I've posted the video for my keynote at EWSN 2010 below. You can check out the full resolution version at blip.tv.
Subscribe to:
Post Comments (Atom)
-
The word is out that I have decided to resign my tenured faculty job at Harvard to remain at Google. Obviously this will be a big change in ...
-
My team at Google is wrapping up an effort to rewrite a large production system (almost) entirely in Go . I say "almost" because ... -
I'm often asked what my job is like at Google since I left academia. I guess going from tenured professor to software engineer sounds l...
Matt, fascinating talk, thanks for putting it up here.
ReplyDeleteI was quite amused by the comment about Industry at the end. The funny thing is that some of bulk in sensor networks research is really not important in Industry either. We recently used CC2530 based motes to put together a home automation system , only to realize that we could drive down the cost signficantly by actually not sticking to Zigbee
or even 802.15.4. One of the fascinating solutions that I believe could be cheap for that application is actually based a generalization of the google meter like concepts.
Clearly, I am not a big fan of sensor networks being treated as just constrained wireless ad-hoc networks, there is more to it. Ofcourse would be good to see applications, and hence AI getting increasingly integrated with sensor networks research. Off to read your papers on Robobees
hi
ReplyDeleteit was very interesting i live in Iran and i want to use this video in my conferans about wsn if you allow to me
thanks for such interesting thing
ghader- You are certainly welcome to use this video as long as you adhere to the Creative Commons license terms that I have released it under:
ReplyDeletehttp://creativecommons.org/licenses/by-nc-nd/3.0/
Matt
Hello Mark,
ReplyDeleteThank you for this very interesting video. Could you please send me the video where the behavior of the bee hive is shown? My e-mail is franciscoheredia@hotmail.com
Many thanks
Francisco
Matt - a very thoughtful and humorous talk. I enjoyed it. Forgive me for being a bit contentious to make a point: you do not appear to have any idea what is going on in commercial wireless sensor networks. A friend of mine in the WSN industry pointed me to your talk with the statement "It's amazing what this guy doesn't know."
ReplyDeleteBased on my experience with commercial deployments, I would say that your list of solved and unsolved problems is out of touch with the state of the art in industry.
I'm astonished that in a talk on the next 10 years of WSN you made no mention of commercial deployments and standards: Zigbee, WirelessHART, IPSO, etc. . There were about $100 million in sales of wireless mesh sensors sold in industrial automation alone last year. The performance and scale of those WirelessHART networks would clearly suprise you. I'm no big fan of Zigbee, but the scale of their deployments is about three to four orders of magnitude above the networks you refer to. I don't think that ABB, BP, Chevron, Emerson, GE, Siemens etc. would agree that security, energy management, fault tolerance, and network management are unsolved problems. So *of course* it appears to you that industry is losing interest in WSN - academics have no idea what industry is really doing, and so academics work on problems that have no relevance to industry, so industry ultimately gives up and ignores academia.
The reality is that this industry is *finally* taking off in a big way, with almost zero credit due to academic research.
What really puzzles me is why this has happened in WSN. I've never seen such a disconnect between industry and academia. In MEMS, ICs, and IC CAD, academics still often work on far-out projects, but always with an awareness of where industry is. For some reason, the "inner circle" of academics in WSN does not do this, much to the detriment of the field.
ksjp
Kris, you hit the nail on the head. I *don't* really know what's happening in the WSN industry, not to that level of detail. I'm not sure that most academics do, and unless they are working closely with companies in this space I'm not sure how they are supposed to find out.
ReplyDeleteI don't know how to solve this problem. It might be that the level of competition in the WSN space is so high that companies are not that forthcoming about what they have accomplished. Certainly we see essentially zero paper submissions from industry to the major conferences. Maybe I need to start reading more trade mags, though they make me want to gouge my eyes out when it's impossible to tell what's real and what's marketing hype.
This problem is not specific to WSNs, of course; only in the last few years has there been more visibility of the "real world" problems from places like Google, Amazon, etc. - brought on largely because they started attending, and publishing in, the major academic conferences. Even so I'd argue that a lot of academics are working on "industrially irrelevant" problems largely because they are unaware of what's out there.
Can you help us fix this???
Hi Prof. Welsh, great video. The concept of robobee is very interesting. It reminds me of particle swarm optimization algorithm I was once working on. A TinyOS program running distributed particle swarm optimization protocol will be cool.
ReplyDeleteMatt, thanks for an interesting and inspiring talk, and citing my OASIS volcano project. Also thanks for your good words about us. Regarding your comments whether iMOte2 and TinyOS is a suitable choice, I have some different view: (1) iMote2 is sufficient capable for this application, and also has low energy consumption; it is a good tradeoff, although it has a funny small size comparing to the spider box. We did not see that there is no need to use a more capable board, except that its CC2420 radio has too low power output. (2) We adopted TinyOS platform, because we need to compare our design to other existing approaches in the community (for writing papers :-)) and TinyOS has largest user community and implementations of WSN. In such a real project, we have to consider design tradeoffs and time overhead ...
ReplyDelete