Sensor Networks
1
OUTLINE
•
•
•
•
Introduction
Applications
Communication Architecture
Protocol Stack
– Research Issues
• Another Challenges
• Conclusion
2
Introduction - Definition
• Sensor Network
– Wireless network consisting of low cost, densely deployed (may be
mobile) sensor nodes.
– Distribution is done in an ad hoc fashion.
– Close to event to be monitored.
– Usually have a limited amount of energy.
• Sensor Nodes
– Battery Power source, low power wireless communication.
– Match Box size form factor and power aware CPU.
– Small embedded OS (TinyOS) and program & data memory is few
KB.
– MEMS sensors (measures light, temp, seismic, acoustics, stress).
http://www.ensc.sfu.ca/~ljilja/cnl/presentations/shameem/project816.pdf
3
Introduction - Advantages
• Improved Signal-to-Noise Ration (SNR)
– combine sources with different spatial perspectives.
•
•
•
•
Greater fault tolerance through redundancy
Coverage of large area
Multiple sensor types can improve performance
Sensors close to object/phenomena of interest can
overcome environmental noise effects
• Can be deployed in regions where infrastructure for
replenishing energy is not available
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part1.pdf
4
Introduction - Difference
What makes a wireless sensor network unique?
• Cellular networks, ad-hoc networks are designed to
– Optimize QoS & Provide high bandwidth
– Provide good throughput/delay characteristics under
medium/high mobility conditions
– Energy consumption of secondary importance
• Sensor networks
– Many nodes, autonomous operation
– Generally stationary devices (or low mobility)
– Traffic periodic or intermittent, low data rate, frequently unidirectional
– Energy management is critical
• Sensing application cannot be ignored !
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part1.pdf
5
Introduction - Uniqueness
• Data-centric networks
– Identity/address of a sensor node is not critical –
its data is the important aspect
• Application specific
– Intermediate nodes can perform data aggregation
or in-network processing
– Network operation driven by global objectives; not
by individual data transfers.
• Resource constraints call for more tightly
integrated layers
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part1.pdf
6
Sensors and Wireless Radio
Sensors
Wireless
Radio
From CDMC http://www.ececs.uc.edu/~dpa
Types of sensors:
- Pressure,
- Temperature
- Light
- Biological
- Chemical
- Strain, fatigue
- Tilt
• Capable of surviving harsh
environments (heat, humidity,
corrosion, pollution, radiation, etc).
• Could be deployed in large
numbers.
7
Applications
• Medical monitoring (e.g., heart rate, glucose
level) and localized drug delivery
• Monitoring structural integrity in buildings
• Tracking vehicles, people, chemical agents,
pollution, weather phenomena
• Seismic monitoring, contaminant/pollution
monitoring
• Precision agriculture
The movie “Twister” is an example of sensor network.
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part1.pdf
8
Ecophysiological Modeling Using
Sensor Array Data
• Spatially and temporally dense microclimate data will allow
• significant advancements in modeling plant production
http://cens.ucla.edu/Estrin/
9
Habitat and Environmental Sensing
UC James Reserve
Habitat Sensing Testbed
Dense micro-climate sensor networks
Extensible Sensor System (ESS)
NIMS: mobile ground and canopy
climate sensors, data mules, and
robotic samplers
Cavity nest micro-climate, remote
observation, bioacoustic sensing
Soil microclimate and chemical
sensors, root/fungi imaging systems
(mini-rhizotron)
James Reserve and Hall Canyon Research Natural Area
http://cens.ucla.edu/Estrin/
10
Error Resilient Contaminant Monitoring
Sensor network error resiliency in complex
media (air-water-soil)
• Working in the context of a real problem in
Palmdale, CA
– partnering with LA County Sanitation District
•
Working in concert with Sensor Group on
broadly applicable sensors, scalable
sensors nitrate and other ionic species
– microsensors matching COTS perfomance
•
Real-time analysis instead of “logging”
– model calibration, forecasting
http://cens.ucla.edu/Estrin/
11
Contaminant Transport Futures
“backbone”
network
adapted from
CA DWR website
http://cens.ucla.edu/Estrin/
•
•
•
•
•
Larger scale, multimedia problems
Linking remote and in situ sensing over multiple scales
Management, visualization, exploration of massive,
heterogeneous data streams
NSF CLEANER Initiative
Challenges:
– Multimedia, Multiscale problems (time and space)
– Multidisciplinary
– Management, visualization, exploration of massive,
heterogeneous data streams
12
Seismic Applications
•
•
Multi-Hopped Radio Linked Array
features
– Time synchronization
– Network event detect
– Sequenced event transmission
– Deployments planned for UCLA
campus and the San Andreas
