Distributed Robust Power Minimization for the Downlink of Multi-Cloud Radio Access Networks
Permanent link to this recordhttp://hdl.handle.net/10754/623823
MetadataShow full item record
AbstractConventional cloud radio access networks assume single cloud processing and treat inter-cloud interference as background noise. This paper considers the downlink of a multi-cloud radio access network (CRAN) where each cloud is connected to several base-stations (BS) through limited-capacity wireline backhaul links. The set of BSs connected to each cloud, called cluster, serves a set of pre-known mobile users (MUs). The performance of the system becomes therefore a function of both inter-cloud and intra-cloud interference, as well as the compression schemes of the limited capacity backhaul links. The paper assumes independent compression scheme and imperfect channel state information (CSI) where the CSI errors belong to an ellipsoidal bounded region. The problem of interest becomes the one of minimizing the network total transmit power subject to BS power and quality of service constraints, as well as backhaul capacity and CSI error constraints. The paper suggests solving the problem using the alternating direction method of multipliers (ADMM). One of the highlight of the paper is that the proposed ADMM-based algorithm can be implemented in a distributed fashion across the multi-cloud network by allowing a limited amount of information exchange between the coupled clouds. Simulation results show that the proposed distributed algorithm provides a similar performance to the centralized algorithm in a reasonable number of iterations.
CitationDhifallah O, Dahrouj H, Al-Naffouri TY, Alouini M-S (2016) Distributed Robust Power Minimization for the Downlink of Multi-Cloud Radio Access Networks. 2016 IEEE Global Communications Conference (GLOBECOM). Available: http://dx.doi.org/10.1109/glocom.2016.7841744.
SponsorsThe work of O. Dhifallah, T. Y. Al-Naffouri and M. -S. Alouini was supported by KAUST. Hayssam Dahrouj would like to thank Effat University in Jeddah, Saudi Arabia, for funding the research reported in this paper through the Research and Consultancy Institute.