Research On Robust Power Allocation In Cognitive Radio Network With Imperfect CSI

Increasing demand of wireless services has made the radio spectrum a very scarce and precious resource. Cognitive radio(CR) can effectively improve the spectrum utilization where second users(SUs) are allowed to communicates over the same bandwidth that has been allocated to the existing primary users(PUs). OFDM has already been widely recognized as a candidate transmission technology for CR mainly due to its great flexibility in dynamically allocating the unused spectrum among SUs. Power allocation is one of the most important issues as high performance is expected, and the imperfect channel state information(CSI) in actual environment has brought new challenges to the power allocation. In this thesis, we study the power allocation to maximize CR capacity with imperfect CSI.First, minimum-mean-square-error(MMSE) model is used to obtain channel gain in cognitive radio with single CU and single PU. SU optimizes transmission power that maximizes the ergodic capacity with respect to average power and interference outage constraint. We use the Lagrangian method to find general expression for the optimal power and ergodic capacity. In addition, an asymptotic analysis of the ergodic capacity is presented at low and high signal noise ratio. Also, the key parameter is obtained that effect the ergodic capacity at low and high signal noise ratio.Second, the imperfect CSI of each user at each frequency is deterministically modeled under an ellipsoidal approximation in multi-user OFDM cognitive networks. Also, a robust iterative water-filling algorithm is proposed. The power allocation problem of the SU network is formulated as a noncooperative game, in which each SU maximizes his capacity, subject to power constraints and robust interference constraints imposed by PUs. We analyze the robust game along with its convergence properties using the advanced theory of variational inequalities, then the condition of uniqueness of the Nash equilibrium is obtained. Furthermore,we get the optimal power allocation solution of SU using water-filling method.Finally, performance evaluations of the proposed robust power allocation algorithms are taken under MATLAB simulation. The performances of SU capacity, interference constraints imposed to PU with imperfect CSI are analyzed, and the results show the reasonability of proposed algorithms on satisfying the PU interference limit.