Exploiting macro-diversity in cellular networks using the sum-product algorithm

We present a distributed macro-diversity algorithm for the uplink process in a cellular network, where base stations cooperate via message-passing to improve signal quality. We show that the cellular network can be modeled as a factor graph with loops and a simple distributed cooperative scheme for base stations can be derived from the sum-product algorithm. Assuming Gaussian signaling, we analyse the convergence of the algorithm and show its dependency on cellular system parameters such as the intercell interference, the signal-to-noise (SNR) ratio and the network size. We show that the algorithm converges if the intercell interference level is below some threshold, which is dependent on the SNR level in general. We analyse the rate of convergence with respect to the intercell interference and SNR level and find that the convergence rate is slower for higher intercell interference and higher SNR level. We also find that under certain channel conditions, the algorithm may not converge if the network size is too large. However, if the algorithm converges, the convergence rate is insensitive to the network size.