Self-configuration algorithms for mobile ad hoc networks

Amit J Shirsat, Purdue University

Abstract

This thesis presents a localized distributed algorithm for topology control of mobile ad hoc wireless networks by engaging control over the local placements of the nodes. We allow a subset of nodes to adjust their positions locally and self-configure a topology that satisfies certain network graph properties. The nodes selectively put themselves into a power-saving passive mode if they are not required to form the part of the routing backbone for the ad hoc network. The resulting network has better life-time characteristics achieved through node power savings, lower interference and efficient routing paths. The concept of self-adjustment is referred to as localized mobility and is designed to guarantee several desirable network properties: (1) The network maintains a backbone of active nodes which form a connected dominating set over the unit disk graph. (2) The number of such active nodes in any given unit area is bounded by a small constant. As the network size grows, their overall count is asymptotically optimal. (3) The live nodes maintain a planar routing graph for the use of stateless geographic forwarding protocols. (4) The graph described by the live nodes is a spanner over the topological metric. These properties are known to alleviate the issues of interference, routing overheads and power consumption. The thesis develops a local distributed geometric algorithm to detect the geographical voids formed by the regions with low node densities. The hole is defined as the closed polygon comprised of the nodes and links which enclose the voids. In contrast to the approach of identifying the polygonal loop by traversing its boundary, we employ geometric properties to design the local hole detection algorithm. We support this claim using simulations over random distributions of nodes. We describe Geographic Location Aware MANETs (GLAM). GLAM is a protocol that integrates greedy routing scheme with location tracking. It is a location service that hashes the unique ids of nodes to distributed virtual coordinates organized in an abstract hierarchy. We provide a scheme for mapping the virtual coordinates to actual nodes (location servers) using the hole detection framework. The protocol is distributed, scalable, fault-tolerant and self-configurable.

Degree

Ph.D.

Advisors

Bhargava, Purdue University.

Subject Area

Computer science

COinS