Fault (100m-10 km)
– Easily reconfigurable
– Worldwide application
Factor Building site
– 72 channels of 24-bit data
– 500 samples per second
continuous data recording
– Internet accessible real time data
monitoring
– Observation of 4 strong
earthquakes, including Alaska &
Japan
http://cens.ucla.edu/Estrin/
Fiberoptic link
Radio link
13
Science Application Systems
• Biology/Biocomplexity
– Microclimate monitoring
– Triggered image
capture
Ecosystems, Biocomplexity
• Marine
microorganisms
Marine Microorganisms
http://cens.ucla.edu/Estrin/
– Detection of a harmful
alga
– Experimental testbed
w/autonously adapting
sensor location
14
Architectural Decisions
• Small size, rugged design, energy-efficient
operation and low cost
• Limited transmission range -> multi-hop
network
• Communication is energy-expensive
– 3 J energy to transmit 1Kb over 100m – equivalent
to 300 million instructions for a 100 Mops
processor
– Rough energy rule: 1 bit = 1000 instructions
– Local processing of information to limit amount of
data that must be exchanged
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part1.pdf
15
Strategies
• Cooperative signal processing
– Collaboration enhances energy efficiency
– Substantial redundancy in data from closely-spaced sensors
• Exploit redundancy of hardware elements
– Deploy higher density of nodes than necessary
– Adjust duty cycle so neighboring nodes are not always active
• Adaptive signal processing
– Maintain balance between energy, accuracy and rapidity of
results
• Hierarchical architecture
– Higher energy, more powerful devices act as cluster heads
– Cluster heads control operation of a set of more limited
devices
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part1.pdf
16
Communication Architecture
Sensor nodes are usually scattered in a sensor field. Each
of these scattered sensor nodes has the capabilities to
collect and route data back to the sink by a multi-hop
infrastructureless architecture through the sink.
17
Communication Architecture (Cont.)
• Design Factor
– Fault Tolerance
– Scalability
– Production Costs
– Hardware Constraints
– Sensor Network Topology
– Environment
– Transmission Media
– Power Consumption
18
Communication Architecture (Cont.)
• Design Factor - Fault Tolerance
– The failure of sensor nodes should not affect the
overall task of the sensor network
–
 λk t
R
(t)

e
wherek
: the failure rate of sensor node k
t :time
period
k
– Reference [2]
19
Communication Architecture (Cont.)
• Design Factor - Scalability
– The density can range from few sensor nodes to few
hundred sensor nodes in a region, which can be less
than 10 m in diameter.
–
 (R)  ( N R2 ) / A
– Reference[3]
20
Communication Architecture (Cont.)
• Design Factor - Production Costs
– The cost of each sensor node has to be kept low
– The state-of-the-art technology allow a Bluetooth
radio system to be less than US$10.
– The cost of a sensor node should be much less than
US$1 in order for the sensor network to be feasible.
21
Communication Architecture (Cont.)
• Design Factor - Hardware Constraints
22
Communication Architecture (Cont.)
• Design Factor - Sensor Network Topology
– Predeployment and deployment phase
– Post-deployment phase
– Redeployment of additional nodes phase
23
Communication Architecture (Cont.)
• Design Factor - Environment
– Sensor nodes usually work unattended in remote
geographic areas.
– At the bottom of an ocean
– In a biologically or chemically contaminated field
– In a battlefield
– In a home or large building
24
Communication Architecture (Cont.)
• Design Factor (Transmission Media)
– RF circuit
μAMPS, WINS
– Infrared
– Optical
Smart Dust mote
25
Communication Architecture (Cont.)
• Design Factor - Power Consumption
– Limited power(<0.5Ah, 1.2V)
– Power Consumption
• Sensing
• Communication
• Data processing
26
Protocol Stack
This protocol stack
combines power and
routing awareness,
integrates data with
networking protocols,
communicates power
efficiently through the
wireless medium, and
promotes cooperative
efforts of sensor nodes.
27
Protocol Stack (cont.)
• Physical Layer
– 915MHz (ISM) band
– Ultra wideband (UWB)
– Impulse radio (IR)
– Open research issue
• Modulation schemes
• Strategies to overcome signal propagation effect
• Hardware design
28
Protocol Stack (cont.)
• Data Link Layer (Medium Access Control)
– Existing MAC protocols cannot be used
– SMACS (Self-Organizing Medium Access
Control for sensor Networks) and the EAR
(Eaves-drop-And-Register) Algorithm
– CSMA-Based Medium Access
– Hybrid TDMA/FDMA-Based
29
Protocol Stack (cont.)
• Data Link Layer (Power Saving Modes of
Operation)
– Sensor nodes communicate using short data packets.
The shorter the packets, the more the dominance of
startup energy.
– Energy-efficient only if the time spent in that mode is
greater than a certain threshold.
