Robust Power Allocation in Cognitive Radio Networks with Uncertain Knowledge of Interference

We study the secondary user power allocation in OFDM-based cognitive radio networks with underlay mode where secondary user is allowed to share the spectrum licensed to primary user, provided that limited interference is generated to primary user. While this problem has been studied extensively in the literature, most of the previous works in this context assume that secondary user has the exact information of how much interference it generates to primary user, which may not be the case in practice. We assume that the interference channel gains are known to secondary user with uncertainty, and develop a power allocation algorithm that can keep interference under a desired level. In particular, we formulate the power allocation problem as a chance-constrained program that finds a power allocation that guarantees limited interference with high probability. The formulation is non-convex in general, and we apply techniques from robust optimization theory to convexify the formulation. We show through simulations that our approach finds a power allocation that remains feasible under time-varying channel conditions.