Amulya kg
15MSS0019
Sensor system technology
PROSPECT ON WIRELESS SENSOR
NETWORK APPLICATIONS AND SYSTEM
FOR DETECTION OF LEAKAGE
MONITERING
Under the guidance of
Prof. Renuka devi
HOD
Communication engineering
ABSTRACT
• wireless sensors network have been used in the recent
years and are in great zeal as of now.it is because of the
miniature devices that wireless sensor networks enable
various applications. They are useful in gathering real
time data with less latency. The protocols and the sensor
nodes chosen are application oriented. The use of sensors
and the possibility of forming them into a network have
revealed many research issues. This paper aims at
presenting WSNs in the application point of view and also
its capabilities and specifications are explained.
INTRODUCTION
• The small node brings us to an allied branch i.e MEMS enables to
carry real word data which was just restricted for logic functions
earlier. Measuring parameters and actuating it in a real time scenario
is made possible by sensors and actuators on silicon.
• The co-ordination and synergy of sensing, processing,
communication and actuating is the next evolution which brings us to
wireless sensor network. This has also given rise to another branch
called internet of things
AREAS OF APPLICATION INCLUDED IN
THE PAPER
• Military applications
• Medical applications
- Research applications
- Industry applications
• Applications in industry and automation
• Application in urban areas
• Main paper – leakage detection and size estimation.
MILITARY APPLICATIONS
• Self healing land mines: The Self-Healing Minefield is an antitank landmine
system, it is an intelligent, dynamic obstacle that responds to an enemy breaching
attempt by physically reorganizing. It contains antitank, each antitank mine monitors
its neighbor‘s state, senses threats to itself, and responds autonomously to those
threats by moving. The mechanism of sensing is based on a distributed self-contained
acoustic location system and accelerometer sensors.
• Soldier detection and tracking: it uses acoustic sensors and daylight cameras for
sensing. The network transmits only those images which are consistent with the
acoustically generated tracks, providing a very high hit rate. Footstep, speech and
some specific impulsive sounds are detectable at various distances from the source.
These algorithms can be combined with the weapon fire detection algorithms. IR
sensors are also used for this applications.
• Sniper detection and localization: The system can detect and
locate any kind of optical sight systems: optical scopes used
by snipers or optronic sight systems. It can also do target
identification using HD cameras. To identify snipers‘ in an
operational area before troupes are deployed - two acoustic
arrays and a daynight video camera are used.
• Time difference of arrival blast localization using a
network of disposable sensors (BL): A system, using a
mesh network of inexpensive acoustic sensors. The system
performs a three-dimensional, time-difference-of-arrival
(TDOA) localization of blasts of various yields in several
different environments. Localization information of the blast
is provided to the end user by exfiltration over satellite
communications. The system is able to perform accurately in
the presence of various sources of error including GPS
position, propagation effects, temperature, and error in
determining the time of arrival (TOA).
CONT’D
MEDICAL APPLICATIONS IN RESEARCH
• Wireless body area network: A body-integratable network, so-called
WBAN, can be formed by integrating these devices. WBAN with sensors
consuming extremely low power is used to monitor patients in critical
conditions inside hospital. Outside the hospital, the network can transmit
patients‘ vital signs to their physicians over internet in realtime. WBAN
usually uses Zigbee, or UWB standard
MobiHealth: major research under BAN. home healthcare using BAN-based
sensors and wireless telephony technology. They are using GPRS/UMTS
wireless communication technology for transferring data.
Fig1: wireless pulseoximeter sensor Fig2: small wearable mote
CONT’D
CodeBlue: CodeBlue is a sensor networks based medical research project being
developed at Harvard. Specific goals for this project include pre-hospital care and in-
hospital emergency care, stroke patient rehabilitation and disaster response. Some
devices and software produced in the project include
 Wireless Vital Sign Sensors
 Wireless two-lead EKG
 Accelerometer, gyroscope and electromyogram (EMG) sensor for monitoring
patients with strokes CodeBlue Software Platform.
there are other research applications based on
Wireless Physiological Sensors for
Ambulatory and Implantable Applications
2 lead EKG sensor
INDUSTRY BASED MEDICAL APPLICATIONS
• LifeSync: Wireless ECG System LifeSync Wireless ECG System is a basic ECG
device that operates using Bluetooth wireless technology. It can collect patient
ECG and includes continuous monitoring in a mobile environment respiratory
data and transmit using two-way radios. It is also designed to interface with other
medical devices.
• HealthTrax: it is a tracking application. RFID tags are attached to
equipment/patients where a real time tracking is possible. The advantages and
disadvantages of this are discussed in the paper.
