wireless sensor and sensoractuator networks: research trends...
TRANSCRIPT
Wireless Sensor and SensorActuator Networks: Research Trends, Protocols,
and Applications
Muhammad Farukh Munir Department of Mobile Communications
Institut EurecomINCC 2008
01/05/2008 M. Farukh Munir Institut Eurecom 2
Outline Wireless Sensor Networks
− Motivating Applications
− Research Challenges
− MAC and Routing
Wireless SensorActuator Networks− Motivating Applications
− Self Organization
− Coordinations
− Networking
01/05/2008 M. Farukh Munir Institut Eurecom 3
Wireless Sensor Networks Large number of distributed
sensor nodes
Self Organize in a multihop fashion
Wide range of application potential...
01/05/2008 M. Farukh Munir Institut Eurecom 4
Available Devices Mica motes
Sensor boards
Embedded routers
Wireless gateways
Wireless basestations
Dataloggers
01/05/2008 M. Farukh Munir Institut Eurecom 5
Motivating Applications Industrial Automation
Environmental Monitoring
Marine Monitoring
Intelligent Transportation
Disaster Response
01/05/2008 M. Farukh Munir Institut Eurecom 6
Industrial Automation Microsensors, onboard processing,
and wireless interfaces all feasible at very small scale
Can monitor phenomena “up close”
Embedded Networked Sensing Potential
Will reveal previously unobservable phenomena
01/05/2008 M. Farukh Munir Institut Eurecom 7
Environmental Monitoring• Embed numerous distributed devices
to monitor and interact with physical world
Understand response of wild populations (plants and animals) to habitats over time.
Develop in situ observation of species and ecosystem dynamics
− Automatic identification− Measurements− Harsh environments
01/05/2008 M. Farukh Munir Institut Eurecom 8
Marine Monitoring Ocean life monitoring
Containment flow monitoring...
Identify risky situations before they become exposures.
Subterranean deployment
01/05/2008 M. Farukh Munir Institut Eurecom 9
Intelligent Transportation Transportation and Urban monitoring
Improve safety and reduce vehicle wear, transportation times and fuel consumption.
Optimization of transport systems
Cooperative Vehicle and Highway Systems for safety and control applications
01/05/2008 M. Farukh Munir Institut Eurecom 10
Disaster Response Enhance firstresponders’ ability to
access patients on scene, ensure seamless transfer of data among caregivers, and facilitate efficient allocation of hospital resources
Wearable wireless pulse sensors Devices can be programmed to alert
medical personnel when vital signs fall outside normal conditions
CodeBlue
01/05/2008 M. Farukh Munir Institut Eurecom 11
Common Vision
• Embed numerous distributed devices to monitor and interact with physical world
Exploit spatially and temporally dense, in situ, sensing and actuation
Network these devices so that they can coordinate to perform higherlevel tasks
Requires robust distributed systems of hundreds or thousands of devices
01/05/2008 M. Farukh Munir Institut Eurecom 12
Research Challenges Realtime analysis for rapid response
Massive amount of data
Tight coupling to the physical world and embedded in unattended “control systems”
Untethered, small formfactor, nodes present stringent energy constraints
Communications is primary consumer of energy in this environment
01/05/2008 M. Farukh Munir Institut Eurecom 13
New Design Themes• Longlived systems
− Lowduty cycle operation with bounded latency− Exploit redundancy − Tiered architectures
• Stable systems − MAC layer− MultiHop Routing
• Self configuring systems − Measure and adapt to unattended/unpredictable environment− Exploit spatial diversity and density
01/05/2008 M. Farukh Munir Institut Eurecom 14
Approach Leverage data processing inside the network
− Exploit computation near data to reduce communication Achieve desired global behavior with adaptive localized algorithms
(i.e., do not rely on global interaction or information)− Dynamic, messy (hard to model), environments preclude pre
configured behavior− Cant afford to extract dynamic state information needed for
centralized control or even Internetstyle distributed control
01/05/2008 M. Farukh Munir Institut Eurecom 15
Internet Protocols• Internet routes data using IP Addresses in Packets and Lookup tables in
routers– Humans get data by “naming data” to a search engine– Many levels of indirection between name and IP address– Works well for the Internet, and for support of PersontoPerson
communication• Embedded, energyconstrained, unattended systems cant tolerate
communication overhead
01/05/2008 M. Farukh Munir Institut Eurecom 16
Building LongLived Systems Exploiting redundancy
− Adaptive SelfConfiguration− Supporting lowduty cycle operation
Exploiting Stability Exploiting heterogeneity
01/05/2008 M. Farukh Munir Institut Eurecom 17
The Layered Architecture Layers operate independently without any interaction
Extensively used in PRN literature
Problem: For a given sampling rate, one needs to jointly optimize on channel access rates and the routing in order to optimize on delays
− Restrictions...
