Wireless Sensor Networks
Xenofon “Fontas” Fafoutis
14/04/2011Wireless Sensor Networks2 DTU Informatics, Technical University of Denmark
Sensors
14/04/2011Wireless Sensor Networks3 DTU Informatics, Technical University of Denmark
Wireless Sensor Networks
Sink
Sensor
Sensed Area
14/04/2011Wireless Sensor Networks4 DTU Informatics, Technical University of Denmark
Outline
• Wireless Sensor Networks
–Types and Topologies
–Applications
–System Issues and Standards
• Energy Harvesting
• Networking Challenges
• Open Research Problems
14/04/2011Wireless Sensor Networks5 DTU Informatics, Technical University of Denmark
Types of Nodes
Sensor
–Low resources
–Inexpensive
–Energy constraints• Main challenge!!
Sink
–High resources
–AC power supply
–Internet connection (typically)
Typically traffic is generated by the sensors and it is directed to the sink
14/04/2011Wireless Sensor Networks6 DTU Informatics, Technical University of Denmark
Unstructured vs. Structured
Unstructured
–Dense
–Ad hoc
Structured
–Fewer sensors
–Strategic positions
Forest Railway
14/04/2011Wireless Sensor Networks7 DTU Informatics, Technical University of Denmark
Topologies
Multi-sink
–Increased cost
–Increased performance
–Reliability
Mobile sinks
–Move and gather data
14/04/2011Wireless Sensor Networks8 DTU Informatics, Technical University of Denmark
Topologies: Single- vs. Multi-Hop
Single-Hop
–Short coverage
–Less challenging
–Higher deployment costs per m2
Multi-Hop
–Large Coverage
–Challenging
14/04/2011Wireless Sensor Networks9 DTU Informatics, Technical University of Denmark
Topologies: Wireless Mesh Network
Very large sensed areas
–Wireless links of several hundreds of meters
Sink w/o Internet Connection
Central Station w/ Internet Connection
14/04/2011Wireless Sensor Networks10 DTU Informatics, Technical University of Denmark
Outline
• Wireless Sensor Networks
–Types and Topologies
–Applications
–System Issues and Standards
• Energy Harvesting
• Networking Challenges
• Open Research Problems
14/04/2011Wireless Sensor Networks11 DTU Informatics, Technical University of Denmark
Application Types
• Monitoring
– Environmental, industrial and health monitoring
– Factory and process automation
– Logistics storage support
• Tracking
– Tracking objects, animals, people and vehicles
– Military, business, public transportation networks
14/04/2011Wireless Sensor Networks12 DTU Informatics, Technical University of Denmark
Traffic Classification
• Continuous
– Mainly on monitoring applications
– Predictable and static
• Event-Driven
– Mainly on tracking applications
– Threshold alerts
– Unpredictable triggering
• Request-Reply
– Predictable triggering
• Hybrid
14/04/2011Wireless Sensor Networks13 DTU Informatics, Technical University of Denmark
Application Requirements
• End-to-end delay
– Tracking, alerting applications
• Reliability
– Long-term monitoring for off-line analysis
• Main trade-off / challenge of WSNs
– Application requirements vs. Energy constraints
Can you name some applications?
