Skip Header NavigationIntranet 
CENTER FOR EMBEDDED NETWORKED SENSINGContactDirectionsEmploymentEventsNews
HomeAbout UsResearchEducationResourcesPeople

Research Project


EmStar

Technology > Systems > EmStar

On this page: Overview | Approach | Systems/Experiments | Accomplishments | Future Directions | People

Lead Investigators:

Deborah Estrin

Overview

We ported the state-of-the-art Centroute routing protocol to the EmStar. Starting with this porting, we are working on using sockets for the inter process communication (IPC) in the EmStar. The use of sockets will remove the kernel dependency of the EmStar and simplify its porting to other platforms.

We deployed the acoustic sensing array at the Rocky Mountain Biological Laboratory (RMBL). We used the lessons learned from the deployment to design the next generation of the ENSbox and a laptop-based acoustic sensing system. We envisage that the acoustic laptop will make distributed acoustic sensing more accessible.

Approach

Acoustic ENSbox: In this effort, we followed up our prior work on a self-configuring acoustic localization system. We developed a second version of the acoustic ENSBox that is easier to deploy and more suitable for field experiments. We improved the software and focused on bio-acoustic field experiments. The acoustic laptop performs in the same way as the acoustic ENSbox platform, with the added advantages of more powerful processor, hardware floating-point support, more storage space for audio data recordings, and more RAM.

Centroute in EmStar: We developed a new centralized routing protocol, Centroute, to improve upon many aspects of the distributed old routing protocols, such as the time required to form a routing tree and the robustness of the established tree. We ported the Centroute protocol to EmStar.

Systems/Experiments

Acoustic ENSbox: We deployed a 6-node ENSBox system at the RMBL.  The deployment lasted for 5 days, during which time we collected several gigabytes of acoustic data and worked out some last-minute bugs in the field. Working with several colleagues from the UCLA Biology dept, we set up a network to detect and localize marmot alarm calls.  In this process, we tested a real-time marmot call detection algorithm based on a Constant False Alarm Rate (CFAR) algorithm, and measured the offline performance of Approximate Maximum Likelihood (AML)-based localization algorithm.

These experiments were successful and data analysis showed that our localization results are within about 1 meter of the estimated true location of the marmot, based on visual observation by researchers in the field. We hope to establish more accurate ground truth results in the future experiments.

As part of the RMBL experiments, we also tested the accuracy of the AML-based algorithm for estimating direction of arrival (DOA), and its dependence on both the distance from the source and the direction of the speaker. Tests with different directions are important because the animals often are pointed away from the sensors when they call. We found that neither case introduced large amounts of error.  However, we did observe that reverberation often causes problems, and that sensors located farther from the source tended to be less subject to reverberation problems.

Regarding the acoustic laptop, we upgraded the drivers for both the sound and network cards to support the newer 2.6.15 kernel. Similar to the existing ENSbox, the laptop uses VXpocket440 PCMCIA soundcard, SMC 802.11 PCMCIA wireless network card, speakers, and microphones.

Porting of EmStar to the Nokia 770: We ported EmStar to the Nokia 770 – a handheld, WiFi-enabled, touch screen tablet device, which is used for the urban sensing research at CENS. The Nokia 770 runs a customized version of the Linux. We enabled the Nokia 770 to interface with a Mica2 mote through the 770's onboard USB connection and a USB MIB520 programming board, allowing for interaction between a Mica2 network and software running on the Nokia 770. To test the porting, we implemented MP2 (a simple application that is used to examine mote connectivity) using EmStar, Python, and TinyOS.

Accomplishments

The Centroute protocol is ported to EmStar. We are working on using sockets-based version of the Centroute in the EmStar. Based on the lessons learned from the acoustic experiments, we made a number of improvements to the ENSBox’s design. These improvements resulted in a much smaller, more integrated design, which we expect will greatly simplify the deployment process.

Future Directions

Use of sockets for the IPC: We are making the Centroute protocol functional without the dependence on the device file structure required by the EmStar.  This will allow the Centroute (and in the future other EmStar-based applications) to be more easily accessible to the users, who do not wish to manage kernel dependencies.  The first such user for this work is the Tenet project.

Acoustic and seismic array tutorial: We are preparing a tutorial on the CENS acoustic and seismic array technology to be given at the 6th International Conference on Information Processing in Sensor Networks (IPSN 2007). Boston, MA on April 24, 2007. The tutorial will introduce the software and hardware platform, with particular focus on those features that contribute to the robustness and scalability of the seismic sensing system. For the system, we will describe the key components of the system in detail, set up the array, demonstrate the use of the system, run the self-configuration software, delve into the details of the sampling and system software, and analyze data coming from the system; we will also write some simple programs to modify the nodes behavior, giving hands-on training for wireless seismic sensing systems.

Acoustic ENSBox: We will continue the improvement of the ENSBox’s design, which will include developing new boards that can be made readily available. We will deploy the system in Mexico, Colorado, and Northern California to learn from the process, as well as to make important achievements in answering questions about biology. We are developing a laptop version of the ENSbox that is more accessible and requires less investment to acquire.

People

Faculty:

Staff:

Graduate Students:


External Research Partnerships