Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Lecture 8Wireless Sensor Networks:
Overview
Reading: • “Wireless Sensor Networks,” in Ad Hoc Wireless Networks: Architectures and
Protocols, Chapter 12, sections 12.1-12.2.• I. Akyildiz, W. Su, Y. Sankarasubramaniam and E. Cayirci, “A Survey on
Sensor Networks,” IEEE Communications Magazine, August 2002.• D. Estrin, R. Govindan, J. Heidemann, and S. Kumar, "Next Century
Challenges: Scalable Coordination in Sensor Networks," Proc. Mobicom '99, August 1999.
• “Wireless Sensor Network Protocols” by M. Perillo and W. Heinzelman. To appear in: Fundamental Algorithms and Protocols for Wireless and Mobile Networks, CRC Hall, 2005.
2
Wireless Sensor Networks (WSNs)Microsensors
Low power, cheap sensorsSensor module (e.g., acoustic, seismic, image)A digital processor for signal processing and network protocol functionsRadio for communicationBattery-operated
Sensors monitor environmentCameras, microphones, physiological sensors, etc.Gather data for some purpose
Hundreds or thousands of nodes scattered throughout an environmentEach sensor can collect dataData routed via other sensors to a sink or base station node
3
WSNs (cont.)Sensor data limited in range and accuracy
Each node can only gather data from a limited physical area of the environmentData may be noisyData aggregation enables higher quality (less noisy) data to be obtained that gives information about a larger physical area than any individual data signal
Networking sensors enablesExtended range of sensing improved quality Fault tolerance due to redundancy in data from different sensorsDistributed processing of large amounts of sensor dataScalability: quality can be traded for system lifetime“Team-work”: nodes can help each perform a larger sensing task
4
WSN ApplicationsNew wireless networking paradigm
Requires autonomous operationHighly dynamic environments
Sensor nodes added/failEvents in the environment
Distributed computation and communication protocols required
ApplicationsHome securityMachine failure diagnosisChemical/biological detectionMedical monitoringSurveillance and reconnaissanceAnimal/plant monitoring (e.g., for research)
5
Example Application: Environmental Monitoring
Traffic patterns many-to-oneRaw sensor data or high level descriptions about environmental phenomenaExample projects
ZebraNetEcology of rare plants in Hawaii
6
Example Application: Health Monitoring
Sensors monitoring vital signsBlood pressure, heart rate, EKG, blood O2
One or more sensors indicate abnormalityInform personTake corrective measures
Control pacemaker functionsAutomatically dispense medication
Alert emergency teamSensing, processing, understanding, feedback/controlRequires protocols that are
ReliableFlexibleScalableSecure
7
Sensor Platforms
Example platformsSmart Dust (UC Berkeley)Berkeley MotesTelos Motes (MoteIV)iBadge (UCLA)WINS (UCLA)
8
Sensor Platforms
9
WSN LimitationsSensor energy
Each sensor node has limited energy supplyNodes may not be rechargeableEventually nodes may be self-poweredEnergy consumption in sensing, data processing, and communication
Communication the most energy-intensiveMust use energy-conserving communication
SENSORS
Power consumption of node subsystems
0
5
10
15
20
Pow
er (m
W)
CPU TX RX IDLE SLEEP
RADIO
10
WSN Limitations (cont.)Communication
The bandwidth is limited and must be shared among all the nodes in the sensor networkSpatial reuse essentialEfficient local use of bandwidth needed
11
WSNs vs. MANETs
Many-to-one traffic patternOne-to-one traffic pattern
Data-centricAddress-centric
Very constrained energy and bandwidthConstrained energy and bandwidth
CooperativeCompetitive
Application-specific QoSQoS: delay, etc
Typically immobileTypically mobile
Large-scaleSmall-scale
Require self-configurationRequire self-configuration
Unreliable communicationUnreliable communication
Sensor NetworksGeneral Ad Hoc Networks
12
Design FactorsWhat are the important features of WSNs?Fault tolerance/reliability
Network should be robust to individual node failuresFailures due to running out of energy, hardware failures, malicious intercept of sensor, etc.
