Title: Power Saving Operations in Target Tracking and Surveillance Sensor Networks
1Power Saving Operations in Target Tracking and
Surveillance Sensor Networks
- Chao Gui, Prasant Mohapatra
- Networks Lab. UC. Davis
- Seminar, Jan 2004
2Agenda
- Target tracking and surveillance sensor networks
- Power-saving operations and the network
- Sleep planning during surveillance
- Performance Evaluation
3Multiple Target Tracking (MTT) Problem
- Problem Statement (Sittler-1964, Sastry-2004)
- A varying number of targets
- Arise at random in space and time
- Move with continuous motions
- Persist for a random length of time and disappear
- Positions of targets are sampled at random
intervals - Measurements are noisy and
- Detection probability lt 1.0 (missing
observations) - False alarms
- Goal For each target, find its track!!!
4Wireless sensor networks
- Wireless sensor node
- power supply
- sensors
- embedded processor
- wireless link
- Many, cheap sensors
- wireless ? easy to install
- intelligent ? collaboration
- low-power ? long lifetime
5Cooperative Tracking with SN
- Advantages
- Easy deployment
- Track multiple targets simultaneously
- Difficulties
- Very limited resources
- Work with local information
6Frisbee A Networks Model for Target Tracking
Applications
7Frisbee A Networks Model for Target Tracking
Applications
- Interesting events happen infrequently, and
only take place at certain locations. - Make the sensors sleep during the long interval
of inactivity. - When and where event occurs, only a limited zone
of network is kept in full active state. - For moving target, the active zone moves along.
8Agenda
- Target tracking and surveillance sensor networks
- Power-saving operations and the network
- Sleep planning during surveillance
- Performance Evaluation
9Work cycle
- Two states for a sensor node
- Surveillance state
- no events of interest in the field, but ready
to detect any possible occurrences - Tracking state
- reacts in response to any moving target,
sensors collaborate in tracking - Power-saving operations be considered in full
work cycle.
10Research Issues
- Intuitive and basic scheme
- Sleep, wakeup periodically
- Check for messages and sensing for targets
- If target detected, stop sleeping
- Basic scheme adopted
- Two issues should be considered
- Quality of surveillance
- In-time, in-place transition between two work
stages
11Problems for Sleeping
- Sleeping reduces the detectability of the network
- When a target first enters the SN field, and all
the nodes are not in ready-tracking state, how
long will it travel without exposure to any
sensor?
12Quality of Surveillance
- QoSurv in general
- Coverage, p-coverage
- Minimum exposure path (MEP)
- Aka. maximum breach path
- QoSurv on moving targets
- Special metrics needed
- Coverage --not necessary for moving object
- MEP --cannot yield direct guideline on sensor
deployment - Proposed metric
- How long will a target travel without exposure to
any sensor?
13QoSurv on Moving Targets
Assumptions (1) No sleeping (2) Non-uniform but
static sensing range (3) Uniform and random
deployment of sensors (4) Area X not fully
covered.
14Approximation of ALUL(X)
- Theorem For any straight line of length l in the
field, the expected number of intersections the
line with the disc boundaries is
- Proposition For any straight line of length l,
and let e be expected number of intersections
within disc boundaries, l/e approximates ALUL(X).
15Experiment Result
16Agenda
- Target tracking and surveillance sensor networks
- Power-saving operations and the network
- Sleep planning during surveillance
- Performance Evaluation
17Role of Sleep Planning
- QoSurv study guides sensor deployment
- Number of nodes
- Distribution
- Over-deployment of sensor nodes
- Make spare nodes sleep
- Active nodes follow deployment guideline
18Sleep Planning Methods
- Random Independent Sleeping (RIS)
- Independent decide when to wake-up
- Randomized schedule
- Neighbor Collaborative Sleeping (NCS)
- Collaborate on whose turn to be on duty
- Planned Distribution Methods
- Use location info., make the active nodes
distribute in a planned manner
19Random Independent Sleeping
- Alertness level
- Each node remain active for a percent of total
time. - Timeslots
- Within each timeslot, active for aTslot
- Randomized timeslot boundaries
20Neighbor Collaborative Sleeping
- PEAS (Ye-2003)
- Each node initially sleep for random time
- When wakeup, do Probe Environment
- If find working node, decide next sleep time,
then go sleep - If not found, become working node till energy
used up
21Examples
2
Reply
4
1
2
1
Rp
3
3
(a) 2 and 3 are working, 1 probes and starts
(b) 4 probes and 2 replies 4 goes back to
sleep
2
1
6
1
3
3
(c) 2 fails or exhausts energy
(d) 6 probes and replaces 2
22Revised PEAS
- Why revise
- Once starts working, a node does not sleep
- Goal Nodes take turn for duty, balance energy
consumption among all nodes - PECAS Probe Environment and Collaborated
Adaptive Sleeping
23PECAS by example
2
Reply (Ts)
4
1
2
1
Rp
3
3
2 and 3 are working, 1 probes and starts
4 probes and 2 replies with Ts 4 goes back to
sleep for Tse
2
1
4
1
3
3
Ts time later, 2 goes to sleep
4 probes and replaces 2
24Planned Distribution Methods 2-D Mesh
- Only nodes at planned locations are active
- Use 2-D mesh for spatial pattern
- Parameters
- r Sensing range,
- lGGrid spacing
r
lU
lG
252-D Mesh
- For i-th horizontal line, nodes with x-coordinate
in range ilG-s, ilGs are active - Similar for j-th vertical line
- Each line forms a covered stripe of width 2r2 s
- Each uncovered area is approximately a square of
side length lU - lUlG-2r-2 s
r
lU
lG
26Deterministic ALUL(X)
- Theorem Let X be the monitored area of size LL.
The physical deployment of sensors is uniformly
random distribution of adequate density. The
distribution of active sensors follows the 2-D
mesh planned pattern. Then, ALUL(X) is
27Agenda
- Target tracking and surveillance sensor networks
- Power-saving operations and the network
- Sleep planning during surveillance
- Performance Evaluation
28Performance Evaluation
29Performance Evaluation
30Performance Evaluation QoSv
31Performance Evaluation QoSv
32QoSv of mesh with errors
33Performance Evaluation
- Performance Metric
- total energy consumed
- delay of detection
- Simulation Setup
- 800 nodes,
- transmission range 55.9 m,
- sensing range 20m
- target speed 10 m/s
34Performance Evaluation Relative Energy Save
35Performance Evaluation Path Exposure
- Definition Path Exposure
- Summation of networks sensing intensity at each
point along a path. - Sensing intensity of a sensor x at a point p is
defined as S(x,p) 1/ x, p4.
36Performance Evaluation Path Exposure
37Questions, remarks?
38References
- S. Tilak, N.B. Abu-Ghazaleh, W. Heinzelman, A
Taxonomy of Wireless Micro-Sensor Network
Models, Mobile Computing and Communications
Review, Vol. 6, No. 2. - A. Cerpa, J. Elson, M. Hamilton, J. Zhao,
Habitat Monitoring Application Driver for
Wireless Communications Technology, First ACM
Sigcomm Workshop on Data Communications in Latin
America and the Caribbean, Apr. 2001 - K. Mechitov, S. Sundresh, Y. Kwon, G. Agha,
Cooperative Tracking with Binary-Detection
Sensor Networks, Technical Report
UIUCDCS-R-2003-2379, Computer Science, UIUC,
Sept. 2003