Tuesday, February 20, 2007

SenSys'05 conference.

Some notes from the SenSys'05 conference. I attended only tangentially (I assisted at the first keynote, but had to leave right afterwards).

  • Radio Interferometric Geolocation. Miklos Maroti, Branislav Kusy, Gyorgy Balogh, Peter Volgyesi, Andras Nadas, Karoly Molnar, Sebestyen Dora, Akos Ledeczi (Vanderbilt University)
    This paper might be very interesting for multiple applications outside of the sensor network world, despite sensors being its primary focus. The idea is to send two signals with slightly different frequencies from two sources, and two observe the phase of the cumulative signal at the receiver. Based on the phase, the receiver can obtain its distance from the sources. This allow to perform geolocation (localization, positioning) at little cost, since there is no need for GPS or dedicated (cricket-like) hardware.
    Application is obvious for positioning of a handset (caveat: I am not aware of the techniques currently used, it might already be something similar). This could be used for indoor localization, or for emergency services localization. The technique is simple enough so that it could be easily implemented on a handheld device.

  • High-Accuracy, Low-Cost Localization System for Wireless Sensor Network. Radu Stoleru, Tian He, John A. Stankovic, David Luebke (University of Virginia)
    This is a paper on localization, which, unlike the previous one, has no applicability to cellular networks. It uses dedicated hardware to measure the distance of the distributed sensors. Localization cannot be performed in real-time, and requires the sensor nodes to be static.

  • A New Approach for Establishing Pairwise Keys for Securing Wireless Sensor Networks. Arno Wacker, Mirko Knoll, Timo Heiber, Kurt Rothermel (Universit├Ąt Stuttgart)
    I have not read this paper, its applicability to my work seems limited.

  • TSAR: A Two Tier Sensor Storage Architecture Using Interval Skip Graphs. Peter Desnoyers, Deepak Ganesan, Prashant Shenoy (University of Massachusetts Amherst)
    I have not read this paper in details, it is about distributed storage and a hierarchical architecture to perform storage, not unlike a distributed hash table.

  • A Macroscope in the Redwoods. Gilman Tolle, Joseph Polastre, Robert Szewczyk, Neil Turner, Kevin Tu, Stephen Burgess (UCB), David Gay, Phil Buonadonna, Wei Hong (Arch Rock Corporation), Todd Dawson, David Culler (UCB)
    This paper is describes an experimental approach for observing an ecosystem using sensor network. I saw most of the material presented in this paper during a keynote talk by David Culler for the MobiHoc 2005 conference. The sensor network is used to extract data streams, such as the temperature and humidity in a redwood grove as a function of the elevation into the tree. Very interesting as an environment monitoring tool.

  • Design and Deployment of Industrial Sensor Networks: Experiences from the North Sea and a Semiconductor Plant. Robert Adler, Phil Buonadonna, Jasmeet Chhabra, Mick Flanigan, Lakshman Krishnamurthy, Nandakishore Kushalnagar, Lama Nachman, Mark Yarvis (Intel)
    This is the poster child for sensor network in industrial monitoring application. It is the example given by Intel people as to what is the application for sensor networking. I was surprised to see this presented here, as I was convinced it had been presented already. Basically, it describes the deployment of a sensor network for monitoring instrumentation on an oil tanker, and for monitoring vibrations in a silican wafer manufacturing plant. In both cases (for which the cost of wiring is extremely harsh), the wireless sensor network proves superior.

  • A Unifying Link Abstraction for Wireless Sensor Networks. Joseph Polastre, Jonathan Hui, Philip Levis (University of California, Berkeley), Jerry Zhao (ICSI Berkeley), David Culler (University of California, Berkeley), Scott Shenker (ICSI Berkeley and University of California, Berkeley), Ion Stoica (University of California, Berkeley)
    This paper argues that the "narrow waist" of the protocol stack for sensor networks should be at the link layer, and not at the network layer, as it is in IP.

  • Z-MAC: A hybrid MAC for wireless sensor networks. Injong Rhee, Ajit C. Warrier, Mahesh Aia, Jenogki Min (NCSU) Prashant Patel (Progress Energy)
    I did not look into the details of the protocols, but it looks like another TDMA-CSMA hybrid. It might be interesting, but it is definitely not exciting.

  • Packet Combining in Sensor Networks. Henri Dubois-Ferriere (EPFL), Deborah Estrin (UCLA), Martin Vetterli (EPFL)
    The very good idea of the paper is relatively simple: when transmitting a packet over multiple hop, say from A to B to C, C might overhear some of the transmission from A to B. Thus it does not require a perfect transmission from B to C, but just enough so that it can reconstruct the packet. Of course, using coding so that the packet from A to B is the original data, and the packet from B to C is some packet of same length made of parity bits, which allows a more efficient decoding at C than just transmitting the orginal packets twice between A and B and between B and C.

    [update: these notes were written in 2005. But as I put them here, I see that I just read a paper by Ed Knightly's group to appear in Infocom'07 which puts out a very similar idea.]

