Research Directions in Wireless Sensor Networks

1
Research Directions in Wireless Sensor Networks

• Sandeep Gupta

2
Topics

• Types of Applications
• Difference between WSN traditional sensor and
  ad hoc networks
• Research Issues

3
Types of Applications

• According to Culler et al. CES04
• Monitoring space
• Habitat monitoring, precision agriculture, indoor
  climate control, surveillance, treaty
  verification, intelligent alarms
• Monitoring things
• Structural monitoring, ecophysiology,
  condition-based equipment maintenance, medical
  diagnostics, urban terrain mapping
• Monitoring the interactions of things with each
  other and the encompassing space
• Wildlife habitat monitoring, disaster management,
  emergency response, ubiquitous computing
  environments, tracking asset, healthcare,
  manufacturing process flow.

4
Types and Features of Traditional Sensor Networks
SAC04

• Types I -Remote deployment
• Sensors are placed far from sensing phenomenon.
• Large sensors use complex signal processing
  techniques to distinguish the targets (sensed
  objects) from environmental noise.
• Type II in-situ depolyment
• Several sensors that perform only sensing task
• Position of sensors and communications topology
  is carefully engineered
• Sensors transmit time-series of the sensed
  phenomenon to the central node which performs
  computations and fuses data.

5
Features of Wireless (Micro/Embedded) Sensor
Networks

• In-situ or embedded
• Sensor are deployed either close or inside the
  object of study
• Dense deployment
• Ad hoc topology
• Sensors are resource constrained (energy,
  bandwidth, processing, memory)
• In-network processing (data aggregation, fusion,
  array signal processing)

6
Differences between Ad Hoc Networks and Sensor
Networks

• nodes several order of magnitude higher
• Sensor nodes may not have global identification
  (ID)
• Sensor nodes densely deployed
• Sensor nodes more prone to failure
• Topology changes frequently
• Sensor nodes mainly use broadcast communication
  paradigm whereas ad hoc networks use
  point-to-point communication
• Sensor nodes are more limited in power,
  computational capabilities, and memory
• Sensor networks are usually application-specific

7
Research Directions Sensor Network Organization

• How should the sensor networks be organized?
• Hierarchical (Tiered) is preferred from energy
  consumption perspective
• How many levels?
• E.g. Two Tiered Sensor -gt PDA (microserver) -gt
  Basestation (sink)

8
Research Direction Routing and In-network
processing

• Routing responses to attributed-based queries
  as opposed to data from one network address to
  another.
• Efficiency demands in-network processing
• Routing needs to be integrated with and
  influenced by the application
• Different from Internet-style routing

9
Research Direction Automatic Localization

• Needed for integration of information gleaned by
  various sensors
• Location information of sensors in phenomenon Ps
  region can be used to determine additional
  information such as its size and speed.
• E.g. combining binary decision (within P or
  outside P) to determine its shape.

10
Research Direction Automatic Time Synchronization

• Useful for sensing time-varying or mobile
  phenomenon.
• Timestamped data from different sensors can be
  combined to estimate phenomenon's velocity.
• In some cases GPS can provide position and a
  global clock
• Requires line-of-sight to several satellites
• Does not work inside building, underwater, when
  jammed by enemy, on Mars
• May be too expensive high cost, energy
  consumption

11
Research Direction Distributed Signal Processing

• Traditionally, centralized sensor fusion
• Signal processing resolve low-level sensor
  signals (acoustic or seismic) into higher level
  sensors (e.g. cars or earthquake).
• Target tracking, signal enhancement (beamforming
  combining signals from several sensors to
  reduce noise)
• Challenge how to get best signal processing
  results within the bandwidth and computational
  constraints of the sensing platforms?

12
Research Direction Storage, Search and Retrieval

• Sensor network can produce large volume of data
• Limited amount of data may be stored locally at
  sensors
• Challenge Data mining over a massively
  distributed database which is under energy,
  bandwidth, and storage constraints
• Data mining over data streams rather than data
  stored on secondary storage

13
Research Direction Actuation

• Actuation extend the capability of network in two
  ways
• can enhance sensing task
• pointing camera, aiming antenna, repositioning
  sensors
• can affect the environment
• Opening valves, emitting sound
• Common actuation task is sensor mobility
• Size and shape of a target can be determined more
  accurately if the sensors can move
• Challenge making the right actuation decisions.

14
Research Direction Simulation, Monitoring, and
Debugging

• How can a designer evaluate a system where, by
  definition, the information necessary for the
  evaluation is not available?
• Most of the raw data is not seen outside the
  network
• How can we be sure that the final, high-level
  sensing results delivered by the system is an
  accurate reflection of the state of the
  environment when sensors are deployed where
  there is insufficient energy and bandwidth to
  record all the raw data?
• Simulation becomes crucial
• simulators may be the only environment in which a
  sensor network can both run at a large scale and
  record each raw datum.

