zaitcev.mee.nu

Spent the meeting discussing the power grid issues and its relevance to CS, albeit in very general terms. The Astrolabe paper was the reading, and as usual the TCP haters really made my blood boil. Their protocol is squarely aimed at ignoring the congestion, and thus poisoning the well for everyone else (by taking a constant bandwidth it effectively reduces the bandwidth available for congestion control). Prof.B.'s critique of the it centered on the tendency of Astrolabe people to attempt to apply their tool to all problems (e.g. if you have a hammer, everything is a nail), which is clearly valid too.

N.B. Prof.B. says they were working "pseudo-multicast" RTP, which kinda looks like a "galactic expansion" RTP. Alternatively, an RTP on top of a spanning tree. I didn't follow the RT proto area, V.curious.

Obviously any such thing is going to be complex. Van Renesse, IIRC, comes from the Tannenbaum school of simple solutions. Aside from Amoeba, which was infamous for encoding the Intel i82586 inter-packet gap bug into its high-level protocols, he also worked on reliable broadcast based on sequential numbers. That might have burned him off any super-RTP and begat Astrolabe.

We also talked about energy spot markets, cooperative behaviours, etc. As usual the privacy/secret protection was a big question.

N.B.2 Until Prof.B. explained it, I forgot that tracker was involved into BitTorrent's economic model. For some reason I thought that peers simply accounted what they received. Not so. They also report to the tracker how much they receive from others, and pull this information to include into decisions to throttle. So, a global data space, but with no central calculation.

By Thursday: Mandatory: Kosch, Timo ; Adler, Christian ; Eichler, Stephan ; Schroth, Christoph ; Strassberger, Markus : The Scalability Problem of Vehicular Ad Hoc Networks and How to Solve it. Heavily suggested: Prashanth Mohan, Venkata N. Padmanabhan, and Ramachandran Ramjee, TrafficSense: Rich Monitoring of Road and Traffic Conditions using Mobile Smartphones.

Soon: One paragraph about the selected project. It's ahead of the project proposal paper.

Tags: cs-591-005

Posted by: Pete Zaitcev at 12:37 PM | No Comments | Add Comment
Post contains 345 words, total size 2 kb.