• ecgAnywhere: ecgAnywhere is a electrocardiograph recorder, that records and
transmits ECG data to a central data warehouse. The device collects useful data
and sends it over the internet via a PC, PDA or other wireless devices.
INDUSTRY AND AUTOMATION APPLICATIONS
• From industrial point of view, ISA SP100 workgroup introduces six classes
(Class 5 – Class 0) for wireless communications based on analysis of industrial,
inter-device wireless communication applications (ISA SP100.11, 2006).
• Class 5 defines items related to monitoring without immediate operational
consequences. This class covers applications without strong timeliness
requirements.
• Class 4 defines monitoring with short-term operational consequences. This
includes high-limit and low-limit alarms etc.
• Class 3 covers open loop control applications, in which an operator, rather than a
controller, ―closes the loop‖ between input and output.
• Class 2 consists of closed loop supervisory control, and applications usually
have long time constants, with the time scale measured in seconds to minutes.
• Class 1 closed loop regulatory control, includes motor and axis control as well as
primary flow and pressure control.
APPLICATIONS IN URBAN AREAS
• Performance monitoring of electrical distribution
system: monitoring system for Electrical Distribution System where
power system's voltage, current and frequency variations are monitored
with the help of wireless sensor network which is both cost effective
and requires less maintenance.
• The sensor nodes are placed on pole transformers and relay the
information measured to the collection center via multi-hop protocols.
The measured values include RMS values of current and voltage,
power factor, temperature, noises like total harmonic distortion.
Automatic remote meter reading system: Each node in wireless
ad hoc network consists of an ammeter, an acquisition card that contains
microprocessor which incorporates ammeter reading into percentage
electricity usage and a GPRS transmitter for communication with rest of the
network.
• The ISM band based ZigBee is utilized for communication between
sensor nodes. Communication is carried out in two tiers; communication
between sensor nodes and data collector is by using the Zigbee
technology while connection between data collector and server is
through Internet.
• Modified LEACH based routing protocol are also proposed.
SMART GRID SENSOR AND ACTOR
NETWORK APPLICATION:
• WMSAN is an integration of multimedia sensor nodes (which sense
the desired parameters e.g. take video, audio and still images as well
as scalar data in smart grid) and actor nodes (that respond to decisions
made at supervisory level by operating/ mobilizing devices).
• The drawbacks of using WMSANs include limited bandwidth
allocation, limited battery life and low reliability. Reliability is a
crucial quality in power systems as a single misinterpretation that can
result in a blackout.
Main application considered
WIRELESS SENSOR NETWORK
BASED MACHINE LEARNING FOR
LEAKAGE DETECTION AND SIZE
ESTIMATION
INTRODUCTION
• pipeline (gas/liquid) health monitoring is very challenging in fluid
transportation systems. Development of smart devices capable of micro-
sensing, on-board processing, and wireless communication capabilities, the
wireless sensor networks are able to equip online learning and reliable event
monitoring and reporting for pipelines.
• This system uses wireless communication and machine learning (WML) to
learn, make decisions and report the critical events like slow /small leakages
in natural gas/oil pipeline autonomously.
• Machine learning is performed on negative pressure wave (NPW) to identify
events based on raw data gathered by individual sensor nodes in network.
• The proposed technique was investigated for performance and capabilities by
a series of trials on a field deployed test bed, with regard to performance of
leakage detection and size estimation in pipelines
• The proposed technique was investigated for performance and
capabilities by a series of trials on a field deployed test bed, with
regard to performance of leakage detection and size estimation in
pipelines
LEAKAGE SYSTEM ARCHITECTURE
• Data Collection and Communication Module: For communication, the system
utilizes ZigBee modules that can be connected to a standard UART connector.
ZigBee standard compliant transceiver can provide an outdoor range of 3200
m (2 miles), indoor range of 90 m (300 ft), transmit power of +18 dBm and
receiver sensitivity of -102 dBm. A 2-cell 7.4 V Lithium Polymer Battery
Pack is used with the high capacity of 13,500 mAh.
• Learning module: A few learning approaches for WML leakage detection
system are considered KNN, GMM, SVM and Naive bayes. Details of these
techniques are not included due to limited space
• Inference module: The knowledge gained in learning module is applied in this
module to make real time decisions. This is the last module in the WML
architecture for leakage detection as shown in Fig. Every senor node in the
network is allowed to interfere from the features it collects based on the
knowledge of training during the learning phase.
FEATURE SELECTION AND
DIMENSIONALITY REDUCTION
• Pre-processing: Wavelet analysis is used to remove signal noise as
well as provide insight into frequency content of the signal 20. The
signal of NPW is denoised using daubechies wavelets and low pass
filter .