14/04/2011Wireless Sensor Networks14 DTU Informatics, Technical University of Denmark
Typical applications
• Environmental monitoring
– Indoor environment control: light, temperature, status of windows and doors, indoor air pollution
– Great Duck Island: Sense the environment that birds live (temperature, pressure, humidity)
• Military applications
– A line in the Sand: Sensors that can detect metallic objects, tracking and classifying moving objects
• Support for logistics
– Storage management of barrels by BP: Detect incompatibilities in storage that may lead to explosions
• Human-centric applications
– Support for senior citizens: Identify behaviors, indicate early stages of disorders, recording if they are taking medication and detect emergencies
• Other
– Six-sensor glove: Movement and gesture recognition
14/04/2011Wireless Sensor Networks15 DTU Informatics, Technical University of Denmark
Outline
• Wireless Sensor Networks
–Types and Topologies
–Applications
–System Issues and Standards
• Energy Harvesting
• Networking Challenges
• Open Research Problems
14/04/2011Wireless Sensor Networks16 DTU Informatics, Technical University of Denmark
Operating System and Standards
• Standards for Low-Rate Wireless Personal Networks (LR-WPN)
– IEEE 802.15.4
• Defines the PHY and MAC layers
– Zigbee
• Defines the NET and APP layers
• TinyOS
– A tiny operating system designed for sensor networks
14/04/2011Wireless Sensor Networks17 DTU Informatics, Technical University of Denmark
Challenges: Localization
Localization
– The problem of determining a node’s position
• Challenging in unstructured topologies
– Important for applications, routing protocols (e.g. geographic routing)
– Straightforward solution: GPS
• But, requires line of sight to satellites, consumes energy, increases cost
– Alternative estimation approaches
• E.g. Received Signal Strength Indicator (RSSI) methods
14/04/2011Wireless Sensor Networks18 DTU Informatics, Technical University of Denmark
Challenges: Synchronization
Synchronization
– The problem of assuring that different nodes have a common notion of time
– Important for applications (correlating data) and networking protocols (time scheduling, coordinated duty cycles)
– Known problem of distributed systems
• Typical solutions are unsuitable due to the limited resources
14/04/2011Wireless Sensor Networks19 DTU Informatics, Technical University of Denmark
Challenges: Security
Security
– The problem of protecting sensed data and resources from attacks and misbehaviors
– Attacks classification
• On secrecy and authentication
– Confidentiality, authorization, authentication
• On availability
– Denial-of-Service (DoS) attacks
• On service integrity
– False data values from compromised nodes
14/04/2011Wireless Sensor Networks20 DTU Informatics, Technical University of Denmark
Outline
• Wireless Sensor Networks
–Types and Topologies
–Applications
–System Issues and Standards
• Energy Harvesting
• Networking Challenges
• Open Research Problems
14/04/2011Wireless Sensor Networks21 DTU Informatics, Technical University of Denmark
Energy Harvesting
• Battery-based WSNs
– Sacrifice performance for lower energy consumption
– Eventually will die and need battery replacement
• Often not even possible (e.g. underground sensors)
• Energy-Harvesting WSNs
– Extracting energy from the environment
– Infinite lifetime but energy not always available
• Energy sources have spatiotemporal variations
– Batteries operate as energy buffers
14/04/2011Wireless Sensor Networks22 DTU Informatics, Technical University of Denmark
Classification of Energy Availability
• Uncontrollable but predictable
– E.g. Solar energy
• Uncontrollable and unpredictable
– E.g. Vibrations in an indoor environment
• Fully controllable
– E.g. Flush-lights used to generate energy
• Partially controllable
– E.g. A deployed energy source
Can you name some energy harvesting sources?
14/04/2011Wireless Sensor Networks23 DTU Informatics, Technical University of Denmark
Energy Sources
• Electromagnetic radiation
– Solar power
– Ambient indoor light
• Thermal energy
– Room radiator
– Machines
– Body temperature
• Mechanical energy
– Wind power, air currents
– Water flows in natural channels (e.g. rivers) or in pipes
– Blood flow and breathing
– Vibrations
• Acoustic noise
– High noise levels (e.g. concerts)
14/04/2011Wireless Sensor Networks24 DTU Informatics, Technical University of Denmark
Design Principles
Battery-Based WSNs• Maximize the lifetime while maintaining a minimum performance
• Save as much energy as possible
• Distribute the tasks and computation load as much as possible
Energy-Harvesting WSNs• Maximize performance while maintaining energetic sustainability
• Use the available harvested energy
• Use the nodes that have access to more energy to cover for nodes that they don’t to
14/04/2011Wireless Sensor Networks25 DTU Informatics, Technical University of Denmark
Outline
• Wireless Sensor Networks
–Types and Topologies
–Applications
–System Issues and Standards
• Energy Harvesting
• Networking Challenges
• Open Research Problems
14/04/2011Wireless Sensor Networks26 DTU Informatics, Technical University of Denmark
Transmission Power
• The power of the signal transmission
•Directly affects the transmission range
– Number of neighbors
•Main energy consumption factor
– The relation of transmission power and range is not linear!
