research directions in sensor networks · research directions in sensor networks kendall e. nygard...
TRANSCRIPT
Research Directions inSensor Networks
Kendall E. NygardNorth Dakota State UniversityNorth Dakota State University
IARIA InfoSys Conference Sint Maarten
March 27, 2012
Project sponsor acknowledgements: U.S. Department of Energy, ArmyResearch Office, Office of Naval Research, Air Force Office of ScientificResearch, NASA, National Science Foundation
Outline and Problem Categories
• Sensors and Sensor Networks
• Coverage and Sensor Distribution
• Hierarchies and Clustering
• Mobile Sensors• Mobile Sensors
• Sensor Failures and Self-Healing
• Sensor Scheduling
• Routing
• Security and Encryption
CrossbowMica mote
Sensors, Sensor Nodes, Mobile Sensors
CrossbowStargate
NASA – Mobile Platform
SmartPhone
TransceiverSensor 1
Sensor Node Architecture
Micro-ControllerAnalog/Digital
Converter
External Memory
Sensor 1
Sensor 2
PowerSource
Sensor Node Characteristics
• Small
• Battery power or harvested power
• Limited power → Limited lifetime
• Limited RF communication radius
• Environmental conditions can be harsh• Environmental conditions can be harsh
• Low reliability → node failures
• Static in many applications, but can be mobile
• Network topology is can be dynamic as nodes fail orothers join
• Homogeneous or heterogeneous networks
• Unattended operation
AC
BD
Geographical Coverage andDistribution
Double Coverage
Single Coverage
Coverage Holes
Triple Coverage
Research Problems in Coverage and theDistribution of Sensors
Problem Solution Approach
Numbers of sensors toachieve a minimumcoverage level
statistical analysis
coverage level
Self-location Protocols and calculationsusing signal strength,reference nodes,andtriangulation
Research Problem: Coverage HoleDiscovery and Boundary Calculations
Research Problems using Mobile Sensors
Problem Goal of AlgorithmicApproaches to DirectMovements
Cover holes in sensingcoverage.
Cover Holes
Improve networkconnectivity and topology
Reduce connectivity gapsand increase redundantconnectivity and topology and increase redundantcommunication paths
Dynamically respond to newsensing tasks
Establish or increasecoverage where needed
Compensate for nodefailures
Cover gaps caused by thefailures
Reconfigure when newnodes are added
Optimize the new topology
Routing Voids Caused by Sensor Failures
Mobile Sensors Move to Improve Coverage
Research Problem: How to Direct MobileSensors to Move and cover the holes?
Distance
Small Medium
50 m 100m
Large
Small LargeMedium
Fuzzy Logic Functions to DriveMobile Sensor Movement
RelativeHole Size
Small LargeMedium
0.33 0.67
Small LargeMedium
3 6
Number of New links
Distance Size Number of NewConnections
Decision
L - - N
M M/S S N
M M/S L Y-
M L L Y
De-Fuzzification Movement Rules
S L S Y+
S M S Y
S S S N
S L M Y+
S M M Y+
S S M Y
S L/M L Y
S S L Y-
High and Low Level Sensors
• High-level sensors (like Stargate) havebetter capabilities than Low-level sensors(like Mica) in terms of communications,computation, memory/storage, energycomputation, memory/storage, energysupply, and reliability.
• Call them H-sensors and L-sensorsrespectively
Research Problems regarding Architectureswith Low and High Level Sensors (Hierarchies)
Problem Goal of Algorithmic Approaches toUtilize the H-Sensors
Design 2-level topology tosupport energy-efficiency andhigh data rates to a basestation
Cluster the L-sensors around each H-sensor which serve as relay heads, andestablish intra and inter cluster routing
Manage H-sensors to support Establish secure routing protocols at the 2Manage H-sensors to supportprivacy, authenticity, andintegrity
Establish secure routing protocols at the 2levels
Coordinate sensing tasksover time and track mobileobjects.