  • Siphon: Overload Traffic Management using Multi-Radio Virtual Sinks. Chieh-Yih Wan (Intel Research), Shane B. Eisenman (Columbia University), Andrew T. Campbell (Dartmouth College), Jon Crowcroft (Cambridge University)
    The idea is to detect congestion in the network, which is typically due to funneling effect: all the traffic fans in towards the data sink. The combat this fanning in issue, another sink, using a different radio connection towards the initial sink, can be placed in the network to redirect traffic.
    This is a relatively obvious result: you add capacity where you need it. I fear it might be a technique with limited applicability: coming up with new hardware to reduce traffic in a hot spot might make sense in a static sensor network, but not in a dynamic environment. I guess it all depends on the granularity of the time scales.

  • Estimating Clock Uncertainty for Efficient Duty-Cycling in Sensor Networks. Saurabh Ganeriwal (University of California Los Angeles), Deepak Ganesan (University of Massachusetts), Hohyun Shim, Vlassios Tsiatsis, Mani B. Srivastava (University of California Los Angeles)
    I did not go into the details of this one. Power saving technique might have application outside of the sensor network universe, but it seems cellular clock sync is working. It also seems that the dynamic aspect of the cellular environment might prove too challenging for these techniques.

  • Firefly-Inspired Sensor Network Synchronicity with Realistic Radio Effects. Geoffrey Werner-Allen, Geetika Tewari, Ankit Patel, Matt Welsh, Radhika Nagpal (Harvard University)
    This paper present a node synchronization protocol based on locking in phase with ones neighbor, a process similar to that of the dance of the fireflies. It is cute, but the real-world applicability seems far away.

  • Data Collection, Storage and Retrieval with an Underwater Optical and Acoustical Sensor Network. Iuliu Vasilescu, Keith Kotay, Daniela Rus (Massachusetts Institute of Technology), Peter Corke, Matthew Dunbabin (CSIRO Australia)
    Under-water sensor network is a challenging application, but not of much interest to me.

  • MAX: Human-Centric Search of the Physical World. Kok Kiong Yap, Vikram Srinivasan, Mehul Motani (National University of Singapore)
    This is a paper to allow localization of object using a taxonomy which is easily understandable by a human user. This might be of interest to the UI community.

  • CenWits: A Sensor-Based Loosely-Coupled Search and Rescue System using Witnesses. Jyh-How Huang, Saqib Amjad, Shivakant Mishra (University of Colorado, Boulder)
    The paper describes a search and rescue systems which functions as follows: people wear tags (802.11 based, but one assume RFID could be an alternative technology); some readers are distributed in the environment, and take a reading of the tags. When a catastrophe occurs (say, an avalanche), the tag readings are used to locate the users.
    There is a huge privacy issue here. On the other hand, the system could be integrated on a cell phone, so that the 'tag' becomes the handheld device. I guess it is practical of people of University of Colorado to come up with avalanche rescue systems. I wish I could work there, the experimentation must be fun!

  • Cyclops: In Situ Image Sensing and Interpretation in Wireless Sensor Networks. Mohammad Rahimi (Center for Embedded Networked Sensing - UCLA), Rick Baer (Agilent Technology), Obimdinachi I. Iroezi, Juan C. Garcia (Center for Embedded Networked Sensing - UCLA), Jay Warrior (Agilent Technology), Deborah Estrin, Mani Srivastava (Center for Embedded Networked Sensing - UCLA)
    This is a sensor network platform which carries a low resolution digital camera, ie a so-called image sensor. This is made in partnership with Agilent Technology.

  • Lightweight Detection and Classification for Wireless Sensor Networks in Realistic Environment. Lin Gu (University of Virginia), Dong Jia (Carnegie Mellon University), Pascal Vicaire, Ting Yan, Liqian Luo, Tian He (University of Virginia), Ajay Tirumala (University of Illinois at Urbana-Champaign), Qing Cao, John. A. Stankovic, Tarek Abdelzaher (University of Virginia), B.H. Krogh (Carnegie Mellon University)
    This paper I did not read in details.

  • Intelligent Light Control using Sensor Networks. Vipul Singhvi, Andreas Krause, Carlos Guestrin, Jim Garrett, H. Scott Matthews (Carnegie Mellon University)
    I think the title says it all: it is a sensor network which controls a lighting system. It seems very application specific.

  • Algorithms for Generic Role Assignment in Wireless Sensor Networks. Christian Frank, Kay Roemer (ETH Zurich)
    This paper (and the next two as well) I did not go into details, as it deals with software support for sensor network, a topic I am not familiar with.

  • VM*: A Scalable Runtime Environment for Sensor Networks. Joel Koshy, Raju Pandey (University of California, Davis)
    I did not go into the details of this paper, as it deals with software support for sensor network, a topic I am not familiar with.

  • Sympathy for the Sensor Network Debugger. Nithya Ramanathan, Kevin Chang, Lewis Girod, Rahul Kapur, Eddie Kohler, Deborah Estrin (UCLA)
    I did not go into the details of this paper, as it deals with software support for sensor network, a topic I am not familiar with.

No comments: