Dissertation Defense

Rethinking Wireless: Building Next-Generation Networks

Eugene Chai

We face a growing challenge to the design, deployment and management of wireless networks that largely stems from the need to operate in an increasing spectrum sparse environment, the need for greater concurrency among devices and the need for greater coordination between heterogenous wireless protocols. Unfortunately, our current wireless networks lack inter- operability, are fixed-function deployments, and omit easy programmability and extensibility from their key design requirements.

In this dissertation, we study the design of next-generation wireless networks and analyze the individual components required to build such an infrastructure. Re-designing wireless architecture must be undertaken carefully due to the need to balance new and coordinated multipoint (CoMP) techniques with the backwards compatibility necessary to support the large number of existing devices. These next-generation wireless networks will be predominantly software-defined and will have three components: (a) a wireless component that consists of software-defined radio resource units (RRUs) or access points (APs); (b) a software-defined backhaul control plane that manages the transfer of RF data between the RRUs and the centralized processing resource; and (c) a centralized datacenter/cloud compute resource that processes RF signal data from all attached RRUs.

Towards that goal, this dissertation covers four key pieces of work:

1. Rodin demonstrates an approach to integrate existing wireless devices into the new wireless architecture by bringing spectrum agility to COTS devices.

2. Aileron is a novel approach to spectrum coordination that is necessary for fast and efficient cognitive spectrum access

3. Sidekick efficiently and effectively aggregates disparate blocks of spectrum from different wireless APs, and

4. Spiro illustrates a novel backhaul management and compression technique to enable the transport and processing of coordinated multipoint RF data over commodity Ethernet networks.

The mechanisms and techniques presented in this dissertation serve as the fundamental building blocks for future wireless networks.

Sponsored by

Kang G. Shin