Week 3: Wireless Sensor Networks WSNs

1
Week 3Wireless Sensor Networks (WSNs)

• ???
• lyyu_at_cs.ecnu.edu.cn

2
Outline

• Introduction to WSN
• Research directions
• A WSN for structural monitoring
• Data dissemination
• Directed diffusion
• TTDD
• An energy-efficient MAC for WSN
• System architecture directions

3
Introduction to WSN

• A typical WSN

4
Introduction to WSN

• Main features
• is composed of a large number of low-cost,
  low-power, multifunctional sensor nodes
• sensor nodes communicate untethered in short
  distances
• sensor network protocols and algorithms must
  possess self-organizing capabilities
• only transmit the required and partially
  processed data from nodes to the sink
• the topology changes very frequently

Dont send all raw data
5
Introduction to WSN

• Main applications
• Military applications
• Environmental applications
• Health applications
• Home applications
• Transportation applications

6
Introduction to WSN

• WSN vs. Mobile Ad Hoc Networks
• The number of sensor nodes in a WSN can be
  several orders of magnitude higher than the nodes
  in an MANET
• Sensor nodes are densely deployed
• Sensor nodes are prone to failures
• The topology of a WSN changes very frequently

7
Introduction to WSN

• WSN vs. Mobile Ad Hoc Networks (cont.)
• Sensor nodes mainly use broadcast communication
  paradigm whereas most MANETs are based on
  point-to-point communications
• Sensor nodes are limited in power, computational
  capacities, and memory
• Sensor nodes may not have global ID because of
  the large amount of overhead and large number of
  sensors

One reason why it is challenging and attractive.
8
Introduction to WSN
Quarter Dollar

• Sensor nodes

Crossbow MICA2DOT?? Size(mm) 25 x 6 Price 105
Crossbow MICAz?? (http//www.xbow.com/) Size(mm)
58 x 32 x 7 Price 125
9
Introduction to WSN

• Typical components of a sensor node

10
Outline

• Introduction to WSN
• Research directions
• Core Challenges
• Research Directions
• A WSN for structural monitoring
• Data dissemination
• An energy-efficient MAC for WSN
• System architecture directions

11
Research Directions

• Core Challenges
• Energy Efficient
• Energy is scarce in WSN.
• This requirement pervades all aspects of the
  system's design, and drives most of the other
  requirements.
• Data reduction is key to the energy efficiency of
  the system.
• A perfect system will reduce as much data as
  possible as early as possible.

12
Research Directions

• Core Challenges
• Dynamics
• Reasons
• Nodes failure run out of energy / overheat in
  sun / be carried away by wind / crash due to
  software bugs / be eaten by a wild boar
• Quality of RF communication
• Result
• Self-configuring
• Adaptive

13
Research Directions

• Research Directions
• Tiered architectures
• Routing and in-network processing
• Automatic localization and time synchronization
• Storage, search and retrieval
• Actuation
• Simulation, monitoring, and debugging
• Security and Privacy

14
Research Directions
Memory Cache

• Tiered architectures
• Numerous, small, cheap, disposable nodes
• Lower price, better energy efficiency
• Few, larger, faster, and more expensive hardware
• More capable, more durable

Heterogenous Wireless Sensor Networks
15
Research Directions

• Routing and in-network processing
• Routing is a popular topic in multiple hops
  networks
• In Internet, routing is to find a way to
  transport packets to a particular endpoint
• In a sensor network, efficiency demands that we
  do as much in-network processing (e.g., data
  reduction) as possible
• Aggregating similar data, filtering redundant
  information, and so forth
• Depend on the application

16
Research Directions

• Automatic localization and time synchronization
• Some of the most powerful benefits of a
  distributed network are due to the integration of
  information gleaned from multiple sensors into a
  larger world-view not detectable by any single
  sensor alone.

17
Research Directions

• Automatic localization and time synchronization

• Each sensor can know whether it is within the
  sensed phenomenon or not.
• If the sensor positions are known, integration of
  information from the entire field allows the
  network to deduce the size and shape of the
  target, even though it has no size or shape
  sensors.

