Bond: unifying mobile networks with named data

Much research has been devoted to freeform networks – that is, mobile, multi-hop wireless networks where the nodes move freely and unpredictably. This includes connected mobile ad hoc networks (MANETs) as well as disconnected networks such as vehicular ad hoc networks (VANETs) and opportunistic networks. To date, the fundamental goal of such research has primarily been data delivery between node pairs as identified by their IP addresses. Despite a rich literature of solutions to this problem extending back over 30 years, no sufficiently general, agreed-upon standard exists today. This dissertation returns to the drawing board to rethink how to align modern data communications with the highly dynamic nature of freeform wireless networks and the broadcast nature of wireless channels. This leads to an entirely different approach, which is based on Named Data Networking (NDN). NDN asserts that data, not hosts, are the central element of modern networks. The realization of this approach and core contribution of this dissertation is the Broadcast-Only Named Data (BOND) protocol, the first complete protocol specification based on NDN concepts. BOND has the following unique properties: (1) BOND supports all freeform networks, connected and disconnected, without so much as a configuration change. (2) BOND communication is entirely based on named data. (3) BOND's performance does not depend on the level of physical mobility in the network. (4) BOND uses exclusively broadcast communication for all packets, allowing for more efficient use of the wireless channel and more flexibility in the selection of MAC layer protocols. Under simulation, BOND significantly outperforms existing protocols for both connected and disconnected networks. In all scenarios, whenever nodes are mobile, BOND consistently has the lowest roundtrip times and the highest response rate of any protocol tested, with overhead consistently ranking among the lowest observed.