zaitcev.mee.nu

According to the syllabus, blogging the class is a requirement, and counts towards a big 50% block of the grade that deals with the main project (the final is a 8-10 pages paper on it). Sounds like a perfect opportunity to use the Meenuvia.

First meeting was mostly a class overview by Prof. Patrick Bridges, and the class is mostly a project, after we start up with readings on MapReduce etc. Project areas are:

• Distributed computation to collapse the sensor data for storage, save "interesting" things, discard the rest.
• Cellphone-based applications: MegaGaydar (Prof. Wenbo He's pet project), traffic congestion (using cellphone density as a proxy).
• Power grid (fridges that cooperate, kinda fishy stuff).
• Traffic systems (traditional, government and ground-based).

Alternative projects are allowed too, as long as they are distributed and/or embedded.

The most interesting project is definitely the gaydar thing. I'm thinking about something like Nintendo Pictochat, only running constantly on your cellphone (over Bluetooth), and on your Nokia 880/XO/netbook/etc. (over WiFi). This should include some mesh capability, cross-media routing (with Bluetooth enabled netbooks serving neighbouring cellphones), etc. Prof. Bridges' thoughts went in a different direction: he wants cellphones routing each other web access (e.g. in case Verizon coverage works but not AT&T). IMHO that's kinda low level, but he probably has a better feel for constraints of a one-term project. Heck a regular Linux device driver gestates for 6 months, and here we only have 2 to 3.

The most important part that makes MegaGaydar interesting is how it's a consumer oriented application not sponsored by an outside entity. Everything else (save for the utopical fridges talking) is a pretty hideous Big Brother kinda stuff. Why don't we have people who are paid to do it, do it? Right?

Right here I'm going to invoke the Hypoctitic Oath and mention my alternative project, "Hail", which is, unfortunately, in black-out in this time. I may be getting paid to do it, too. Details are exceeding hazy at this point anyway.

By next class: Read: J. Hill, R. Szewczyk, A. Woo, S. Hollar, D. Culler, and K. Pister, "System Architecture Directions for Network Sensors". If possible the remaining articles (I'm going to start with the MapReduce one: the 2nd homework is to write a working MapReduce app — by 1/29?). So, next meeting we'll discuss... something. Architecture Directions, or our projects.

By next Tuesday: Select the project. I suspect it's a bit premature, especially since we'll be handed our first homework then, so the 2nd week is busted; project presentations only start 1/29 or later. Not that I'm worried, I narrowed the field to 2 candidates already. However, we'll need to write a project proposal too.

Tags: cs-591-005

Posted by: Pete Zaitcev at 10:45 AM | No Comments | Add Comment
Post contains 448 words, total size 3 kb.