18
Research Directions

• Automatic localization and time synchronization
• Nodes require the capability of localizing
  themselves after they have been deployed
• Time synchronization is also a crucial service
  necessary to combine the observations of multiple
  sensors with each other

19
Research Directions

• Automatic localization and time synchronization
• GPS
• Provide nodes with both their position and a
  global clock
• Too expensive, not be practical

20
Research Directions

• Storage, search and retrieval
• Energy, bandwidth, storage, memory, processing
  constrain on sensor nodes
• The traditional database approach is not suitable
  for WSN
• In-network processing

21
Research Directions

• Actuation
• In many cases, a sensor network is an entirely
  passive system
• Actuation can dramatically extend the
  capabilities of a network in two ways
• can enhance the sensing task, by pointing
  cameras, aiming antennae, or repositioning
  sensors
• can affect the environment - opening valves,
  emitting sounds, or strengthening beams

22
Research Directions

• Actuation

23
Research Directions

• Simulation, monitoring, and debugging
• In WSN, simulation and debugging environments
  are particularly important
• e.g. how can we be sure that the final,
  high-level sensing result delivered by the system
  is an accurate reflection of the state of the
  environment?
• Several example TOSSIM, EmStar and sensor
  network extensions to GloMoSim and ns-2

24
Research Directions

• Security and Privacy
• The physical security of the nodes making up the
  network can not be assured
• The limited resources on the smallest sensor
  nodes also can pose challenges
• Sensor networking, similarly, is a technology
  that can be used to enrich and improve our lives,
  or turned into an invasive tool

25
Outline

• Introduction to WSN
• Research directions
• A WSN for structural monitoring
• Introduction
• Three challenges
• Data dissemination
• An energy-efficient MAC for WSN
• System architecture directions

26
A WSN for Structural Monitoring

• Introduction
• Structural health monitoring systems seek to
  detect and localize damage in buildings, bridges,
  ships, and aircraft
• Currently, structural engineers use wired or
  single-hop wireless data acquisition systems to
  acquire such data sets

27
A WSN for Structural Monitoring

• Introduction
• These systems consist of a device that collects
  and stores vibration measurements from a small
  number of sensors
• power and wiring constraints
• impose significant setup delays
• limit the number and location of sensors

Wireless sensor networks can help address these
issues.
28
A WSN for Structural Monitoring

• Introduction
• In this paper, authors describe the design of
  Wisden, a wireless sensor network system for
  structural-response data acquisition.
• Wisden continuously collects structural response
  data from a multi-hop network of sensor nodes,
  and displays and stores the data at a base
  station.

29
A WSN for Structural Monitoring

• Introduction
• Wisden overview
• A typical Wisden deployment will consist of
  several tens of nodes placed at different
  locations on a large structure.
• Each node has an attached accelerometer that is
  capable of sensing up to three channels of
  vibration data, with a configurable sampling rate.

30
A WSN for Structural Monitoring

• Introduction
• Wisden overview
• A base station provides the functionality
  equivalent to a data logger or acquisition
  unitthe ability to store samples and to provide
  near real-time display of samples.
• Nodes self-configure to form a tree topology,
  then send their vibration data to the base
  station, potentially over multiple-hops.

31
A WSN for Structural Monitoring

• Introduction
• Structural response data is generated at higher
  data rates than most sensing applications
  (typically, structures are sampled upwards of 100
  Hz).
• This application requires loss intolerant data
  transmission, and time synchronization of
  readings from different sensors.

32
A WSN for Structural Monitoring

• Introduction
• Three challenges in Wisden
• Reliable Data Transport
• Compression
• Data Synchronization

33
A WSN for Structural Monitoring

• Reliable Data Transport
• In Wisden, nodes self-organize themselves into a
  routing tree rooted at the base station.
• Wisden uses both hop-by-hop and end-to-end
  recovery.

34
A WSN for Structural Monitoring

• Reliable Data Transport
• Hop-by-hop NACK-based reliability scheme
• Each source stores generated vibration data in
  its EEPROM, then transmits the data to its
  parent.
• Parents keep track of sequence numbers of packets
  that they receive, on a per source basis.
• A gap in the sequence number of sent packets
  indicates packet loss.
• Each node maintains a list of missing packets.

35
A WSN for Structural Monitoring

• Reliable Data Transport
• Hop-by-hop NACK-based reliability scheme
• When a loss is detected, a tuple containing a
  source ID and sequence number of the lost packet
  is inserted into this list.
• Entries in the missing packets list are
  piggybacked in outgoing transmissions, and
  children infer losses by overhearing this
  transmission.
• Nodes keep a small cache of recently transmitted
  packets, from which a child can repair losses
  reported by its parent.

