I am an Research
I'm excited about my research and I want you to be excited too. The proliferation of wireless technologies in the past decade has led to near-ubiquitous computing at almost every level and opened up both problems and opportunities at every layer of the network stack. One of the major issues as networks become larger and denser is scalability: How does one efficiently use the limited resources available given increasing number of wireless devices competing for the same medium? We look at addressing the issue of scalability at the routing layer. To tackle the issue of scalability, we identify three trends in computer networking:
- The move toward unstructured, flat, yet scalable networks - Routing in wired, wireless, and overlay networks all began with flood-based protocols (ethernet, DSR, DSDV, AODV, Gnutella, etc). While for a small number of hosts, this was effective and simple, as the number of nodes grew, flooding the network became a major headache. We then saw a move toward hierarchical approaches that limited flooding by building hierarchies (routers, HSLS, ZRP, Kazaa, etc.) and then structured approaches that mapped node information to a specific structure (VRR, CHORD, CAN, etc.). While effective in their own right, these approaches either required the maintenance of a structure or had single points of failures. In recent years, there has been a move from hierarchical and structured approaches to unstructured, flat, yet scalable approaches that leverage random walks to route from source to destination.
- The leveraging of directional antennas for increased capacity - Yi et al. showed analytically in 2005 that the capacity gains by using directional antennas are dramatic - increasing capacity by a factor of 50x even with only antenna spread being 1/8th of a traditional omni-directional antenna coverage. Other work with sector antennas in cell base stations show a 1.714 gain. It becomes interesting, therefore, to leverage directional antennas in efficient medium reuse.
- The push toward complimenting traditional RF communications with higher bandwidth devices such as Free-Space-Optical Transceivers - Current RF-based ad hoc networks utilize omni-directional antennas, consume high power, are constrained by low bandwidth, and are highly error-prone. By contrast, free space optical (FSO) transceivers such as high-brightness LEDs (HBLEDs) are very low cost ($2-$5/transceiver package), highly reliable (10 year lifetime) and low power (100 microW for 10-100 Mbps), operate in license-free frequency bands, transmit at relatively higher bandwidth, and are more secure and spatially efficient due its transmission’s directional nature. The issue with FSO transcievers, however, are that they are very directional in nature resulting in the need for alignment.
Given these trends, it becomes interesting to investigate how to leverage directionality to route packets scalably in unstructured, flat, and even highly mobile networks. To find out more about how we tackle this issue, feel free to peruse my Thesis [PDF in Thesis Format | Condensed PDF for Printing], Candidacy Research Proposal, Candidacy Presentation, the posters below, or our publications.
Routing in Wireless Networks with Directional Communications Methods (Candidacy Document) [ PDF | Presentation ]