UC Santa Barbara

As the wireless medium has become the primary source of communication and Internet connectivity, and as devices and wireless technologies become more sophisticated and capable, there has been a surge in the capacity demands and complexity of applications that run over these wireless devices. To sustain the volume and QoE guarantees of the data generated, the opportunity and need to rethink wireless network design across all the layers of the protocol stack has firmly emerged as a solution to enable the timely and reliable delivery of data, while handling the inherent challenges of a crowded wireless medium, such as congestion, interference, and hidden terminals.

The research work presented in this dissertation builds efficient solutions and protocols with a theoretical foundation to address the challenges that arise in rethinking wireless network design. Example challenges include managing the overhead associated with complex systems. My work particularly focuses on the opportunities and challenges of sophisticated technology and systems in emerging wireless networks. I target the main thrusts in the evolution of wireless networks that create significant opportunity to achieve higher theoretical capacity, and have direct implications on our day-to-day wireless interactions: from enabling multifold increase in capacity in wireless physical links, to developing medium access techniques to exploit the high speed links, and making the applications more bandwidth efficient. I build deployable, and resource-aware wireless systems that exploit higher bandwidths by leveraging and advancing diverse research areas such as theory, analysis, protocol design, and wireless networking. Specifically, I identify the erroneous assumptions and fundamental limitations of existing solutions in capturing the true and complex interactions between wireless devices and protocols. I use these insights to guide practical and efficient protocol design, followed by thorough analysis and evaluation in testbed implementations via prototypes and measurements. I show that my proposed solutions achieve significant performance gains, at minimum cost to overhead.