Sunday, November 7, 2010

SenSys 2010 in Zurich

Photo from http://www.flickr.com/photos/aforero/542248140

I just got back from Zurich for SenSys 2010. I really enjoyed the conference this year and Jan Beutel did a fantastic job as general chair. The conference banquet was high up on the Uetliberg overlooking the city, and the conference site at ETH Zurich was fantastic. We also had record attendance -- in excess of 300 -- so all around it was a big success. I didn't make it to all of the talks but I'll briefly summarize some of my favorites here.

Sandy Pentland from the MIT Media Lab gave a great keynote on "Building a Nervous System for Humanity." He gave an overview of his work over the years using various sensors and signals to understand and predict people's behavior. For example, using various sensors in an automobile it is often possible to predict in advance whether someone is about to change lanes, based on subtle prepatory movements that they make while driving. His group has also used wearable sensors to gather data on conversational patterns and social interactivity within groups, and used this data to study practices that influence a business' productivity. This was an amazing keynote and probably the best we have ever had at SenSys -- very much in line with where a lot of work in the conference is headed.

The best paper award on Design and Evaluation of a Versatile and Efficient Receiver-Initiated Link Layer for Low-Power Wireless was presented by Prabal Dutta. This paper describes a new MAC layer based on receiver initiation of transmissions: receivers send probe signals that are used to trigger transmissions by sending nodes with pending packets. Their approach is based on a new mechanism called backcast in which senders respond to a receiver probe with an ACK which is designed to constrictively interfere with multiple ACKs being transmitted by other sending nodes. This allows the receiver probe mechanism to scale with node density. Because A-MAC does not rely on receivers performing idle listening, cross-channel interference (e.g., with 802.11) does not impact energy consumption nearly as much as LPL.

There were a bunch of talks this year on use of cell phones and other sensors for participatory sensing applications. One of my favorites was the paper on AutoWitness from Santosh Kumar's group at the University of Memphis. In this work, a small tag is embedded within a high-value item (like a TV set). If the item is taken from the home, accelerometer and gyro readings are used to determine its probable location. Using HMM-based map matching they showed that they can reconstruct the path taken by a burglar with fairly high accuracy.

Chenyang Lu from WUSTL presented a paper on Reliable Clinical Monitoring using Wireless Sensor Networks: Experience in a Step-down Hospital Unit. This paper presents one of the first studies to make use of low-power wireless sensors in a real hospital environment with real patients. My group spent about seven years working on this problem and we were often frustrated at our inability to get medical personnel to sign on for a full-scale study. Chenyang's group managed to monitor 46 patients in a hospital over 41 days (but only three patients at a time). Their paper showcases a lot of the challenges involved in medical monitoring using wireless sensors and is a must-read for anyone working in the area.

Finally, Steve Dawson-Haggerty from Berkeley presented his work on sMAP, a framework for tying together diverse sensor data for building monitoring. Steve's observation is that while different companies have worked on various protocols for standardizing building monitoring applications, most of these systems are highly proprietary, vertically-integrated nightmares of multiple entangled protocols. Steve took a "Web 2.0" approach to the problem and designed a simple REST-based API permitting a wide range of sensors to be queried through a Web interface. This is a really nice piece of work and demonstrates what is possible when a clean, open, human-centric design is preffered over a design-by-committee protocol spec with twenty companies involved.

Speaking of companies, one disappointing aspect of this years' conference is that there were very few industrial participants. None of the papers were from companies, and only a couple of the demos had any industrial affiliation. Part of the reason for this is that the conference organizers didn't approach many companies for support this year, since the budget was adequate to cover the meeting expenses, but this had the negative effect of there being essentially zero industrial presence. My guess is that the companies are going to the IEEE sensor nets conferences, but I am deeply concerned about what this means for the SenSys community. If companies aren't paying attention to this work, we run the risk of the wheels of innovation grinding to a halt.

There was one talk this year that was highly controversial -- Tian He's group from University of Minnesota presented a paper on an "energy distribution network" for sensor nets. The idea is to allow sensor nodes to push energy around, in this case, using wires connecting the nodes together. Unfortunately, the presenter did not justify this design choice at all and the only experiments involved very short (1 meter) cables between nodes. It seems to me that if you connect nodes together using wires, you can centralize the power supply and bypass the need for a low-power node design in the first place. The fact that the presenter didn't have any good arguments for this design suggests that the research group has not spent enough time talking to other people about their work, so they've built up a herd mentality that this actually makes sense. I don't think it does but would love to hear some good arguments to the contrary.