• Transmission Power policies
– Static Approaches
– Adaptation Algorithms
14/04/2011Wireless Sensor Networks27 DTU Informatics, Technical University of Denmark
Transmission Power Trade-off
��� � ����
��
Power of received signal
Constant Distance
Transmission Power
Path loss exponent typically in [2-4]
In order to maintain the same received signal strength over the half distance we need 2�times less transmission power!
Transmission range
R/2RThe number of neighbors are decreased. Hence, the delay to reach the sink is increased.
14/04/2011Wireless Sensor Networks28 DTU Informatics, Technical University of Denmark
Duty Cycles
• Time division into active and sleeping periods
– Energy conservation in battery-based WSNs
– Energy neutral operation in EH-WSNs
• Energy consumption equal to harvested energy
– Introduce additional delay: sleeping delay
• Duty cycle algorithms
– Synchronized duty cycles
• Used in battery-based WSNs
• Same duty cycle for all sensors in an area
– Individual duty cycles
14/04/2011Wireless Sensor Networks29 DTU Informatics, Technical University of Denmark
Individual Duty Cycles
• Important for EH-WSNs because of
– Different energy harvesting rates
• Transmitters and receivers need to be awake at the same time
– Straightforward for single-hop topologies
• The receiver (sink) is always awake
– Challenging for multi-hop topologies
• Energy wasted listening for the receiver to wake up
14/04/2011Wireless Sensor Networks30 DTU Informatics, Technical University of Denmark
Routing
• The problem of selecting the optimum path from the source to the sink
– Objectives: performance and energy conservation
• Is the shortest path always the best solution?
– No!
Path A
Path B
Path B is the shortest path but is it the best solution when:• The batteries of Path A nodes are fuller?• The nodes of Path A can harvest more energy?• The conditions on Path B are bad (many retransmissions)?• …
14/04/2011Wireless Sensor Networks31 DTU Informatics, Technical University of Denmark
Routing Metrics
• Cost functions that evaluate a link or a path
– The lower the better
• Factors that routing metrics shall consider
– Shortest path
– Transmission power requirements
– Energy profile (residual battery, harvesting rate)
– Channel conditions
– Duty cycles
– Traffic load (for load balancing)
The importance of each factor is different for different objectives
14/04/2011Wireless Sensor Networks32 DTU Informatics, Technical University of Denmark
Outline
• Wireless Sensor Networks
–Types and Topologies
–Applications
–System Issues and Standards
• Energy Harvesting
• Networking Challenges
• Open Research Problems
14/04/2011Wireless Sensor Networks33 DTU Informatics, Technical University of Denmark
Some Open Research Problems
• Transmission Power Assignment for EH-WSNs
• Routing metrics for EH-WSNs
• Load distribution for EH-WSNs
• …
14/04/2011Wireless Sensor Networks34 DTU Informatics, Technical University of Denmark
Transmission Power Assignment for EH-WSNs
• Problem: Assign the transmission power of each sensor considering the fact that each node has a unique energy profile (residual battery, harvesting rate)
• Objective: Energetically capable sensors can cover for less capable
• Example scenario:
As the shadow moves, the energy harvesting rate of each node changes. Capable sensors can increase their power consumption to help the incapable sensor regenerate.
14/04/2011Wireless Sensor Networks35 DTU Informatics, Technical University of Denmark
Routing metrics for EH-WSNs
• Problem: Design a routing algorithm / cost function that evaluates and select the optimum path for EH-WSNs.
• Approach: Different factors affect differently the performance and the energetic sustainability. The importance of each factor needs to be evaluated.
• Example scenario:
• Which path minimizes the delays?• Which path contributes more to the
energetic sustainability of the network?
• Path A where shadows and keep the harvesting rate low?
• Or Path B where an obstacle creates a lots of channel errors and, thus, needs many retransmissions?
Path A
Path B
14/04/2011Wireless Sensor Networks36 DTU Informatics, Technical University of Denmark
Load distribution for EH-WSNs
• Problem: Distribute the sensing between sensors with different energy profiles.
• Example scenario:
Assuming that the sink is interesting in only 1 value per period for the red circle area. How shall this energetically consuming task should be distributed among the sensors?
14/04/2011Wireless Sensor Networks37 DTU Informatics, Technical University of Denmark
Thanks
• What’s next?
– Questions / Clarifications
– Exercises