• Feature selection and reduction: Many features are extracted to
detect leakage in the pipelines. Time domain features selected are
Expected value ( f1) , Variance ( f2) , Gradient ( f3) and Kurtosis ( f4)
and frequency domain features selected are Pseudo spectrum ( f5)
Entropy ( f6), Power Spectral Density (PSD) ( f7), Percentage of
energy ( f8) and Entropy ( f9). The features are selected from a set of
nine features by performing two tests
• first test is Wilcoxon rank-sum test, which checks classes based on
median values. Second test used is Ansari-Bradley test which tests the
dispersions of benign or non-benign class‘s features with median
values.
• The final features selected are f1, f2, f7, and f9.
Closer view of sensor nodes placed
PERFORMANCE ANALYSIS
• Artificial leaks are created by setting up 5 valves in the layout.
• Pressure transducers are used for sending data to wasp motes
• There are five valves with diameters 0.25, 0.5, 0.75, 1.0, 1.5 inches to create
different sized leaks in the pipelines. As first step, a third level
decomposition is performed using daubechies wavelet for noise removal.
• With an increase in the leakage size, the deflection in the pressure curve
becomes prominent that is why large leaks are comparatively easy to detect
but we focus on detection of small leakages here.
Performance analysis for different dimensions of leak.
CONCLUSION
• Wireless sensor network can leverage the existing machine learning
algorithms to provide robust, reliable and autonomous monitoring.
• The overall performance comparison of the classifiers has been tabulated in
Table.
• The system is developed indigenously and provides capability of reporting
pipeline health, structure and condition related statistics stretched over large
geographic areas.
THANK YOU

Wireless applications in various areas

  • 1.
    Amulya kg 15MSS0019 Sensor systemtechnology PROSPECT ON WIRELESS SENSOR NETWORK APPLICATIONS AND SYSTEM FOR DETECTION OF LEAKAGE MONITERING Under the guidance of Prof. Renuka devi HOD Communication engineering
  • 2.
    ABSTRACT • wireless sensorsnetwork have been used in the recent years and are in great zeal as of now.it is because of the miniature devices that wireless sensor networks enable various applications. They are useful in gathering real time data with less latency. The protocols and the sensor nodes chosen are application oriented. The use of sensors and the possibility of forming them into a network have revealed many research issues. This paper aims at presenting WSNs in the application point of view and also its capabilities and specifications are explained.
  • 3.
    INTRODUCTION • The smallnode brings us to an allied branch i.e MEMS enables to carry real word data which was just restricted for logic functions earlier. Measuring parameters and actuating it in a real time scenario is made possible by sensors and actuators on silicon. • The co-ordination and synergy of sensing, processing, communication and actuating is the next evolution which brings us to wireless sensor network. This has also given rise to another branch called internet of things
  • 4.
    AREAS OF APPLICATIONINCLUDED IN THE PAPER • Military applications • Medical applications - Research applications - Industry applications • Applications in industry and automation • Application in urban areas • Main paper – leakage detection and size estimation.
  • 5.
    MILITARY APPLICATIONS • Selfhealing land mines: The Self-Healing Minefield is an antitank landmine system, it is an intelligent, dynamic obstacle that responds to an enemy breaching attempt by physically reorganizing. It contains antitank, each antitank mine monitors its neighbor‘s state, senses threats to itself, and responds autonomously to those threats by moving. The mechanism of sensing is based on a distributed self-contained acoustic location system and accelerometer sensors. • Soldier detection and tracking: it uses acoustic sensors and daylight cameras for sensing. The network transmits only those images which are consistent with the acoustically generated tracks, providing a very high hit rate. Footstep, speech and some specific impulsive sounds are detectable at various distances from the source. These algorithms can be combined with the weapon fire detection algorithms. IR sensors are also used for this applications.
  • 6.
    • Sniper detectionand localization: The system can detect and locate any kind of optical sight systems: optical scopes used by snipers or optronic sight systems. It can also do target identification using HD cameras. To identify snipers‘ in an operational area before troupes are deployed - two acoustic arrays and a daynight video camera are used. • Time difference of arrival blast localization using a network of disposable sensors (BL): A system, using a mesh network of inexpensive acoustic sensors. The system performs a three-dimensional, time-difference-of-arrival (TDOA) localization of blasts of various yields in several different environments. Localization information of the blast is provided to the end user by exfiltration over satellite communications. The system is able to perform accurately in the presence of various sources of error including GPS position, propagation effects, temperature, and error in determining the time of arrival (TOA). CONT’D
  • 7.