– Number of modes can be characterized by its power
consumption and latency overhead.
30
Protocol Stack (cont.)
• Data Link Layer (Error Control)
– FEC (Forward Error Correction)
– ARQ (Automatic Repeat Quest)
31
Protocol Stack (cont.)
• Data Link Layer (Open Research Issues)
– MAC for mobile sensor networks
– Determination of lower bounds on the energy
required for sensor network self-organization
– Error control coding schemes
– Power-saving modes of operation
32
Protocol Stack (cont.)
• Network Layer
– Power efficiency is always an important
consideration.
– Sensor networks are mostly data-centric.
– Data aggregation is useful only when it does
not hinder the collaborative effort of the
sensor nodes.
– An ideal sensor network has attribute-based
addressing and location awareness.
33
Protocol Stack (cont.)
• Network Layer
– Maximum PA route
– Minimum energy (ME) route
– Minimum hop (MH) route
– Maximum minimum PA node route
34
Protocol Stack (cont.)
a) The power
efficiency of the
routes
b) an example of
data
aggregation
c) the SPIN
protocol [15]
d) an example of
directed
diffusion [5]
35
Protocol Stack (cont.)
• Network Layer
– SMECN (Small Minimum Energy Communication
Network)
– Flooding
– Gossiping
– Sensor Protocols for information via negotiation
– Sequential assignment routing
– Low-Energy Adaptive Clustering Hierarchy
– Directed diffusion
36
Protocol Stack (cont.)
• Network Layer (open research issues)
An overview of the protocols proposed for sensor networks is given in
above table. These protocols need to be improved or new protocols
developed to address higher topology changes and higher scalability.
37
Protocol Stack (cont.)
• Transport Layer
– TCP with its current transmission window
mechanisms does match the extreme characteristics
of the sensor network.
– TCP connections are enabled at sink nodes.
– Communication between the sink and sensor nodes
maybe purely by UDP-type protocols, because each
sensor node has limited memory
38
Protocol Stack (cont.)
• Transport Layer (open research issues)
– UDP-type protocol are used in the sensor
network
– Traditional TCP/UDP protocols are used in the
internet or satellite network
39
Protocol Stack (cont.)
• Application Layer
– SMP (Sensor Management Protocol)
– TADAP (Task Assignment and Data
Advertisement Protocol)
– SQDDP (Sensor Query and Data
Dissemination Protocol)
40
Operational Challenges
• Adaptive, self-configuring systems that
respond to an unpredictable environment
• Data processing inside the network
– Perform computation where data is measured to
extract information (compress)
– Important to reduce communication overhead
– Distributed control and signal processing
• Untethered, unattended large-scale systems
– Low-duty cycle design
– Preserve energy by minimizing communication
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part1.pdf
41
Networking Challenges
and Design Principles
• Localization
• Synchronization
• Coverage
– Device management and scheduling
– Connectivity (topology) maintenance
• Routing
• Reliable Data Transport
• Seciruty
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part4.pdf
42
Localization and Synchronization
• Accurate localization and synchronization
– Need to determine where events occur in space
– Critical for fusion of sensor measurements
– Important for coordination of communication
• GPS provides one solution
– Not always available
– Can be too costly, bulky
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part4.pdf
43
Localization
• Most techniques use recursive
trilateration/multilateration
• Some nodes are assumed to know their position
(through GPS for instance).
• These act as beacons by periodically transmitting
their position
• Nodes hearing these beacons use them to
estimate their position
• May be an iterative process
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part4.pdf
44
Localization (cont.)
• Fine-grained (timing/signal strength) or
coarsegrained (proximity)
• Fine-grained
–
–
–
–
Time of flight
Signal strength
Signal pattern matching
Directionality
• Coarse-grained
– Centroid of beacons
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part4.pdf
45
Localization Challenges
• How many beacons are needed?
– Provide good coverage, avoid excessive interference
• Where should beacons be placed?
– Incremental beacon placement
• Place new beacon in region of maximum (average)
localization error
• Controlling beacons
– Scheduling operation to preserve lifetime
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part4.pdf
46
Synchronization
• Metrics
– Precision
• Peer dispersion or with reference to external standard
– Lifetime
• Ranging from persistent to instantaneous
– Scope and Availability
• Geographic span and completeness of coverage
– Efficiency
• Time and energy expenditure
– Cost and form factor
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part4.pdf
47
Sensor Network Synchronization
• Fine-grained, persistent timing is important in sensor
networks
– Data fusion
• detect/estimate the same event
– Local data processing
• Eliminate duplicates through timestamping
• Existing network timing protocols inadequate
– Often conservative in use of bandwidth (e.g., NTP) but neglect
cost of listening
– Heterogeneity of hardware
– Multiple methods of synchronization should be available
– Algorithms should be tunable (precision vs. energy)
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part4.pdf
48
Coverage
• Measures of coverage
– Area coverage : fraction of area covered by sensors
– Detectability : probability sensors detect event
– Node coverage: fraction of sensors covered by other
sensors
– Maximal breach path : intruder is maximum distance
from sensors over entire path
– Maximum exposure path : minimum distance from
sensors over entire path
– K-coverage : entire sensing region must be within
distance K of a sensor
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part4.pdf
49
Coverage - Maintenance
•
•
•
•
Where to position sensors initially?
