Optimum Power Allocation for OFDM Based Cognitive Radio Systems with Arbitrary Input Distributions

In the literature, optimal power allocation assuming Gaussian input has been evaluated in OFDM based Cognitive Radio (CR) systems to maximize the capacity of the secondary user while keeping the interference introduced to the primary user band within tolerable range. However, the Gaussian input assumption is not practical and Finite Symbol Alphabet (FSA) input distributions, i.e., M-QAM are used in practical systems. In this paper, we consider the power optimization problem under the condition of FSA inputs as used in practical systems, and derive an optimal power allocation strategy by capitalizing on the relationship between mutual information and minimum mean square error. The proposed scheme is shown to save transmit power in a CR system compared to its conventional counterpart, that assumes Gaussian input. In addition to extra allocated power, i.e., power wastage, the conventional power allocation scheme also causes nulling of more subcarriers, leading to reduced transmission rate, compared to the proposed scheme. The proposed optimal power algorithm is evaluated and compared with the conventional algorithm assuming Gaussian input through simulations. Numerical results reveal that for interference threshold values ranging between 1mW to 3mW, the transmit power saving with the proposed algorithm is in the range between 55-75%, 42-62% and 12-28%, whereas the rate gain is in the range between 16.8-12.4%, 13-11.8% and 3-5.8% for BPSK, QPSK and 16-QAM inputs, respectively.