low key recovery probability.