Cognitive Radio Network Duality and Algorithms for Utility Maximization

We study a utility maximization framework for spectrum sharing among cognitive secondary users and licensed primary users in cognitive radio networks. All the users maximize the network utility by adapting their signal-to-interference-plus-noise ratio (SINR) assignment and transmit power subject to power budget constraints and additional interference temperature constraint for the secondary users. The utility maximization problem is challenging to solve optimally in a distributed manner due to the nonconvexity and the tight coupling between the power budget and interference temperature constraint sets. We first study a special case where egalitarian SINR fairness is the utility, and a tuning-free distributed algorithm with a geometric convergence rate is developed to solve it optimally. Then, we answer the general utility maximization question by developing a cognitive radio network duality to decouple the SINR assignment, the transmit power and the interference temperature allocation. This leads to a utility maximization algorithm that leverages the egalitarian fairness power control as a submodule to maintain a desirable separability in the SINR assignment between the secondary and primary users. This algorithm has the advantage that it can be distributively implemented, and the method converges relatively fast. Numerical results are presented to show that our proposed algorithms are theoretically sound and practically implementable.