Proportional Fair Resource Allocation Based on Chance-Constrained Programming for Cognitive OFDM Network

Abstract

Proportional fair resource allocation plays a critical role to balance the spectrum efficiency and fairness for cognitive orthogonal frequency division multiplexing (OFDM) network. However, due to the lack of cooperation between cognitive radio (CR) network and primary network, channel state information between CR base station (CRBS) and primary user (PU) could not be estimated precisely. Therefore, the interference of CRBS---PU couldn't be computed precisely and chance-constrained programming is adopted to formulate the resource allocation problem. In this work, we study the proportional fair resource allocation problem based on chance-constrained programming for cognitive OFDM network. The objective function maximizes the spectral efficiency of cognitive OFDM network over subcarrier and power allocation. The constraint conditions include the interference constraint of PU with the target probability requirement and the proportional fair rate requirement of CR users. In order to solve the above optimization problem, two steps are taken to develop hybrid immune optimization algorithm (HIOA). In the first step, the probabilistic interference constraint condition is transformed as an uncertain function which is computed by a generalized regression neural network (GRNN). In the second step, we combine immune optimization algorithm and GRNN to develop HIOA. Simulation results demonstrate that HIOA yields higher spectral efficiency while the probabilistic interference constraint condition and the proportional fair rate constraint condition could be satisfied very well.