The growth and value of the Internet today are fueled by a variety of wireless edge networks including WiFi access points in homes and public places, wireless local area networks (WLANs) in workplaces, cellular data services such as 3G, multi-hop wireless mesh networks, wireless sensor networks, mobile ad hoc networks (MANETs), and delay-tolerant networks (DTNs). This diversity poses fundamental challenges to traditional network protocols that are fragile under highly fluctuating loss, delay, and topology characteristics, and completely break down in disruption-prone environments. The diversity increases management complexity?a network administrator must pick from an intimidating number of protocol options and optimizations tuned to one environment have to be re-engineered in a different environment?stifling innovation and the long-term value of interconnecting diverse wireless edge networks.
The goal of this proposal is to architect, analyze, and implement a simple network protocol stack that ensures robust performance across diverse wireless edge networks. The key insight is to incorporate uncertainty as a first-class design concern. We adopt an unusual approach: instead of trying to make protocols in otherwise well-connected environments robust to intermittent fluctuations or disruptions, we design for an extreme point in the design space?an always-partitioned network or a DTN?and work back from those insights to ensure robust performance of protocols even in well-connected mesh networks.