Apart from SenSys, I had the chance to (briefly) visit Timothy Roscoe at ETH Zurich as well as connect with some colleagues at the Google Zurich office. ETH Zurich is a very exciting place: lots happening, lots of faculty growth, tons of resources, good students and postdocs. I was very impressed. Even more impressive is Google's office in Zurich, which has the most over-the-top design of any of the Google offices I've visited so far (including Mountain View). The office is beautifully laid out and has a bunch of humorous design touches, including an indoor jungle and firepoles that connect the floors (with a helpful sign that reads, "don't carry your laptop while sliding down the pole.")

17 comments:

  1. "None of the papers were from companies, and only a couple of the demos had any industrial affiliation."

    Hi Matt, thanks for the roundup. One handicap of industry papers / demos might be the often mundane nature of good engineering solutions.

    Cheers,
    tamberg

    ReplyDelete
  2. Thanks for the interesting post.

    However, about the Minnesota group paper: instead of blaming the authors for their ideas, which can be sometimes good, sometimes bad, as anyone else's ideas, I think the responsibility of this very poor presentation and paper should entirely fall on the PC.
    One should ask first: why the top-experts in the world in the area accepted the paper in first instance???
    I think that at the beginning of the keynote talk somebody said that not only SenSys has a very low acceptance rate, but that most of the works submitted to SenSys are high quality ones which normally would be accepted in other top conferences of the area...

    ReplyDelete
  3. tamberg - I don't agree at all that industrial papers are any less innovative or relevant to the conference than those from academia. Look at conferences like NSDI, OSDI, SOSP, where a lot of papers from places like Microsoft and Google appear. The problem here is that the industry players in the sensor nets space are not going to SenSys, and probably don't consider it to be an important venue for them. My concern is that if the conference remains a purely academic venue it will lose relevance over time.

    Anon - I don't agree that the PC should take responsibility for a bad presentation. I haven't read the paper yet but presumably there are some good ideas in there. The speaker did not justify the most important aspect of the design, but that is a presentation problem, not a problem with the work itself.

    ReplyDelete
  4. Hi Matt- thanks for the writeup. I wanted to put in a shameless plug for sMAP -- the code's up here (http://code.google.com/p/smap-data/), and you can poke around with most of the thousands of streams we mention in the paper starting here (http://www.openbms.org/smap/plot). I'm also going to release the A-MAC code this week for people who are interested; that announce will go to tinyos-devel.

    That aside, let me agree wholeheartedly with you on the industrial participation angle. I hope the PC next year can encourage more of it, since a mix is really valuable. Re: the whole "less innovative" angle, it's funny because I was just reading the Facebook Haystack OSDI paper on the plane back. The authors readily concede that there's nothing really new there, except for the fact they store something like 240e9 photos in an on-line system. So, one role industrial papers provide is showing the academic community the scale and breath of problems being solved; also they tend to point to which application areas are really ready for commercialization and where the real problems still are.

    ReplyDelete
  5. Having worked in the industry for many years, I would say it is pointless -- other than to laugh at incorrect assumptions or old idea (relative to those in my company). We were usually five years ahead of academia. The only time we published is either for prior-art purposes or indirect advertisement for an upcoming product.

    ReplyDelete
  6. It seems that Matt is pretty wrong about the paper and the PC appreciates the paper more since judging by info at this page:

    http://www-users.cs.umn.edu/~yugu/

    # Nov. 2010
    Our paper on energy sharing network is among the highest ranked papers at SenSys'10, and fast tracked to ACM Transactions on Sensor Networks (TOSN) for future publication.

    ReplyDelete
  7. @Anonymous (Industry is far ahead of academia) -- this is a perspective I've heard before from industry, although I also think academics do make valuable contributions, such as by Phil Levis to the IETF/RPL process. I wonder if you can comment on exactly where you think industry is ahead and where it is behind. It seems pretty clear to me that industry does a better job at engineering integrated hardware-software solutions whereas academic studies often use the telosb, which is certainly long in the tooth.

