Archive for January, 2006
I’ve just signed up with CiteULike, which is a cool little system for managing libraries of academic papers. Visit my library to see what I’ve been reading. The only problem that I see with CiteULike is that it’s hard for me to look at someone else’s library, and then link directly to the Waikato proxy’d version of it. Also, when I find a good paper through the Waikato proxy, and then I citeulike it, I’m left with a non-working link back to the original ACM or whatever website, so I still have to download and then upload the paper. It kind of defeats the purpose of the system in the first place, but oh well, it’s a nice way to organise and BibTex stuff, so I’ll keep using it. From now on I’ll link to my citeulike entries when summarising a paper.
January 27th, 2006
Jason Hill, Robert Szewczyk, Alec Woo, Seth Hollar, David Culler, Kristofer Pister. ASPLOS 2000
http://www.wand.net.nz/~smr26/sensor/papers/tos.pdf
This paper describes the TinyOS system, a multithreaded operating system for network sensors. TinyOS runs in 178 bytes of memory and supports two levels of scheduling. The paper discusses a system architecture for network sensors, from the hardware level through to the application level. A good “real-world” application of sensor principles.
January 20th, 2006
http://www.wand.net.nz/~smr26/sensor/papers/mobicom-00.pdf
David Steere, Antonio Baptista, Dylan McNamee, Calton Pu, Jonathan Walpole, Proceedings of MOBICOMM 2000, pages 292-299
This paper describes an Environmental Observation and Forcasting System (EOFS), CORIE. CORIE is a network of sensors in the Columbia river that monitors salinity, flow, etc. However, it is not a sensor network as such, and this paper is a “call for help” to the sensor network research community. Each node in the network is manually configured (!), runs a full 586 in a box (!), and communicates via a 900 MHz radio link using TCP (!). Routing is completely static in this multi-hop network and some of the nodes that act as repeaters are battery powered (!). This paper describes a situation that would be suitable for a low-power wireless sensor network, and describes the challenges involved in operating the EOFS as a traditional network. One good quote:
Reconfiguring the network in response to a station failure is as expensive as replacing the failing unit
Overall, this paper is more an example of why we should be developing sensor network technology, rather than how.
January 19th, 2006
http://www.wand.net.nz/~smr26/sensor/papers/Estrin99e.pdf
Deborah Estrin, Ramesh Govindan, John Heidemann, Satis Kumar, Proceedings of MOBICOMM 1999, pages 263-270.
This is an introductory paper on some of the key concepts of sensor networking. It discusses the need for sensor networks to be treated differently from existing networks, in terms of addressing, binding, routing, etc. It proposes that nodes in a sensor network need not be directly addressable, however they should have some way of knowing where they are, at least. Sensor networks are data-centric, and application-specific, and as such coordination of the network should be distributed and make use of localised algorithms for clustering nodes. It discusses the topic of directed-diffusion as a way for the network to diffuse queries and return results.
January 19th, 2006
So, with my undergrad degree over, it’s time to avoid the real-world some more and prepare for graduate study. Having done rapid deployment of wireless networks for my honours project, I’ve decided that it would be good to continue along the same theme. At this stage, I’m looking at low-power wireless sensor networks. I have to come up with something a bit more specific than that, obviously, but for now, I’m collecting links to relevant papers so that I can wade my way through the existing literature to get a feel for where things are at.
Here are some pages that contain lots of links to existing literature on low-power wireless sensor networks:
http://wwwcsif.cs.ucdavis.edu/~bharathi/sensor/snw.html
http://www.research.rutgers.edu/~mini/sensornetworks.html
http://www.cs.duke.edu/~alvy/courses/sensors/Papers.html
January 19th, 2006
In the process of developing the Rapid Deploy project, I found I needed to store location data (lat,long) and calculate distances and bearings between two points. Because Rapid Deploy was a 420 project, it became a horrible hack, but since then I’ve had time to separate it out into a module of it’s own. python-nav does two main things:
- Gives you a GeoPoint object to store location information in decimal or DMS format
- Calculates Great Circle Distances and uses spherical trig to calculate bearings between two points on a sphere (GeoPoints).
python-nav is useful if you’re doing things with GPS data. It’s pure-Python, so installation is a simple matter of “python setup.py install”. There’s documentation in the tarball as well as in the code itself.
Download python-nav.
January 19th, 2006