Research Interests
Scalable Routing Protocols in Mobile Ad Hoc Networks
	Due to the rapid growth of wireless technology and the increasing popularity of mobile networking systems, there has been a growing interest in the capabilities of ad-hoc networks connecting mobile phones, PDAs and laptop computers. A Mobile Ad hoc NETwork (MANET) consists of a cluster of mobile hosts without fixed base station or any wired backbone infrastructure. In MANETs, two mobile hosts can communicate directly with each other if they are located within their radio range; otherwise, packets are relayed via intermediate hosts located between the two hosts. Thus, each mobile host must act as a mobile router. Another interesting challenge is scarce battery power. Since most mobile hosts are operated using batteries, and a battery life is not expected to increase significantly in several years, energy efficiency is a critical issue in ad hoc networks. Because of the extremely random and sporadic nature of MANETs, the issue of delivering packets between any pair of source and destination hosts becomes a very formidable challenge. Although a number of proposals have been made to solve the routing problem in MANET, few of these proposed algorithms can scale well in a very large MANET. Thus, the proposed research will be focused on the development of position-based scalable routing protocols for very large MANETs. The position-based routing protocol uses greedy forwarding to forward packets to nodes that are always progressively closer to the destination. Greedy forwarding fails when no neighbor is closer than the current node to the destination (i.e., the only path requires that one move temporarily farther away from the destination). Although the stateless strategy of geographic forwarding reduces routing overhead caused by topology updates, its lack of global topology knowledge prevents a mobile node from predicting topology holes as well as forwarding failures. Even though there are some methods proposed to route around the holes in the literature, they are used only after the geographic forwarding fails, incurring extra cost in detecting forwarding failure and searching for new routes. Therefore, we will investigate how to avoid or recover greedy forwarding failure efficiently.
Latency-Aware Medium Access Control (MAC) Protocols for Large-Scale Wireless Sensor Networks
	A Wireless Sensor Network (WSN) is a self-organized and self-configured network consisting of hundreds or thousands of battery-powered sensors with very limited processing and communications capabilities. Many sensor applications are challenged by the absence of real-time environmental monitoring systems. When an event of interest is detected, sensors should be able to report the local processing results in real time so that appropriate action can be taken promptly. However, most energy-efficient MAC protocols in the literature toggles between active and sleep modes in order to reduce energy consumption caused by idle listening. This periodic listen and sleep design reduces energy consumption, however it increases latency considerably since a sender must wait for the receiver to wake up before it can send data out. When each node strictly follows its sleep schedule, there is a potential delay on each hop, and so packet latency would be proportional to the number of hops between a source and a destination. Due to the trade offs between energy-efficiency and QoS capability, many energy-efficient MAC protocols for WSNs have several shortcomings such as increased packet latency, jitter and loss, system throughput degradation, and slow adaptation to network traffic dynamics. Our research objectives, therefore, include the development of new methods for forwarding sensor data within a specified latency constraint without sacrificing energy efficiency.
Energy-Efficient Data Gathering Protocols for Large-Scale Wireless Sensor Networks
	The increasing interest in video and imaging sensors applications that demand certain end-to-end performance guarantees has posed additional challenges. Such WSN applications require both energy and QoS aware network protocols in order to maintain an appropriate level of service quality. The QoS requirements in WSN have unique characteristics. First, WSN have only very limited memory capabilities that do not allow for storing session information. So any IntServ related approach will not be suitable. Second, many applications of WSNs have an asymmetric many-to-one data pattern mainly from sensor nodes to a sink node. Third, unlike in traditional networks, a session in WSN may contain several nodes jointly collecting data. Due to these unique characteristic, QoS protocols in other types of networks including wired Internet, mobile ad hoc networks and wireless cellular networks are not directly applicable to WSNs without considerable modifications. Therefore, we will develop and implement energy-efficient QoS-aware data gathering protocols that maintain the quality of service above a desired level through the wireless sensor network.