on demand time sychronizaton for wireless sensor networks-november2009
DESCRIPTION
Masters Project on New technique for time synchronization in an on-demand fashion to optimize the energy requirement.TRANSCRIPT
On-demand Time Synchronization for Wireless Sensor Networks
[Plan B project]
Advised by: Prof. Tian He
Presented by:
Abhishek Rawat
Introduction
Time synchronization middleware service provides time-reference for nodes
Time reference Global Peer-node
Introduction: applications
Example Applications Sniper detection Seismic activity detection Structural monitoring Object tracking Habitat Monitoring
Introduction: services
Provide time-reference for some sensor node functions TDMA scheduling LPL communication Distributed processing Aggregation techniques
Time Synchronization: approaches Proactive techniques
periodically synchronize the nodes periodicity based on precision requirement
Reactive techniques
Actuated by the event
Periodicity based on event frequency
Existing Techniques
Several techniques application in post and pre-event scenarios Varying accuracy and communication cost
Notable techniques RB TPSN FTSP ETA
Existing Techniques [continued] RBS
Receiver- receiver Broadcast based
TPSN Sender-sender MAC time-stamping
Existing Techniques [continued] FTSP
Broadcast-based MAC time-stamping and skew calculation High accuracy Reactive technique Hardware calibration
Existing Techniques
ETA State-of-art Elapsed Time-of-Arrival primitive Reduced communication cost Elapsed time in data-item-no separate time-
synchronization messaging Skew calculation is a problem
Problem Statement
How to minimize the cost of time-synchronization? Optimizing communication requirement with
desired accuracy
Motivation
Reducing communication cost post-facto technique caching: time-reference present in network
Assumptions
Spatiotemporal events Adjacent nodes will detect the event Time reference would be available in mearby
nodes Unicast messaging will reduce communication
cost
Sources of Error
Time-reference communication Sources of delay in timestamp delivery
Send/Receive Time Access Time Transmission and Reception Time Propagation Time Interrupt Handling Encoding/Decoding Time Byte Alignment Hardware Calibration/Clock-Skew
Proposed Approach
On-demand Time Synchronization Protocol
Post-facto approach
Primitive: seeker provider determination post-event
Time-reference exchange using 3-way handshake
Clock offset and Clock skew rate calculation
Approach Details
Seeker Determination
Data > threshold value
Previous time reference : Expired !!!
Approach Details
Provider determination
Approach Details: time reference exchange Time-Synchronization Messaging
3-way handshake
Tx2Tx1initMsg
t1
provider-node
t0
t2 t5
t4t3
seeker-node
Fig2. Messages exchanged during time-synchronization between two nodes
Approach Details [continued]
Skew Calculation
Approach Details [continued]
Offset Calculation
Implementation: platform
Platform
Simulator
Implementation: system design Modules:
TimeLibService TimeSyncModule TimeSyncCtrl TimeSyncCommModule Routing Module SenseDB
Implementation: system design [continued] Diagram
Analysis: applications
Accuracy requirement
Habitat monitoring : order of seconds Seismic activity detection : order of 10 milliseconds Sniper detection : order of 1 millisecond Structural monitoring : order of 10 milliseconds
Results
Data Analysis for Real Time Experiment Seismic activity detection
Data Analysis from TOSSIM based simulations Error- analysis Communication cost with event frequency Communication cost with event density Delivery success rate
Results: Data Analysis for Real Time Experiment
Analysis of Data from Seismic Activity Detection 21 days – 230 events 60 seconds per activity Sampling frequency of 100 Hz.
Results: real-time data analysis Communication complexity for various skew errors-rates over the
experiment-span
0
100
200
300
400
500
600
1 2 3 4 5 6 7 8 9 10
Skew Error in order of 10 ppm
Mes
sag
es E
xch
ang
ed in
10
00's
FTSP OTSP OTSP w ith caching
Results: tossim simulation Average error per hop
Error per hop (for first hop): 4.52 milliseconds Error per subsequent hop : 1.24 milliseconds
Error in Time Synchronization
0
2
4
6
8
10
1 2 3 4
Hop distance from base station
Err
or
in m
illiseco
nd
s
Error in Time Synchronization
Results: performance with frequency
Results: performance with event density
Results: miscellaneous observations Congestion Delivery probability
Drawbacks
Discussion
Further Work