Bird Nest Box Monitoring at James Reserves (Tenet/Cyclops)

Project Overview

We have developed and deployed a wireless, image-based environmental monitoring system. Our system uses low-power Cyclops cameras and the Tenet general-purpose sensing system. It leverages Tenet's built-in support for reliable delivery of high data-rate sensing data, and its flexible scripting language that enables mote-side image compression. Our three-month long deployment at the James San Jacinto Mountain Reserve resulted in over 102173 images collected from a 19-node network deployed over an area of 0.05 square miles, despite highly variable individual node availability. Our biologist users found the on-line, near-real-time access to images to be useful for obtaining data on the nesting behavior of bird species.

- Abstract from ImageSense'08 publication

Logistics

Schedule

May. 9th ~ Aug. 9th (3 months), 2008

Deployment scale

19 Tenet nodes (Mica2+Cyclops nestbox package)
4 stargates and 1 laptop as Tenet Masters

Hardware

Software / Application

We have used 'Tenet' software for this deployment.
All the code that we have used are here: http://enl.usc.edu/enl/trunk/tenet/

Mica2

Mote runs 'Tenet', re-compiled with the 'CYCLOPS_HOST=1' option turned on. TOSH_DATA_LENGTH was set to '76' bytes, which can contain upto 42 bytes of image fragment (possibly compressed) data. One 200x200 B/W image requires around ~800 packets.
http://enl.usc.edu/enl/trunk/tenet/mote/apps/Tenet

Cyclops

200 x 200 Black & White image. Use default camera setting with Infrared LED lighting. Max fragment size = 42 bytes.
http://enl.usc.edu/enl/trunk/tenet/mote/test/CyclopsDevice

Tenet Imaging Application

We run Tenet imaging application with PackBits Run-Length encoding and RCRT transport protocol. Out application's task was: "Get an image and sensor values from all nodes currently alive. Repeat forever". On average, we achieve 1 image per node per ~15 minutes.
http://enl.usc.edu/enl/trunk/tenet/apps/imaging/

Compression

We use 'PackBits' Run-Length Encoding scheme with Run-Value Thresholding. Below links shows the effect of run-value thresholding;
Lossy RLE Test2, Lossy RLE Test1
We have used PackBits with threshold of 5. This was a good mix of compression ratio (~25% reduction in data size) and being able to make out the image in our application. http://enl.usc.edu/enl/trunk/tenet/mote/lib/cyclops/

Topology

Click on the map to see larger image:

(original map)
Here is the PRR and link quality (RSSI) plot:

Bird Nest Data

Here are few of the sample images: SAMPLE

Here is an amazing predation event caught on our cameras:

Here is our image browser: VIEWER

Here is a screen shot of our image browser:

Photos

More photos available [HERE (deploy)]

People

John Hicks (UCLA)
Jeongyeup Paek (USC)
Sharon Coe (UCR)
Ramesh Govindan
Deborah Estrin

Publication

John Hicks, Jeongyeup Paek, Sharon Coe, Ramesh Govindan, Deborah Estrin, An Easily Deployable Wireless Imaging System, In: Proceedings of the ImageSense08: Workshop on Applications, Systems, and Algorithms for Image Sensing, Raleigh, NC, November 2008. [PDF][Slides] [Abstract]

Related Publications

Jeongyeup Paek, Ben Greenstein, Omprakash Gnawali, Ki-Young Jang, August Joki, Marcos Vieira, John Hicks, Deborah Estrin, Ramesh Govindan, Eddie Kohler, The Tenet Architecture for Tiered Sensor Networks, ACM Transactions on Sensor Networks (TOSN), 2009. [Abstract]

Talk

John Hicks, An Easily Deployable Wireless Imaging System, Raleigh, North Carolina, November 2008. [PDF]

Poster

An Easily Deployable Wireless Imaging System, CENS Research Review Poster, September 2008. [PDF]