NeTS-NBD: Physical Layer Dependent Neighbor Discovery and Topology Management in Ad hoc Networks

University Of California-Riverside, Riverside CA

Investigators

Abstract

Physical Layer Dependent Neighbor Discovery and Topology Management in Ad hoc Networks Award 0626912 S. Krishnamurthy, M. Faloutsos and N.E. Young Under realistic wireless channel representations, the project re-examines a node's neighborhood with three physical layer technologies: (a) rate controllable transmissions, (b) directional antenna capabilities, and (c) Multi-Input Multi-Output (MIMO) capabilities. To limit the overhead incurred with these technologies, topology control is incorporated. Specifically, with rate control, the best trade-offs between rate and range are achieved. In cases (b) and (c), per-neighbor channel state information (CSI) needs to be maintained. Maintaining CSI for large numbers of neighbors is overhead intensive. The topology control framework limits the CSI by having each node communicate with only a carefully chosen sub-set of neighbors while ensuring connectivity and short end-to-end paths. For each technology, the project develops: (i) centralized algorithms to study the trade-offs between the operational parameters and (ii) deployable distributed neighbor discovery protocols. The schemes are evaluated via simulations and analyses. In addition, the rate control mechanisms are implemented on an experimental testbed at UCR. Impact: Neighbor discovery and maintenance are critical in any self-organizing wireless network. The project addresses these aspects with realistic physical layer representations. This is critical for the deployment success of wireless networks. Expected Results: New cross-layer protocols that combine neighbor discovery and topology management will be developed. The centralized approaches will provide fundamental bounds on the performance. The distributed protocols will be amenable to implementation. The developed software will be made available on a project website.