36
A WSN for Structural Monitoring

• Reliable Data Transport
• Discussion when does hop-by-hop NACK-based
  reliability not work well?
• First, heavy packet losses can lead to large
  missing packet lists that might exceed the memory
  of the motes.
• More fundamentally, a topology change could cause
  loss of missing packet list information. For
  example, when a node selects a new parent.

37
A WSN for Structural Monitoring

• Reliable Data Transport
• End-to-end recovery scheme
• The end-to-end recovery scheme is essentially
  implemented in much the same way as the
  hop-by-hop scheme.
• It leverages the fact that the base station has
  significantly more memory and can keep track of
  all missing packets.
• The base station attempts hop-by-hop recovery of
  a missing packet.

38
A WSN for Structural Monitoring

• Reliable Data Transport
• Experimental Evaluation
• Deployed 25 mica2 motes on three floors of a
  medium-sized office building
• 15 of those motes were programmed to generate
  artificial traffic
• 10 motes only forwarded traffic, and were placed
  to ensure the multihop topology
• 10 nodes dynamically selected different parents
  to trigger the end-to-end recovery

39
A WSN for Structural Monitoring

• Achieved 100 reliability in all experiments.
• When packet injecting rate is 2 packet/second,
  network essentially collapses and very few of the
  packets are received

• Reliable Data Transport

40
A WSN for Structural Monitoring

• EEPROM ? sources retransmitting
• RAM ? intermediate nodes retransmitting

• Reliable Data Transport

41
A WSN for Structural Monitoring

• Compression
• Motivation
• Even if each of 20 nodes generated only 10
  minutes worth of 3-channel vibration data, it
  would take almost an hour transmit the data to
  the base station assuming a nominal radio
  bandwidth of 2 KBps.

42
A WSN for Structural Monitoring

• Compression
• Approaches
• Event detection only transmit samples that
  exceed a certain threshold
• Progressive storage and transmission stores
  vibration data locally and transmits a lossy
  version (using wavelet compression) of the data
  to the base station

43
A WSN for Structural Monitoring

• Compression
• Event detection
• If samples within a small window have a low value
  and are comparable in value, the structure is
  quiescent
• Such quiescent periods are compressed using
  run-length encoding
• Samples in non-quiescent periods are transmitted
  without compression

Run Length Encoding consists of the process of
searching for repeated runs of a single symbol in
an input stream, and replacing them by a single
instance of the symbol and a run count.
44
A WSN for Structural Monitoring

• Compression
• Event detection
• Discussion does event detection have any
  limitation?
• The number of instrumentation locations is
  constrained by the rate at which a structure is
  expected to vibrate, especially for forced
  vibrations experiment
• This approach does not reduce the user-perceived
  latency of data acquisition. Due to the global
  nature of vibration events, vibration data is
  usually generated from all node simultaneously.

45
A WSN for Structural Monitoring

• Compression
• Progressive Storage and Transmission
• In order to reduce the latency of data
  acquisition
• This approach uses local storage on the motes as
  a in-network cache for raw data and transmits
  low-resolution compressed summaries of data in
  near real-time.
• The raw data can be collected from the
  distributed caches when required.

46
A WSN for Structural Monitoring

• Data Synchronization
• Samples need to be accurately timestamped in
  order to correlate readings from different
  sensors.
• In order to distinguish responses due to
  different events.
• Wisden uses a light-weight approach in that it
  focuses on timestamping the data consistently at
  the base station, rather than synchronizing
  clocks network-wide.

47
A WSN for Structural Monitoring

• Data Synchronization
• In Wisden, each node calculates the amount of
  time spent by a sample at that particular node
  using its local clock.
• This amount is added to an residence time field
  attached to a packet as the packet leaves the
  node.
• Thus, the delay from the time of generation of
  the sample to the time it is received by the base
  station is stored in the packet.
• This is the time the packet resides in the
  network.
• The base station can calculate the time of
  generation of the sample by subtracting the
  residence time from its local time.

48
A WSN for Structural Monitoring

• Data Synchronization
• Time synchronization example

the residence time at the ith hop node (ms)
the propagation delay for the ith hop (ns)
the residence time from A
when base station received the packet
The sample must be generated at
49
A WSN for Structural Monitoring

• Conclusion
• In this paper, the design of a wireless
  structural data acquisition system called Wisden
  is described.
• Reliable Data Transport
• Compression
• Data Synchronization