Cloud radio access network (CRAN) architecture has emerged as a transformative ar-chitecture for the next generation of mobile cellular networks. Unlike in traditional radio access networks (RANs) where baseband units (BBUs) are distributed in base station (BS) sites, in CRANs they are centralized at a single shared site known as the BBU pool leaving only remote radio heads (RRH) at the BS sites. Centralizing BBUs in CRAN is associated with advantages like reduced interference between BS sites, improved spectral e ciency and it requires less set up time. However, the fronthaul link between the RRHs and the BBU pool is usually long such that signals moving between the BBU pool and the RRHs experience high fronthaul link latencies. The fronthaul links in CRAN there-fore need to be made shorter by bringing BBUs closer to the RRHs in order to reduce the time signals spend while moving between the RRHs and the BBU pool. This can be achieved by deploying edge clouds in cloud radio access networks. Edge clouds are distributed within the radio access network, therefore placing the BBUs in edge clouds can reduce the length of the fronthaul links thus reducing on the fronthaul link latencies. This thesis aims at developing an algorithm that is capable of optimizing location of edge clouds with BBUs in cloud radio access network so as to lower the fronthaul link latencies. Furthermore, this thesis investigates how the optimally placed edge clouds with BBUs are utilized. Moreover, power consumed by BBUs in edge clouds and cost of setting up the CRAN with edge clouds were also investigated. Particularly, a nonlinear edge cloud location optimization problem was formulated and solved by fuzzy c-means and heuristic genetic based algorithms. Simulation results show that with planned and optimal placement of BBU in CRAN, the overall fronthaul link length, latencies and cost of ownership are desirably lower than those experienced when the BBUs are placed at a centralized location. Results also show that there is an optimal number of edge clouds required for every size of CRAN depending on the capacity of the edge clouds used and the remote radio head density. In this thesis, results also show that response time ex-perienced by user equipment nearer to the baseband units in edge clouds, is less than that experienced by user equipment far away from the baseband units. Simulation results further reveal that baseband units placed in edge clouds are more e ciently utilized than those placed at a centralized location.

show more