Bringing mobile relays for wireless access networks into practice - learning when to relay

Adding fixed relay nodes (RNs) to wireless access networks requires additional costly infrastructure. Utilising mobile RNs, i.e., user terminals that relay signals intended for other users being the destination nodes (DNs), is an appealing cost-effective solution. However the changing node topology increases the required signalling for relay selection (RS). The signalling overhead consists of control signals that need to be exchanged between the RNs, the source node (SN) and the DN, in order to achieve the objectives of cooperation. To reduce signalling without penalizing performance we propose a three-step approach exploiting statistical knowledge on the likelihood of attaining performance gains by using RNs as a function of the node position (position of DNs and RNs). In the first step only the cell DNs that are likely to gain from relaying request the assistance of RNs. In the second step, for each DN that requests relaying, a limited set of RN candidates is formed. These decisions are made with the aid of thresholds applied to inter-node distances whose values are based on the acquired statistical knowledge. In the final step, RN candidates feed back the relevant channel state information to the SN which performs RS. Furthermore we investigate the attained gains from mobile RNs as a function of the fading environment and we show that mobile RNs can help overcome the effects of severe fading. Index Terms Mobile relays, cooperative diversity, wireless access networks, relay selection, limited signalling.