In a dense small-cell (SC) network with several users to be served, a multi-user detection (MUD) can be employed across SCs, and distributed estimation is a promising technique for such a scenario. Nevertheless, large communication overhead due to frequently exchange of variables among SCs will cause high energy consumption and processing latency. This paper is focused on the reduction of communication overhead for the distributed processing. To this end, two algorithms, Augmented Lagrangian based Cooperative Estimation (ALCE) and Priority-aided ALCE (PALCE) will be presented. In ALCE a new efficient approach is adopted to achieve parallel processing among all SCs, which needs fewer variables to be exchanged. Thus, a considerable amount of overhead will be saved. However, theALCE algorithm is not robust when applied to a network with erroneous backhaul (BH) links, therefore a variant of this approach termed PALCE is proposed using a priority oriented principle to enhance the robustness and maintain low amount of information exchange. The proposed algorithms are investigated by means of error rate and communication overhead showing significant improvement in estimation performance compared to state of the art algorithms.