Cloud-driven architectures for next generation small cell networks

The ability to aggressively reuse spectral resources has made small cells a key component in catering to the growing traffic demand in cellular networks. While efficient resource management solutions are needed to address the challenging interference problem in small cell networks, the deployment and management of small cells themselves incurs capital and operational expenses for operators. Cloud-driven radio access networks (C-RAN) have been proposed as a cost-effective means for the realization of small cells. Unlike typical RANs where the baseband units (BBUs) and the radio units are situated together, the C-RAN concept migrates the BBUs to a datacenter (i.e., the cloud), while providing a transport network (called the "front-haul") to the remote antennas called remote radio heads (RRHs). In this talk, we will discuss how such a physical decoupling of BBU and RRH allows C-RAN to enable better interference (resource) management solutions and hence provide increased network capacity. More importantly, we will focus on the component that is unique to a C-RAN, namely its front-haul network that maps and transports the signals from the BBUs to the RRHs. We will show how different logical configurations of the front-haul manifest as different transmission strategies on the access network and discuss why they are critical in realizing the complete potential of a C-RAN. Then, with the help of a small scale end-to-end experimental C-RAN test-bed deployed in an enterprise environment, we will present the design of a dynamically reconfigurable, software-defined front-haul network for C-RAN that tailors front-haul configurations to the observed traffic profile in the network. This allows it to provide not just improved capacity but also seamless mobility, reduced compute resource consumption (in BBU pool) and front-haul customization (for different operators and access technologies) simultaneously. We will conclude by discussing some of the limitations of C-RAN architectures in general as well as their current state in the industry.