Saving energy in base station with non real time operation system in Cloud-RAN

Recently, it has been proposed that non-realtime operating systems and cloud-computing approach could be utilized for base band processing of a cellular system. In this paper we analyze the trade off between energy consumption and system performance achieved by adjusting the cloud-computer speed. By changing the processor frequency, we aim at minimizing energy consumption while still meeting the quality targets of the communication system. In considered system each packet has to be transmitted at certain transmission deadline. Because of non real time operating system and variable processing speed some of the baseband packets will not meet their deadline. We take an unorthodox approach and consider the processor frequency scaling to be part of the communication channel. In our system late baseband packets, i.e. lost subframes, are handled by automatic repeat request. We use Markov models for analyzing the channel and illustrate how energy saving impacts the channel throughput. Our results show the tradeoff between the lost packets and energy consumption of the cloud-server. The results are also mapped to the channel packet throughput.