Al-dulaimi, Anwer and Al-rubaye, Saba and Ni, Qiang (2019) Energy efficiency using cloud management of LTE networks employing fronthaul and virtualized baseband processing pool. IEEE Transactions on Cloud Computing, 7 (2). pp. 1-12. ISSN 2168-7161
TCC_final.pdf - Accepted Version
Available under License Creative Commons Attribution-NonCommercial.
Download (1MB)
Abstract
The cloud radio access network (C-RAN) emerges as one of the future solutions to handle the ever-growing data traffic, which is beyond the physical resources of current mobile networks. The C-RAN decouples the traffic management operations from the radio access technologies, leading to a new combination of a virtualized network core and a fronthaul architecture. This new resource coordination provides the necessary network control to manage dense Long-Term Evolution (LTE) networks overlaid with femtocells. However, the energy expenditure poses a major challenge for a typical C-RAN that consists of extended virtualized processing units and dense fronthaul data interfaces. In response to the power efficiency requirements and dynamic changes in traffic, this paper proposes C-RAN solutions and algorithms that compute the optimal backup topology and network mapping solution while denying interfacing requests from low-flow or inactive femtocells. A graph-coloring scheme is developed to label new formulated fronthaul clusters of femtocells using power as the performance metric. Additional power savings are obtained through efficient allocations of the virtualized baseband units (BBUs) subject to the arrival rate of active fronthaul interfacing requests. Moreover, the proposed solutions are used to reduce power consumption for virtualized LTE networks operating in the Wi-Fi spectrum band. The virtualized network core use the traffic load variations to determine those femtocells who are unable to transmit to switch them off for additional power savings. The simulation results demonstrate an efficient performance of the given solutions in large-scale network models.