Speaker: Taewoon Kim, ECpE Graduate Student
Advisor: Daji Qiao and J. Morris Chang
Title: Multi-Stage Stochastic Programming Approach for Resource Optimization in C-RAN
Abstract: Cloud/centralized radio access network (C-RAN) is considered as a promising cellular networking architecture for the next generation 5G network. C-RAN shifts most of the tasks that have been operated by the base stations (BSs) to the central cloud computing resource pool. The centralized concept along with the software defined radio and network function virtualization technology brings plenty of benefits, such as reduced total cost of ownership, better resource utilization and enhanced cooperation among BSs. Thus, it has gained much attention from industry as a promising way to increase the network operators’ profit. However, to take full advantage of such potential, and eventually to yield a large profit gain from a network operator’s perspective, it is crucial that the computing and networking resources of C-RAN are optimally scheduled under the inherent uncertainties, such as users’ mobility and varying service demand. In this study, we propose an optimal resource optimization for energy-aware C-RAN that schedules the pool of computing and networking resources such that a network operator can maximize its profit without degrading the quality of service for users. We propose to use a multi-stage stochastic programming approach by which the optimal resource allocation can best respond to the underlying uncertainties, i.e., users’ mobility and their varying service demand. The evaluation results show that the proposed model yields a higher profit gain than other solutions that do not or partially consider such uncertainties.