LLR quantization and resource allocation of constrained backhaul for multicell processing

We consider the uplink of a cellular system where base stations (BSs) cooperate for decoding signals transmitted by mobile terminals (MTs). Assuming thatMTs transmit on orthogonal channels, each BS demodulates the signal coming from each MT, obtaining the log likelihood ratio (LLR) of each encoded bit. LLRs are quantized and quantization indices are forwarded on a backhaul to the radio network controller (RNC), where the reconstructed values are summed, and decoding takes place. BS-RNC links have a constraint on the maximum supported bit rate. We design the uniform quantizers in order to maximize the generalized mutual information (GMI) on the combined LLR over all MTs under the backhaul constraint. To this end, we derive the probability mass distribution (PMD) of the quantized LLR conditioned on the value of the transmitted bit, and we compute the GMI. We then propose a greedy scheduling algorithm for the choice of the number of bits used for the representation of each quantized signal exchanged on the backhaul.