Statistical security model and power adaptation over wireless fading channels

Secrecy capacity is the widely-used metric to evaluate the capability of wireless channels in delivering messages with the assurance of perfect security as well as reliability. However, with the proliferation and diversification of wireless services, there have been urgent needs in fine-grained and decoupled security metrics towards future mobile communications. In this paper, we propose a statistical model to characterize the finegrained security level, while evaluating the secrecy independently from the reliability. Specifically, we model the data eavesdropping process by a queuing system, where the data arrival represents the information overheard by the eavesdropper. The departure process is modeled by a constant leaving rate, implying that the eavesdropped data will become useless after certain time. The eavesdropper has to accumulate sufficient amount of data for successful decipherment. We further define the novel statistical security metric via a queue-length bound and a bound-violation probability threshold. As long as the violation probability is below the specified threshold, the statistical security is guaranteed. This model can fit diverse services well by tuning corresponding parameters for the statistical security metric. Following the statistical security model, we formulate the security-driven power adaptation problem for throughput maximization over fading channels. Applying asymptotic queuing analyses and effective bandwidth theory, we solve the problem and obtain a set of power control schemes under diverse system parameters. Simulation evaluations are also presented to demonstrate the superiority of our scheme over the baseline schemes.