Printable Version of this PageHome PageRecent ChangesSearchSign In
In his thesis, "Constructing Scalable Wireless Networks with Directional Antennas", Michael Neufeld introduces several new techniques for wireless networks utilizing smart antennas. Moreover, he takes the pragmatic approach that any changes made must be backwards compatible with current technology. In addition to this, he gives an in-depth overview and analysis of prior work.

Smart antennas, at a high-level, are antennas
that are able to dynamically reconfigure their radiation pattern to form a beam in some direction(s) and nulls in others. Using these "adaptive arrays" to reduce interference and increase performance in wireless networks is a popular topic. Many people have proposed new smart-antenna-centric protocols to replace the existing dominant protocols (IEEE 802.11x). Neufeld argues, however, that most prior work makes unrealistic assumptions about the way antennas work and are not backwards compatible with existing technology.

The main contributions of his thesis are: A taxonomy for directional MAC (medium access control) protocols, a thorough performance analysis of DNAV, one popular directional MAC, a technique for reducing RTS/CTS (i.e. virtual-carrier-sensing) overhead using a saturation counter to selectively filter some handshaking, a system for picking antenna patterns to cover relevant neighbours, and a technique for locating neighbours (i.e. Angle of Arrival or AoA) by steering a null. All of these contributions are experimentally tested via simulation.

From my point of view, the most interesting and valuable part of this thesis is the pragmatic approach. I do, however, have a couple of criticisms. Firstly, I would have liked to see a real-world implementation in addition to the simulation. Simulators do an especially bad job of modelling RF (radio-frequency) characterstics, hence, all simulation of wireless networks is slightly suspect. Neufeld admits in his conclusion that he used an overly simplistic antenna model in the simulator - this should be fixed (real-world implementation would also solve this gap). I am a bit confused by marketing the algorithms as especially good for anarchic deployments like community wireless networks - most community wireless groups couldn't afford the hardware that would be required to implement these algorithms. Finally, the tightly-synchronized protocols, which are a bit "futuristic" seem a bit of a deviation from the pragmatic focus.

All in all it was an interesting read. I'm glad I read something close to my field as it's clear the thesis required a good deal of background knowledge to make sense of (at least the details). The most useful lesson here was to get an idea about the quantity of work that makes up a Ph.D. thesis, and also a Ph.D. thesis at CU. I think it'd be worthwhile to read a few more thesis, from various institutions, fields, and perhaps even different times to get an idea of the range of approaches.

Last modified 1 December 2007 at 11:37 pm by caleb