− To maintain stability, one needs to be operating far from the max. allowed sampling rate...
− Sampling rate not directly related to the channel access rates...
01/05/2008 M. Farukh Munir Institut Eurecom 18
CrossLayer Architecture (WiOpt'07) Strong Coupling between the Channel Access and the Sampling Process
Node attempts a channel access with some probability
− No packets..., MAC will ask the APP to provide a new packet
− Atleast one packet waiting to be forwarded: serve headofline packet. Note: a node can have atmost one packet in the transmit queue
generated by itself.
01/05/2008 M. Farukh Munir Institut Eurecom 19
Possible Applications Where a sensor network is used to the time variation of a random field over
the space
One can think of temporal priority mechanism for transmitting packets
Reduce the overall transmissions in the network
Node assigns highest priority to the most recent packet generated by itself
− Note: it does not include the forwarding traffic
01/05/2008 M. Farukh Munir Institut Eurecom 20
A Coupled System (PerSeNS'08) Two different Transmit Queues at MAC layer
− For locally generated data
− For forwarding data If node decides to transmit, it checks:
− Both queues are empty: MAC asks APP to sense a packet and attempts a transmission
− One packet is waiting to be forwarded: Transmit packets for each queue in a probabilistic fashion Assign weights to each queue based on traffic loads
01/05/2008 M. Farukh Munir Institut Eurecom 21
MultiHop Routing (CCNC'08) Each route will have some delay
Wardrop equilibrium
Traffic Split matrix
A routing algorithm
− Iterative updation of onehop routing probabilities
− Estimate average delays
− Adapt routing probability according to new delays
01/05/2008 M. Farukh Munir Institut Eurecom 22
Adapting Crosslayered System A new complication: channel access rates
Tune channel access to guarantee long term average data sampling rate
Estimate statistics
− Success rate
− Throughput calculation
Update using a max. min. optimization criteria
01/05/2008 M. Farukh Munir Institut Eurecom 23
Implementation Architecture
01/05/2008 M. Farukh Munir Institut Eurecom 24
Simulated Network
01/05/2008 M. Farukh Munir Institut Eurecom 25
Simulation Results The Layered System
High delays for given sampling rate...
Routing algorithm converges to Wardrop equilibrium
01/05/2008 M. Farukh Munir Institut Eurecom 26
Contd... Delay obtained by varying the
channel access rates
Sensitivity to channel access rates
Convergence to loadbalanced regime is violated
01/05/2008 M. Farukh Munir Institut Eurecom 27
Contd... Lower delays compared to
Layeredsystem
Fast convergence to Wardrop equilibrium
01/05/2008 M. Farukh Munir Institut Eurecom 28
Contd... Traffic split obtained by the
algorithm
Small traffic volume on some routes to ensure probing on alternate routes
Can cope with changes in traffic patterns and network dynamics
01/05/2008 M. Farukh Munir Institut Eurecom 29
Contd... Convergence of channel access
rates for the cross layered system
The converged values are just enough to serve the load on the system
01/05/2008 M. Farukh Munir Institut Eurecom 30
Contd... Delays in Coupled system are
much smaller than the delays in the Layered system
Results show convergence to equilibrium state
01/05/2008 M. Farukh Munir Institut Eurecom 31
Wireless SensorActuator Networks (SANETs)
Consists of a large number of distributed sensors and few actuator/actor nodes
Self Organize in a multihop fashion
Actuator take decisions and then perform appropriate actions upon the environment, which allows remote, automated interaction with the environment.
01/05/2008 M. Farukh Munir Institut Eurecom 32
SANETs Motivating applications
Coordination
− Sensoractuator coordination
− Actuatoractuator coordination Selforganization
Networking
01/05/2008 M. Farukh Munir Institut Eurecom 33
Motivating Applications Disaster relief operations
− Detecting and extinguishing forest fire. Microclimate control in buildings
− In case of very high or low temperature values, trigger the audio alarm actuators in that area.
Distributed Robotics & Sensor Network− (Mobile) robots dispersed throughout a sensor network
Battlefield Applications− Sensors detect mines or explosive substances − Actuators annihilate them or function as tanks
01/05/2008 M. Farukh Munir Institut Eurecom 34
SubKilogram Intelligent Telerobots (SKITs):
Networked Robots having Coordination & Wireless Communication Capabilities
01/05/2008 M. Farukh Munir Institut Eurecom 35
Robotic Mule: Autonomous BattlefieldRobot designed for the Army
01/05/2008 M. Farukh Munir Institut Eurecom 36
Low Flying Helicopter Platform
01/05/2008 M. Farukh Munir Institut Eurecom 37
Grand Challenges SensorActuator Coordination &
Communication ActuatorActuator Coordination &
Communication− REALTIME COMMUNICATION!!!!
