Relay-based cellular networks are likely to play an important role in the race for energy efficiency. However, potential gains greatly depend on how relay stations are deployed within the cell. Using a geometrical model for energy-efficient analysis, we investigate the impact of the number and location of relays on energy consumption, and its dependence on the relay coding scheme and the propagation environment, i.e. the pathloss and the line-of-sight conditions. In addition to the transmit energy, we account for the economic cost of relay deployment, as well as the overhead energy dissipated at each relay stations due to data processing and network maintenance. We then bring out four key trade-offs which balance the cost and flexibility of relay deployment, the energy efficiency and the coverage extension.
I. INTRODUCTIONA major challenge for next generation cellular systems is to provide high-data-rate services in an energy-efficient manner to all users, including at cell-edge. In this regard, relaying is considered as a promising option for significant gains for both coverage extension and energy consumption. The issue of relay deployment embraces both the number of relay stations and their position within the cell.A wide literature proposes analysis and algorithms for optimal relay placement regarding coverage [1-3], outage probability [4], energy consumption [5] or spectral efficiency [6]. However, obtained results are only valid for the considered radio propagation model and coding scheme. Moreover, relay deployment is also constrained by the urban topography and by administrative by-laws, which implies that it may be impossible to locate a relay station at its optimal location. Therefore, the performance analysis of sub-optimal relay placement is relevant and the deployment flexibility should be considered as well, i.e. the possibility for a designer to have a wide range of potential relay locations without harming the network performance too much.The system performance is also greatly affected by the number of relays in the cell, as shown in [5][6][7][8]. These works converge to the existence of some threshold beyond which adding more relays does not improve the system performance significantly. With this perspective, a trade-off between capacity enhancement and deployment cost has been highlighted in [7]. Though, trade-offs between energy efficiency and relay deployment have been little explored in the literature.