Where to add new sensors?
Where to move sensors?
When to schedule sensors? (overlapping
sensors should not operate at same time)
• Locate coverage holes, breach areas, best
areas.
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part4.pdf
50
Coverage – Node Scheduling
• Turn off redundant nodes
– Extend system lifetime
– Maintain reliability and coverage/detectability
– Rotate active sensor set
• Optimal Schedules
– Possible to find optimal schedules, but
– Requires global knowledge -> much communication
– Not robust to changes of network state
• Distributed approaches
– Topology control: select active routers
– Sensor mode selection: select active sensors
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part4.pdf
51
Topology Control
• Goals
– Ensure that devices are activated so that all sensors can route
data to sink(s)
– Allow non-selected devices to sleep
– Rotate active routers to balance energy
– Ensure robustness: sensor losses do not disconnect the network
• Example protocols:
–
–
–
–
Geographic Adaptive Fidelity (GAF)
Span
Adaptive Self-Configuring Sensor Network Topologies (ASCENT)
Sparse Topology and Energy Management (STEM)
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part4.pdf
52
Topology Control Examples
• Geographic Adaptive Fidelity (GAF)
– Overlay virtual grid over network
– Activate only one device per cell in grid
• STEM
– Reactively turn on routers when sensors need to send
data
– Paging channel used to wake up neighbours
– A wake-up message – tone or beacon
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part4.pdf
53
Sensor Mode Selection
• Goal
– Select sensing modes (active/inactive, frequency,
resolution) to ensure network objectives achieved
• Example criteria: K-coverage, minimum tracking accuracy,
maximum missed detection probability
– Allow non-selected sensors to sleep
• Example Protocols
– Probing Environment and Adaptive Sleeping (PEAS)
– Node Self Scheduling Scheme (NSSS)
– Coverage Configuration Protocol (CCP)
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part4.pdf
54
Routing
• Routing Challenges
–
–
–
–
–
–
–
Node deployment
Time-critical applications
Node/link heterogeneity
Fault tolerance
Scalability
Network dynamics – environment and nodes
Connectivity and coverage
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part4.pdf
55
Routing - Categories
• Data-centric
– Eevent-driven or query-driven
– Sensor Protocols for Information via
Negotiation (SPIN), Directed Diffusion,
GRAdient Broadcast (GRAB), Rumour
Routing
• Hierarchical (Clustering)
– LEACH, PEGASIS, TEEN and APTEEN,
MECN, SOP, Sensor Aggregates Routing ...
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part4.pdf
56
Routing – Categories (cont.)
• Location-based (Geographic)
– Geographic Adaptive Fidelity (GAF), Geographic and
Energy-aware Routing (GEAR), Greedy Other
Adaptive Face Routing (GOAFR)
• QoS-aware
See pros and cons of some protocols in
http://www.ensc.sfu.ca/~ljilja/cnl/presentations/shameem
/project816.pdf.
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part4.pdf
57
Reliable Data Transport
• Transport layer design is difficult because of
application-specific nature of sensor networks
• Networking layers tend to become fused
(particularly transport and application)
• Goal: design customizable transport layer
• Provide the primitives for reliable transport
• Allow the network designer the flexibility to use
them according to the application needs.
• Example: PSFQ (Pump slowly, Fetch Quickly)
http://www.ece.mcgill.ca/~coates/publications/shortcourse-part4.pdf
58
Security architecture
http://www.i2r.a-star.edu.sg/icsd/SecureSensor/papers/security-map.pdf
59
Sensor security categories
http://www.i2r.a-star.edu.sg/icsd/SecureSensor/papers/security-map.pdf
60
Conclusion
Interesting Areas of Research
• Sensor network still at an early stage in terms of technology
– Currently there is still nothing in use in the real world
• Needs improved or new routing protocols
– The protocols present today have their own set of problems
• No work done on contention issues or for high network traffic
– Currently most researchers claim that although important the present
networks thought about does not have high network traffic
• Most protocols deal with energy efficiency
– There is significant work that can be done with robustness, scalability
• Most results are empirical very little analytical work done
– General perception is that it is hard to do such work
– Better ways to categorize the model
http://www.ensc.sfu.ca/~ljilja/cnl/presentations/shameem/project816.pdf
61
Current Research Projects
62