Build secure and efficient timesynchronization schemes
Build intelligence into thesensor network
Build Neyman-Pearson and/or BayesainModels
Build self-healing into thesensor network
Reconfigure routing tables when sensorsfail
Hierarchies and Clustering
H-Sensor
L-Sensor
Voronoi Polygon
Base Station
Intra and Inter Cluster Routing
• Shortest path trees intra andInter cluster
• L-sensors can have smalltransmission radius
• Backbone H-sensors havelonger transmission rangeand higher bandwidth
• H-sensors can carry outdata fusion and otherintelligent operations
Tabu Search to Improve Clusters
Step 1: Create Voroni clusters as an initial solution solution i. Set i* = iand k=0.
Step 2: Set k=k+1 and generate candidate neighborhood clusters bycarrying out non-Tabu boundary node exchanges between clusters.
Step 3: Generate candidate neighborhood clusters by carrying out non-Tabu cyclic transfers among clusters.Tabu cyclic transfers among clusters.
Step 4: Choose the best cluster set i from Steps 2 and 3 in terms of arouting metric (distance or hop count)
Step 5: Compare metrics. If f(i) < f(i*) then set i* = i.
Step 6: Update the Tabu list.
Step 7: If a stopping condition is met then stop. Else go to Step 2.
L-Sensordata at time t
Data Fusion at an H-Sensor Relay Node
H-Sensor DataFusion Rules
Data Sent toBase Station
SomethingObservable
at Time t
L-Sensordata at time t
L-Sensordata at time t
Neymann-Pearson,Bayesian
Research Problems in SchedulingDifferentiated Sensors
Problem Goal of Algorithmic Approaches
Extend useful sensor lifetime Build on/off times for all sensors whilemaintaining a threshold percentage ofcoverage
Self-heal sensor schedulewhen sensors fail
Rebuild on/off times for all sensors whilemaintaining a threshold percentage ofcoveragecoverage
Scheduling Problem Approach
• Superimpose a grid
• Calculate grid cellscovered by eachsensor
AC
B
1 2
Dsensor
• Discretize timeperiods
• Configuresupply/demandnetwork
Generalized Network Flow Model ForSensor Scheduling
3
3
3
g2 ,T1
S1 ,TN
S1 ,T1
S1 ,T2
gK ,T1
g1 ,T1
s1 ,T2
s1 ,T1
s1 ,TN
S11
(0 ,6)
(2, M)
(3, M)(0 ,1)
(0 ,1)
(0 ,1)
(1, M)
(0 ,1)
(0 ,1)
(0 ,1)(0 ,∞)
(0 ,∞)
24
3
SM,T2
SM ,TN
SM, T1 sM ,T1
sM ,TN
g2,TN
gK ,TN
g1 ,TN
SM
2
2
2
1 (0 ,5)
(1, M)
(0 ,1)
(0 ,1)
(1, M)
(0 ,1)
(2, M)
(4, M)
(0 ,1)
(0 ,1)
(0 ,1)
(0 ,1)sM ,T2
(0 ,∞)
(0 ,∞)
(0 ,∞)
(0 ,∞)
Generalized Network Flow Model
Security and Key Management inHeterogeneous Sensor Networks
Problem Goal of Algorithmic Approaches
Reduce communicationoverhead, computationoverhead, and storagerequirements for securityprotocols, while supportingscalability.
Build a management framework usingRabin’s cryptosystem for encryption keysthe supports confidentiality, authenticity,and availability and allows H-sensors toestablish key space information to L-sensorsscalability. sensors
Miscellaneous Projects
Instrumentation/Sensors in theSmart Grid
2828/06/2012
Cooperating Unmanned Air Vehicles
TETWALKERSCooperative Control• Mission goals negotiated• Waypoints determined• Rough path planning• Rough obstacle avoidance
Autonomous Control• Path planning• Obstacle avoidance• Step and gait choices• Step and gait choices• Collision avoidance
Wrap-up: Interesting ResearchProblems Abound in….
• Coverage and Sensor Distribution
• Hierarchies and Clustering
• Mobile Sensors
• Sensor Failures and Self-Healing• Sensor Failures and Self-Healing
• Sensor Scheduling
• Routing
• Security and Encryption