15
Research Direction Security and Privacy

• Physical security of nodes embedded in the
  environment cannot be ensured
• Security protocols different from those in
  Internet servers
• Attackers can modify node h/w, replace it with
  malicious counterpart, fool the sensors in making
  observations that do not accurately reflect the
  environment
• Limited resources
• End-to-end encryption prohibits in-networking
  processing (data aggregation)
• Sensor network can be used to gather private data
  (record location, time, action etc. of
  individuals).

16
Research Issues at various Layers

• Application Layer
• Sensor Management Protocol
• Task Assignment and Data Advertisement
• Sensor Query and Data Dissemination
• Transport Layer
• Event to Sink (or Base Station)
• Sink to Event
• Network Layer
• Data-centric, energy-efficient routing protocols
• Data Link Layer
• medium access with built in power conservation,
  mobility management, and failure recovery
  strategies
• Application specific Error control
• Physical layer
• Efficient frequency selection, carrier frequency
  generation, modulation and data encryption

17
Sensor Management Protocol

• Allows accessing a sensor network remotely via
  Internet
• Makes underlying h/w and s/w transparent to
  sensor management application
• Uses attribute-based naming and location-based
  addressing to access nodes
• Perform several management tasks e.g.
• Introduce rules related to data aggregation,
  attribute-based naming and clustering to the
  sensor nodes
• Time synchronization
• Turning sensor nodes on and off
• Authentication, key distribution
• Moving sensor nodes
• Localization

18
Transport Layer

• Traditional transport layer issues
• Bridge application and network layers by
  application multiplexing and demultiplexing
• Provide data delivery service between the source
  and the sink with error control mechanism
  tailored to the reliability requirements of
  application layer
• Regulate the amount of traffic injected in the
  network using flow and congestion control.

19
Event-to-Sink Transport

• Event-to-Sink reliability (reliable event
  detection at sink) rather than end-to-end
  reliability (reliable delivery of individual
  packets from a source to destination)
• Sink is only interested in the collective
  information of sensor nodes within the event
  radius and not their individual data

sensor
Event radius
sink
20
Sink to Sensors Transport

• Sink to sensors (reverse) path used to send
  operational or application-specific data from
  sink to sensors
• OS binaries, programming retasking configuration
  files, application-specific queries and commands
• Requires 100 reliability
• Requires retransmissions and acks
• Local retransmisions and nacks preferred
• Requires multihop one-to-many (multicast)
  communication

21
Some Approaches to Routing in Wireless Sensor
Networks
22
Directed Diffusion

• Used for Event-Driven Queries
• Inject a Query into the Sensor Network
• Flood the Request throughout the Network
• Transmit Responses from Observers to Sink
• Combine Similar Responses for Efficiency
• Provide Data Aggregation at Junctions
• Maintain Efficient Paths from Observer
• Develop Alternative Paths around Faulty or
  Sleeping Nodes

23
Tree-Based Forwarding
Tree is created root is the base station
Most communication is localized. Hierarchical
structure. Workload of sensors is not
balanced. Limited fault tolerance.
24
Leach

• Used for Periodic Communication
• Set-up Phase picks Clusterheads
• Number of Clusterheads Predefined
• Randomly choose Clusterheads
• Steady-State Phase supports communication
• Nodes listen for Clusterhead advertisements
• Join closest Cluster-head
• Clusterhead Aggregates Data
• Clusterhead Transmits to the Base Station

25
Trajectory-Based Forwarding

• Useful for dense sensor networks
• Requires position information
• Sensors follow a parametric path
• No routing tables or routing information
• Local computation to find next hop
• Highly scalable

Path follows a Sine wave
26
Redundant Nodes

• Extra wireless sensors can aid routing protocols
  in several ways
• Compensates for unbalanced loads
• Allows double-checking of sensor readings
• Extends sensor network lifespan
• Useful for sensors with passive power supplies
• Enables sensors to recharge
• Reduces overall energy demands

27
References

• CES04 D. Culler, D. Estrin, and M. Sirvastava,
  Overview of Sensor Networks, IEEE Computer,
  41-49, 37(8), Aug. 2004.
• SAC04 W. Su, O. Akan, and E. Cayirci, Chapter
  2 Communication Protocols for Sensor Networks,
  in Book Wireless Sensor Networks, Raghavendra,
  Shivalingam, Znati (Eds), KAP, 2004.
• ES04 J. Elson and D. Estrin, Chapter 1 Sensor
  Networks A Bridge to the Physical World, in
  Book Wireless Sensor Networks, Raghavendra,
  Shivalingam, Znati (Eds), KAP, 2004.