01/05/2008 M. Farukh Munir Institut Eurecom 38
SensorActuator Coordination Challenges:
− Which sensor(s) communicate with which actor(s)?
− How should the communication occur?− What are the requirements of the
communication? (i.e., realtime, energy efficiency)
01/05/2008 M. Farukh Munir Institut Eurecom 39
ActuatorActuator Coordination Challenges:
− Which actor(s) should execute which action(s)?− Should they coordinate?− What is the optimal number of actuators
performing the actions?
01/05/2008 M. Farukh Munir Institut Eurecom 40
SANET Communication Architecture
Sensor – Sink – Actuators
No sensoractuator and actuator coordination needed
Similar to conventional WSNs
SemiAutomated ArchitectureSemiAutomated Architecture
Sink
Event Area
01/05/2008 M. Farukh Munir Institut Eurecom 41
SANET Communication Architecture
Sensors – Actuators
Localized information exchange
Low latency
Distributed sensoractuator and actuatoractuator coordination required
Automated ArchitectureAutomated Architecture
Sink
Event Area
01/05/2008 M. Farukh Munir Institut Eurecom 42
SensorActuator Coordination
Multiple Actuator (MA)
Clustering is required
Sensors only need to communicate with sensors within neighborhood to form clusters or groups
Actor Sensor
01/05/2008 M. Farukh Munir Institut Eurecom 43
SensorActuator Coordination
Single Actuator (SA)
Selection of the most appropriate actuator
To select, sensors need to coordinate to each other
01/05/2008 M. Farukh Munir Institut Eurecom 44
Single Actuator (SA) Selecting an actuator may be based on:
− The distance between the event area and actuator
− The energy consumption of the path from sensor to actuator
− The action range of the actuator
01/05/2008 M. Farukh Munir Institut Eurecom 45
SA vs MA SA has lower latency between sensing and acting MA provides the center of an event for actuators MA needs actoractor coordination based on the negotiation
which may result in high communication overhead and latency
01/05/2008 M. Farukh Munir Institut Eurecom 46
ActuatorActuator Coordination
Main Problems
− Which actuator(s) should execute which action(s)?
− Or, how should multiactuator task assignment be done?Task Assignment Problems
− SingleActuator Task (SAT) vs. MultiActuator Task (MAT)
− Centralized Decision (CD) vs. Distributed Decision (DD)
01/05/2008 M. Farukh Munir Institut Eurecom 47
ActuatorActuator Coordination(SAT & MAT)
SingleActuator Task (SAT) − How is the single 'best' actuator selected ?
− For SA case, delaysensitive or insensitive− For SA case, announcement message
MultiActuator Task (MAT)− Selection of most fit (‘best’) actuators among the capable ones for that task− Only a subset of actuators covering the entire event region may perform the
task to save action energy− For MA, DD case, negotiate locally− For MA, CD case, to a decision center
01/05/2008 M. Farukh Munir Institut Eurecom 48
Protocol Stack for SANETs Network congestion:
Transport/Routing/MAC Packet size optimization:
Routing/MAC/PHY
Physical Layer
Application Layer
Transport Layer
Routing Layer
MAC Layer
Power M
anagement Plane
Mobility M
anagement Plane
Fault Managem
ent Plane
Coordination Plane
CrossLayer Melting
01/05/2008 M. Farukh Munir Institut Eurecom 49
LEAD: LowEnergy and Delay Adpative SelfOrganization Framework (EW'06)
Distributed Topology learning
− Energy efficient
− Controlled flooding
− Incremental network deployment
Path acquisition to multiple actuators
− Builds multihop routes to actuators
− Builds multipath connectivity
Actuator selection based on the outcome of a cost function
− Energy constraints or Delay constraints
− Multimetric constraints
01/05/2008 M. Farukh Munir Institut Eurecom 50
LEAD Requires only single hop communication for repair
− Node failures
− Poor signaltonoise ratio on selected links Optimization at Routing results in 'Optimal' Network Lifetime (PEWASUN'07)
− For static deployments
− Easily implementable TDMA MAC with coordinated sleepcycle (PIMRC'07)
− Based on a hybrid scheduling policy
− Network information obtained through LEADlearning framework
Framework 'Optimally' adapted to mobile scenarios with delayenergy constraints
01/05/2008 M. Farukh Munir Institut Eurecom 51
Future Directions Develop technology architecture, software, and middleware components in
the context of driving application prototypes
− Seismic activity and structure response
− Contaminant flow monitoring
− Intelligent transportation systems
− Disaster response
− Underwater sensor networks Integrating actuators into WSNs is a great leap