    The goal of academia is always to be ahead of what would be "a good idea" or profitable, but not all papers get there by any means. In other fields industry has a healthy participation -- I would cite networking and OS both as similar fields where companies publish in Sigcomm or OSDI. Now, it is often as an advertisement, ego-boosting, or recruitment tool, but I think they do help the community get their heads around the scale and types of problems being solved in industry. I guess I don't see why the incentives in this space are different then they are there. Smaller companies? More classified stuff?

    ReplyDelete
  8. In another shameless plug (Steve, started the trend!), I had a demo on a very similar idea to Tian He's work, which articulated the idea of energy-transference using a controllable mirror and a LED lamp as a light source. (http://www.isi.edu/~asyed/papers/syed_energy_transference_demo.pdf) so we had wireless transfer that could potentially be routable (think of a collection of mirrors!)

    I agree that the perhaps that presentation did not make as strong a case as the paper did (which I think is a *very sound* systems paper, do read), but I see the argument for energy-transference in a sensor network to be of an enabling one; i.e. it enables deployments that would not have happened otherwise. As Matt said when we talked at the demo, the idea is cute, but before it si taken seriously we need to articulate, precisely, what the delta will be if we add energy transfer capabilities into our sensornet.

    Regarding the academia vs. industry thread, I think the question to ask (perhaps people from the industry) what would make them want to publish at Sensys? Perhaps people from google power-meter ( a place where there is proven record of publishing in academic conferences). But I think there will always remain an inherent tension for the industry to publish unless there is a clear monetary incentive, as frankly otherwise it doesnt justify an employee's time.

    The industry doing a great job of popularizing sensor nets, but they solve what are the current problems. The academia tries, and should never shy from, taking exploratory sojourns in seemingly iffy areas, and if they dont seem too promising it should -in theory- prune itself back. We should introspect and see if we are doing that...

    ReplyDelete
  9. To the anon who wrote:

    "# Nov. 2010
    Our paper on energy sharing network is among the highest ranked papers at SenSys'10, and fast tracked to ACM Transactions on Sensor Networks (TOSN) for future publication."

    I think that exactly on this blog Matt wrote tim ago that he did not like the idea of having the same paper published in e.g. SenSys and then ACM TOSN, since most of the times the modifications to the paper already published are too small to justify an additional publication.
    What happened then in the case of this paper?

    ReplyDelete
  10. Speaking of industry vs academia and energy transfer, you are aware that there are already products from the industry that does wireless energy transfer, right? Check out powercast: http://www.powercastco.com/ and http://www.powercastco.com/products/development-kits/

    ReplyDelete
  11. The products pointed out are severely limited...
    First of all, the resonant frequency of these power harvester is 915 MHz, and this frequency band has several other limitations for WSN applications. All the WSN applications using 2.4 GHz can not use this technology...
    Second, the power generated is extremely low (extremely! nothing to compare with the energy generated by e.g. pizezos) and strictly related to the distance between the source of the radio signal and the receiver (tens of centimeters)...
    I would say that if you wanted to point out the supremacy of industry over academy, you really used the wrong example...

    ReplyDelete
  12. Coming from an industry background I would like to add a few thoughts into the industry - academia discussion.

    - There is almost no reason to go for a conference as the papers are available right after through portals.
    - Time is rare to prepare a bulletproof paper for such conferences.
    - The IP approval process in organizations usually forces you to delay publication to a point where it is nothing new anymore - thus almost not publishable anymore.
    - Travel costs hit the always tight budget.
    - Academia is usually not considering patent background searches and claims to show the first attempt of doing XYZ. If you look at the patent space however, you will quickly discover that more than half of the so called "innovations" have been patented years ago by companies. Thus industry of course perceives this as an "old hat". (Patent literature usually misses the detailed analysis though.)

    ReplyDelete
  13. To the Anon about industry power-transfer, yes we are aware of these and others (witricity ?).

    However, while that is a technical mechanism to do energy transfer, our research is trying to motivate the paradigm shift that energy be decoupled from the sensing location and allow optimal harvesting at locations where its most available and transfer it to where it is most needed. There are many design choices that can fulfill that requirement, powercast devices may be one (but as another poster pointed out, not the best matched), but there are several others. We plan to explore them while the industry keeps finding better (in efficiency, distance, interference) modes of transfer.
    (btw: that above is my current view of the roles of industry and academia, in a healthy field atleast)

    ReplyDelete
  14. Adding to the academia-industry argument, it is usually difficult to have meaningful conversation at conferences. You can never give a good answer to the question: what are you working on at the moment? My answer: 'important stuff'.

