On Statistical Properties of Co-channel Interference in OFDM Systems

Co-channel interference is a key problem in orthogonal frequency division multiplexing (OFDM) wireless communication systems. Channel coding is an effective means of combating the interference. In order to achieve a good decoding performance, the accurate interference distribution should be known at receiver. In most of the previous studies, the co-channel interference was usually modeled as a Gaussian random variable, however, was found to be inaccurate for OFDM cellular systems recently in few works. In this paper, we study the statistic properties of co-channel interference in wireless OFDM systems. By using Monte Carlo simulation and curve-fitting, non-Gaussian distributions are found to be more accurate than the Gaussian distribution. Based on the derived interference distributions, the log-likelihood ratio (LLR) in a maximum a posteriori (MAP) decoder is calculated to enhance the error correction ability. Simulation results show that a dramatic performance gain can be achieved by using the derived interference distributions compared to the conventional decoder using the Gaussian model.