ScalabilityProtocols must scale to thousands or millions of sensor nodesRequires intelligent management of high density nodes
CostMust have cheap sensors
13
Design Factors (cont.)Topology
Deployment: random or deliberate placement of nodesChanges in topology during network operation
New nodes added to the systemNodes failingEnvironmental changes
Energy consumptionSensor functions: sensing, communication, data processingAll require energy
14
Evaluating WSNsWhat are the performance metrics for WSNs?System lifetime
E.g., time until network partitionE.g., time until probability of missed detection exceeds a threshold
Quality of result of sensor networkApplication-specific measureLatency of data transferSNR of aggregate data signalProbability of missed detection or false alarm
Tradeoffs can be made among network parametersE.g., can reduce quality of result of sensor network to increasesystem lifetime
15
Taxonomy of WSN ArchitecturesIn what ways do sensor networks for various applications differ?Data sink(s)
Embedded within networkNetwork edgesMobile access pointOne or several
Sensor mobilityTypically stationary sensorsSome projects use mobile sensors
ZebraNetMilitary operationsSelf-propelled sensorsRobots
16
Taxonomy (cont.)Sensor resources
MemoryProcessingTransmit power (fixed vs. variable)Locations/density
Traffic patternsEvent-driven applicationsContinuous data generationQuery-driven applications
17
Design IssuesNew protocols neededMAC
Cooperative nature of sensor networks (fairness not an issue)Exploit traffic patternsEnergy efficiency extremely important
Reduce idle listeningReduce unnecessary reception
RoutingDifferent traffic modelsData dissemination rather than point-to-point routingData-centric rather than address-centricLocation-aware sensorsResource-aware routing neededExploit local aggregation
18
Design Issues (cont.)Topology control
Reduce idle power consumption nodes sleepCreate fully-connected dominating set from active routers
Transmission power controlHow to avoid “hot spot” problem?Provide connected network
QoS ManagementQoS determined by content of data rather than amountTransport layer
Intelligent congestion managementThrottle back irrelevant data rather than each node’s sending rate
CoverageEnsure enough sensors provide dataK-coverage: each location monitored by at least K sensors
19
Design Issues (cont.)Time synchronization
Very important in sensor networksNeeded to determine if event sensed by two sensors is in fact the same eventNeeded to determine object speedApproaches
GPS – expensive, not energy-efficientNTP (used in computer networks) – not enough precisionNewer approaches being researched
Romer’s AlgorithmReference-Broadcast Synchronization (RBS)
20
Design Issues (cont.)Localization
Important for same reasons as time synchronizationOften times, only relative position is necessary
GPS is overkill and unattractive for energy reasonsRSSI used to infer distancesTime of Arrival (ToA)Time Difference of Arrival (TDoA)Angle of Arrival (AoA)
Sensor can find its own location using received beaconsSensor can have other nodes measure its location
Sensor sends beacon message and neighbors use trilaterationbased on signal strength measurements
Problem – small scale fading
21
Research Issues (1)Appropriate QoS model
Traditional networks: delay, packet delivery ratio, jitter Sensor networks: probability of missed detection of an event, signal-to-noise ratio and network sensing coverageDifficult to translate these data-specific QoS parameters into meaningful protocol parameters
Cross-layer ArchitecturesEntire protocol stack tailored to specific needs of WSN applicationProtocols should be integrated with hardwareTrade-off: generality and ease of network design to achieve lifetime increases
22
Research Issues (2)
ReliabilityLinks and sensors may fail, temporarily or permanentlyMust design protocols to provide reliable service with these failures
Heterogeneous ApplicationsSensor nodes may be shared by multiple applications with differing goalsProtocols must efficiently serve multiple applications simultaneously
Heterogeneous SensorsHow to make best use of resources in heterogeneous sensor networks
23
Research Issues (3)Security
How much and what type of security is really needed? How can data be authenticated? How can misbehaving nodes be prevented from providing false data? Can energy and security be traded-off such that the level of network security can be easily adapted?
ActuationEventually sensor networks will “close the loop”Data do not need to reach base station Current models for sensor networks may not be valid
24
Research Issues (4)Distributed and Collaborative Data Processing
How to best process heterogeneous data?How much data and what type of data should be processed to meet application QoS goals while minimizing energy drain?
Integration with Other NetworksSensor networks may interface with other networks, such as a WiFi network, a cellular network, or the InternetWhat is the best way to interface these networks? Should the sensor network protocols support (or at least not compete with) the protocols of the other networks? Or should the sensors have dual network interface capabilities?
25
Discussion