Protocols for mobile ad hoc networks

A mobile ad hoc network is a collection of mobile terminals equipped with wireless transceivers that form an autonomous network without any pre-planned infrastructure or centralized administration. Because of its constantly changing topology, the major challenging issues in such network are the medium access control (MAC) and routing protocols. The purpose of our research is to design some novel protocols for mobile ad hoc networks and evaluate their performances. In this dissertation, we first develop an accurate model for wireless packet transmission and reception in an ad hoc network environment. We implement these models in the ns-2 network simulator to improve its accuracy of capturing the effect of radio interference at the receiver. The modified ns-2 simulator is used as a tool for evaluating the performances of several ad hoc network protocols. We develop a new directional antenna based MAC protocol with power control (DMACP), in which a sender maintains a minimum transmission power level for effective transmission of data packets using directional antennas. Simulation results show the throughput advantage and the savings in the average consumed power in the network using the proposed protocol. In addition, we propose a hybrid routing protocol, using a pre-emptive route discovery to replace an existing route by a shorter route when the route has been used for a given interval of time. It effectively reduces the average end-to-end delay in packet transmission as well as the routing overhead in the network.; This research is also extended to wireless sensor networks, which are ad hoc networks of a large number of wireless sensors. Constraints in size, cost, and energy available at the sensor nodes introduce additional challenges in designing protocols for such networks. An additional requirement is that the nodes need to be able to locate themselves in order to coordinate amongst themselves. We present a new technique, for location discovery, employing an angle-of-arrival estimation technique to let a sensor node determine its location in the network. The proposed localization scheme exhibits excellent accuracy and requires very little additional hardware at the sensor nodes. Finally, we develop a distributed signal detection and target location framework for wireless sensor networks. The framework is based on nonparametric signal detection methods, which show very little variation under different noise levels and characteristics. It also has special features to minimize the communication cost. Target detection is accompanied by locating the detected target by analyzing the information sent by the sensors. We develop a special technique for target location estimation that uses off-line measurements and iterative computations at a centralized control unit that monitors the data transmitted from the wireless sensors. Simulation results demonstrate the accuracy of this target detection and location estimation scheme.