Project
Overview
Overall
Objective
The overall objective of this multi-university interdisciplinary
research project is to develop a next-generation wildlife monitoring
technology for behavior analysis, interaction modeling, disease
tracking and control. More specifically, the research team will develop
theories and technologies in efficient wireless networking and video
sensing, and design a wireless sensor network on wildlife species,
e.g., deer, to collect video information about their daily activities,
which is the essential information needed in wildlife behavior analysis
and interaction modeling.
Background
The biological relationship between wildlife and humans has never been
more intertwined. Outbreaks of various infectious wildlife diseases
threaten wildlife populations, human health, food safety and national
economy, as well as our homeland security if the wildlife species are
used by bio-terrorists to spread deadly diseases. Current technologies
available for wildlife studies, such as VHF ratio-telemetry and GPS
tracking, significantly limit our capability in studying the behavior
and interaction of the wildlife species, and the dynamics of the
free-ranging wildlife remains largely unknown. Lack of scientific
knowledge about the behavioral interactions and dynamics of wildlife
systems, our ability to prevent, manage, and control wildlife diseases
is very limited.
Why
Video is needed?
Without access to visual information about the
animals' daily activities,
- We donot know what the animals are doing, either feeding,
walking , or bedding? How they are doing?
- Even if we know the animals present some strange behavior,
such as unusual movement, from other sensor information, without
images, we donot know what has caused this strange behavior (under
attack?) and the environmental context for this behavior.
- Another major motivation for video monitoring is
interaction modeling and disease tracking. If we only know the location
of the animals, we have no idea if they are interacting or not.
Wildlife diseases propagate through animal-animal and
animal-environment interactions. From the images obtained from the
camera mounted on the animal, we can detect and track the interactions,
and therefore build an interation model for disease tracking.
Major Challenges
The fundamental challenges of the proposed research lie in the
following:
- Tremendous amount of video data for monitoring behaviors
and interactions is collected in real-time and delivered over the
unreliable wireless networks;
- Limited energy supply, mobility of the animal-mounted video
sensors, and long operational lifetime requirement to achieve an
unobtrusive observation pose serious constraints on wireless networking
design in supporting real-time video delivery;
- Wildlife behavior and interaction modeling requires
simultaneous processing and analysis of a number of lengthy video
streams collected from the video sensors. Disease tracking based on
interaction models should be carefully investigated.
Major Research Components
To address the significant technical challenges and accomplish the
far-reaching research objective, the research team brings together
extensive expertise in wireless ad hoc networking, real-time embedded
video communication, wildlife tracking and behavior monitoring, and
propose to organize its efforts among three technical thrust areas:
- Resource-aware video sensing: The research team will
explore the unique characteristics of wildlife videos, and develop
theories and technologies in energy-scalable video encoding, accurate
resource control and optimum resource allocation for resouce-efficient
real-time video compression.
- QoS-assured wireless sensor networking: Realizing the
resource limitation in processing and transmission and high
unreliability of sensor nodes, the research team will develop efficient
wireless medium access control (MAC) and robust routing mechanisms to
efficiently coordinate communication and the video traffic delivery in
wireless video sensor networks.
- Wildlife behavior monitoring, disease tracking and control:
The research team proposes to develop vision and networking-assisted
wildlife video analysis technologies, and build behavior and
interaction models to predict, track and control wildlife diseases.
|
