A Bee-Inspired Power Aware Routing Protocol for Wireless Ad Hoc Sensor Networks

Abstract

Design and development of power-aware, scalable and performance efficient routing protocols for Wireless Sensor Networks (WSNs) is an active area of research. In this dissertation, we show that insect colonies based intelligence – commonly referred to as Swarm Intelligence (SI) – provides an ideal metaphor for developing routing protocols for WSNs because they consist of minimalists, autonomous individuals that through local interactions self-organize to produce system-level behaviors that show life long adaptability to changes and perturbations in an external environment. In this context, we propose a new routing protocol for WSNs – BeeSensor – inspired by the foraging principles of honey bees. We follow a three phase novel protocol engineering cycle. In the first phase, we study the foraging principles of a bee colony and utilize the inspirational concepts to develop a distributed, simple and energy-efficient routing protocol for WSNs. We then evaluate and compare the performance of this protocol with existing classical and SI-based WSN protocols. The simulation results demonstrate that BeeSensor consistently outperforms the existing well-known protocols in terms of packet delivery ratio and energy efficiency. However, its performance degrades slightly as the network size is increased. To gain more insights into the parameters governing the behavior of BeeSensor in large-scale networks, in the second phase, we develop a generic mathematical evaluation framework to model two key performance metric of an ad hoc routing protocol: routing overhead and route optimality. We then develop specific routing overhead and route optimality models of the BeeSensor protocol. The metric models unfold several interesting insights about the performance of BeeSensor in large-scale networks. For instance, with an increase in the average hop length, route discovery probability of BeeSensor decays exponentially. We also model the reliability of packet delivery of the BeeSensor protocol. The model shows that the reliability of packet delivery is a concave function of the total number of paths. Therefore,maintenance of a set of paths beyond a certain threshold limit does not result in a proportional increase in the packet delivery ratio. Based on the insights inferred through the formal modeling, we revise the design of BeeSensor protocol in the third phase. To conclude this dissertation, we per- form simulation studies – using prowler simulator – to analyze and compare the performance of the final BeeSensor design with existing protocols. In the first set of experiments, we compare its performance with SI-based energy-efficient WSN protocols. The simulation results demonstrate that BeeSensor outperforms its com- petitors in all assumed scenarios and metrics. We then implement the BeeSensor protocol in NS-2 simulator to further investigate its performance in mobile networks and large-scale static sensor networks. The results clearly show that BeeSensor not only performs well in large-scale networks, but is equally good in MANETs as well. Therefore, BeeSensor is a viable protocol for hybrid ad hoc networks.