Transient Analysis of Data Traffic in Cognitive Radio Networks: A Non-Equilibrium Statistical Mechanics Approach

Abstract

In contrast to traditional communication networks, the data traffic of cognitive radio network is interrupted by primary users. To evaluate the impact of primary user activities, the transient data traffic dynamics of cognitive radio networks, after being interrupted by primary users, are studied. The Onsager's regression approach, a powerful principle in non-equilibrium statistical mechanics, is applied. The autocorrelation function in the equilibrium state is used to study the time evolution of system deviation from the equilibrium state, according to the principle of Onsager's regression. Analytic approximations have been derived for the autocorrelation function for both generic and fluid approximation cases. Numerical simulations are carried out for a Jackson network model. The validity of the Onsager's regression is verified. The derived approximation of the correlation function is demonstrated to match the true value at the beginning of the relaxation. A scheduling algorithm for fast recovery of equilibrium is also proposed and demonstrated by numerical results.