Collaborative Research: Remote Imaging of Community Ecology via Animal-borne Wireless Networks

pdf

Abstract

The goal of this multidisciplinary project is to conceive and construct a wireless network of embedded devices, for monitoring the behavior of animals in the wild, that will be deployed and tested in a biologically-­‐relevant application (in Newfoundland, Canada). This project will be the first to deploy animal-­‐borne wirelessly networked devices that are capable of providing not only geo-­‐location data, but also executing cooperative strategies that save battery-­‐life by selectively recording bandwidth-­‐intensive audio and high-­‐definition video footage of occurrences of animal group behavior of interest, such as predation. In addition to enabling autonomous video capture, the proposed wireless network will register the relative positions of the animals and other sensory information that will be useful in sociobiological characterizations.

The following is a brief description of our ongoing research on the primary challenges of this project:

  • We are investigating new architectures for animal-­‐borne devices that can process GPS data, inertial measurements and wireless information exchange in real time. An initial prototype has been constructed that supports peer-­‐to-­‐peer communication and flexible power management of all sub-­‐modules. A network architecture has been proposed that will enable not only dissemination of time-­‐critical relevant data, but it will also support distributed cooperative decision strategies.

  • Consider a network of interacting agents (LTI) with an observer at each. We have discovered a new class of distributed observers that can achieve omniscience of the state of the agents, under certain conditions on the network. Our approach is valid for a large class of networks, and it is the first to not require the use of multiple time scales. This significantly reduces delay and simplifies performance analysis. We are also investigating new team-­‐decision theoretic methods to optimally disseminate information in a network, subject to power and topology constraints.

  • We have proposed new metrics and methods to determine the effect of measurement noise in the fidelity of models for collective motion identified from data-­‐streams. Quantifying information centrality in an experiment in a dance studio was used to illustrate our new ideas. A framework for modeling motion dynamics is also being developed to characterize predator-­‐prey and mother-­‐calf interactions. Connections with differential game theory are also being explored.

 

A site visit to Newfoundland and preliminary lab tests of the algorithms and hardware are scheduled for the coming months.

Education, Collaborations and Outreach

In addition to graduate student research, this award is supporting various related undergraduate projects, such as the development of a new simulation environment by a team from the University of Maryland at College Park. We have also established a research collaboration with Ryan Kastner (UCSD) for the development of hardware modules through undergraduate projects.

Award ID: 11357268

 

  • Architectures
  • Architectures
  • CPS Domains
  • Quantitative Verification
  • Communication
  • Embedded Software
  • Modeling
  • Wireless Sensing and Actuation
  • Robotics
  • Validation and Verification
  • CPS Technologies
  • Education
  • Foundations
  • National CPS PI Meeting 2012
  • 2012
  • Poster
  • Academia
  • CPS PI MTG 12 Posters & Abstracts
Submitted by Nuno Martins on Thu, 10/04/2012 - 09:25