Random Matrices and the Impact of Imperfect Channel Knowledge on Cooperative Base Stations

Base station cooperation is a very promising technique for reducing inter-cell interference and enhancing spectral efficiency in next generation cellular systems. However, imperfect precoding at the base stations is limiting the potential gains. In this paper, we present a downlink signal model with impaired zero-forcing precoding and analyze self-user signal and inter-user interference. Properties of Wishart-distributed random matrices are used for calculating the average inverse eigenvalues. We provide closed-form expressions for signal to interference ratio (SIR) bounds, which are found inversely proportional to the mean square error of the base station's channel knowledge. Further, it is found that the SIR increases with the ratio of number of base stations to number of users. Numerical evaluation agrees with analytical results.