Lattice Codes for C-RAN Based Sectored Cellular Networks

This paper demonstrates the advantage of utilizing lattice codes in uplink sectored cellular networks employing Cloud radio access network (C-RAN). We give a novel decomposition of the cellular network into interfering but non-overlappping clusters. We employ compute-and-forward (CoF) and quantized-CoF (QCoF) for each cluster by allowing each sector to compute best equation and to treat out-of-cluster as noise. CoF may perform poorly due to rank deficiency of the integer coefficient matrix. This is solved using QCoF, which shows highly favorable performance in terms of average sum-rate. To reduce implementation complexity of the proposed scheme for QCoF, we investigate the effect of employing less number of nested lattices than number of users, i.e., some users apply the same lattice code, and hence, the rate of the lattice code is chosen as the minimum of rates of these users. Simulation results show that the sum rate degradation of QCoF due to using fewer nested lattices is not significant especially if the fronthaul capacity is limited. It is also shown that QCoF with reduced number of nested lattices outperforms CoF even if one lattice code is employed for all cluster users.