    ReplyDelete
  15. Industry versus academia

    I spend quite a lot of time thinking about this, probably too much :)

    1. Example where industry > academia: There's this CoNext conference a couple of years back, where the presenter talked about a platform for providers to support other providers. Amongst the feedback given was that Cisco already has that sort of platform, and more, it was already in operation in various provider networks. I'm not really sure who the reviewers are, or how that paper got in.

    2. I work with large teams in the networking industry, as well as with 'researchers' who often publish papers. I have to say that 'researchers' tend to focus on their individual goals and achievements, while the (majority) of the team members focus on getting the project going. What this results in is the fracturing of the project: rather than have a single simple approach or methodology or solution, the 'researchers' want certain things to be done their way, breaking the cleanliness of the overall architecture. Just so they can put their names on it and say, "Look, here's my contribution!". Of course, this is a sweeping statement but I wonder if the PhD course tends to produce such individualists.

    Breaking the simplicity of the architecture has many repercussions later, the most prominent of which is the difficulty in maintaining and extending the system. So, while in the short-term the 'researchers' may be happy to have their conference papers, they won't stick around when their colleagues suffer down the road. After all, they've already got what they want: papers. So, this craving for conference papers, for instant gratification, is not really a healthy thing for the industry.

    3. Suppose you're someone in the industry, and you bring money home at the end of every month and put food on the table for your family. Suppose one day you get this paper to review. And this paper basically says there's a better way to do what you or your company is doing now. Ignoring the egoistic side of you, say if you accept the paper, you will help the idea to spread thereby causing your company or you to become antiquated. Will you accept or reject the paper? Remember that your livelihood, spouse, kids and mortgage depend on your decision.

    Because of this, I realized why many of the networking papers don't seem to reflect reality. The more stake people have in the network, the more they have to lose. This means that the truly world-changing ideas are less likely to make it through the review process, more likely to get knocked down. This also means that it's the papers, while interesting but won't really have any significant impact and thus won't cost someone's livelihood, that will be accepted. The other kind of paper that gets accepted are those that have impact, but reside in their own small space and thus don't really harm anyone else.


    This is why I think the reviewers should be the class of people who have small egos and don't really have anything to lose. But alas, these kind of people are few in number, and given the many conferences, they'll be totally swamped :)

    ReplyDelete
  16. Anon (Industry vs academia): nice post. Like your #1 and #2, but totally confused by #3. A good idea will always be accepted somewhere, so there is little benefit in stalling the paper. If a better idea is put forward, I would immediately assemble a bunch of engineers together and implement it and make sure the next revision of the product has the better idea. As we all know, it's about first in the market.

    In academia, we play the childish game: look ma, I'm better than the other kids.

    Talking about CoNext, there was a best paper that claims to be the first to propose an idea by not citing a whole area of research.

    ReplyDelete
  17. Anon re: industry vs. academia. Very interesting points. I have never considered the "individualist" perspective before, but I think you're onto something.There is a tremendous gap between trying to do something great because it's useful and the right thing to do, and trying to be "innovative". Academics are trained to be innovative for the sake of being innovative, often to the detriment of doing anything useful. I've lost count of how many times I've seen someone INVENT a problem to solve just so they can be the first to do it and publish it, even when the problem itself is totally artificial. This mode of thinking does not translate well into group settings where the focus is on building products, although it can have enormous value when you're trying to do something disruptive - as in a startup. So I think it depends on what kind of "industry" you're talking about. One shouldn't put Cisco and Twitter in the same category.

    I don't agree on your point regarding the self-serving suppression of conference papers. That might happen in some communities but not in the top systems and networking conferences that I'm familiar with. Then again, these conferences tend to be dominated by academics who have less stake in the game, but I've never seen an industry person try to shoot down an obviously interesting paper because the idea is disruptive to their business model. That would not be acceptable behavior on a program committee in any event.

    ReplyDelete

Startup Life: Three Months In

I've posted a story to Medium on what it's been like to work at a startup, after years at Google. Check it out here.