Abstract

In this talk, I will describe a number of selected topics that we are currently studying in my Wireless Networks Lab at Cornell University. In particular, I will breeze through topics such as topological design of ad hoc and sensor networks, routing in ad hoc and sensor networks, and security in ad hoc and sensor networks. In the remaining part of the talk, I will concentrate on two topics: application of sensor networks to animal habitat monitoring and extension of wireless 2D networks to 3D networks.

Animal Habitat Monitoring: To study the application of the sensor networking technology to animal habitat monitoring, we have introduced the Shared Wireless Infostation Model (SWIM). As opposed to data forwarding over pre-established paths, SWIM creates, what we referred to as, new store-carry-forward networking paradigm, using virtual paths for topology control in networks with temporally and spatially intermittent connectivity. SWIM exploits mobility in the system, producing those virtual links to cooperatively propagate information to the destinations in a timely manner. This new networking paradigm has a broad range of applications, especially in the area of telemetry collection and sensor networks. It could be used, for example, for several types of animal tracking systems, for medical applications with micro-sensors, or to relay traffic and accident information to the public emergency system using the mobility of the vehicles. SWIM is able to reduce delay in delivery of packets at the expense of small increased storage at the nodes and moderate increase in energy consumption. In other words, SWIM improves the overall capacity-delay tradeoff and optimizes the energy-delay tradeoff. These and other tradeoffs are examined in this talk.

3D Wireless Networks: Most wireless terrestrial networks are based on two-dimensional (2D) design, although in reality, such networks operate in three-dimensions (3D). Since most often the size (i.e., the length and the width) of such networks is significantly larger than the differences in the third dimension (i.e., the height), the 2D assumption is somewhat justified and usually it does not lead to major inaccuracies. However, in some environments, this is not the case; underwater, atmospheric, or space communications being such apparent examples. In fact, recent interest in underwater acoustic ad hoc and sensor networks hints at the need to understand how to design networks in 3D. Unfortunately, the design of 3D networks is surprisingly more difficult than the design of 2D networks. For example, proofs of Kelvin's conjecture and Kepler's conjecture required centuries of research to achieve breakthroughs, whereas their 2D counterparts are trivial to solve. In this paper, we consider the coverage and connectivity issues of 3D networks, where the goal is to find a node placement strategy with 100% sensing coverage of 3D space, while minimizing the number of nodes required for surveillance. Our results prove that the use of the Voronoi tessellation of 3D space to create truncated octahedral cells results in the best strategy. In the truncated octahedron placement strategy, the transmission range must be at least 1.7889 times the sensing range in order to maintain connectivity among nodes. If the transmission range is between 1.4142 and 1.7889 times the sensing range, then a hexagonal prism placement strategy or a rhombic dodecahedron placement strategy should be used.