    MEDICAL APPLICATIONS INRESEARCH • Wireless body area network: A body-integratable network, so-called WBAN, can be formed by integrating these devices. WBAN with sensors consuming extremely low power is used to monitor patients in critical conditions inside hospital. Outside the hospital, the network can transmit patients‘ vital signs to their physicians over internet in realtime. WBAN usually uses Zigbee, or UWB standard MobiHealth: major research under BAN. home healthcare using BAN-based sensors and wireless telephony technology. They are using GPRS/UMTS wireless communication technology for transferring data. Fig1: wireless pulseoximeter sensor Fig2: small wearable mote
  • 8.
    CONT’D CodeBlue: CodeBlue isa sensor networks based medical research project being developed at Harvard. Specific goals for this project include pre-hospital care and in- hospital emergency care, stroke patient rehabilitation and disaster response. Some devices and software produced in the project include  Wireless Vital Sign Sensors  Wireless two-lead EKG  Accelerometer, gyroscope and electromyogram (EMG) sensor for monitoring patients with strokes CodeBlue Software Platform. there are other research applications based on Wireless Physiological Sensors for Ambulatory and Implantable Applications 2 lead EKG sensor
  • 9.
    INDUSTRY BASED MEDICALAPPLICATIONS • LifeSync: Wireless ECG System LifeSync Wireless ECG System is a basic ECG device that operates using Bluetooth wireless technology. It can collect patient ECG and includes continuous monitoring in a mobile environment respiratory data and transmit using two-way radios. It is also designed to interface with other medical devices. • HealthTrax: it is a tracking application. RFID tags are attached to equipment/patients where a real time tracking is possible. The advantages and disadvantages of this are discussed in the paper. • ecgAnywhere: ecgAnywhere is a electrocardiograph recorder, that records and transmits ECG data to a central data warehouse. The device collects useful data and sends it over the internet via a PC, PDA or other wireless devices.
  • 10.
    INDUSTRY AND AUTOMATIONAPPLICATIONS • From industrial point of view, ISA SP100 workgroup introduces six classes (Class 5 – Class 0) for wireless communications based on analysis of industrial, inter-device wireless communication applications (ISA SP100.11, 2006). • Class 5 defines items related to monitoring without immediate operational consequences. This class covers applications without strong timeliness requirements. • Class 4 defines monitoring with short-term operational consequences. This includes high-limit and low-limit alarms etc. • Class 3 covers open loop control applications, in which an operator, rather than a controller, ―closes the loop‖ between input and output. • Class 2 consists of closed loop supervisory control, and applications usually have long time constants, with the time scale measured in seconds to minutes. • Class 1 closed loop regulatory control, includes motor and axis control as well as primary flow and pressure control.
  • 11.
    APPLICATIONS IN URBANAREAS • Performance monitoring of electrical distribution system: monitoring system for Electrical Distribution System where power system's voltage, current and frequency variations are monitored with the help of wireless sensor network which is both cost effective and requires less maintenance. • The sensor nodes are placed on pole transformers and relay the information measured to the collection center via multi-hop protocols. The measured values include RMS values of current and voltage, power factor, temperature, noises like total harmonic distortion.
  • 12.
    Automatic remote meterreading system: Each node in wireless ad hoc network consists of an ammeter, an acquisition card that contains microprocessor which incorporates ammeter reading into percentage electricity usage and a GPRS transmitter for communication with rest of the network. • The ISM band based ZigBee is utilized for communication between sensor nodes. Communication is carried out in two tiers; communication between sensor nodes and data collector is by using the Zigbee technology while connection between data collector and server is through Internet. • Modified LEACH based routing protocol are also proposed.
  • 13.
    SMART GRID SENSORAND ACTOR NETWORK APPLICATION: • WMSAN is an integration of multimedia sensor nodes (which sense the desired parameters e.g. take video, audio and still images as well as scalar data in smart grid) and actor nodes (that respond to decisions made at supervisory level by operating/ mobilizing devices). • The drawbacks of using WMSANs include limited bandwidth allocation, limited battery life and low reliability. Reliability is a crucial quality in power systems as a single misinterpretation that can result in a blackout.
  • 14.
    Main application considered WIRELESSSENSOR NETWORK BASED MACHINE LEARNING FOR LEAKAGE DETECTION AND SIZE ESTIMATION
  • 15.
    INTRODUCTION • pipeline (gas/liquid)health monitoring is very challenging in fluid transportation systems. Development of smart devices capable of micro- sensing, on-board processing, and wireless communication capabilities, the wireless sensor networks are able to equip online learning and reliable event monitoring and reporting for pipelines. • This system uses wireless communication and machine learning (WML) to learn, make decisions and report the critical events like slow /small leakages in natural gas/oil pipeline autonomously. • Machine learning is performed on negative pressure wave (NPW) to identify events based on raw data gathered by individual sensor nodes in network. • The proposed technique was investigated for performance and capabilities by a series of trials on a field deployed test bed, with regard to performance of leakage detection and size estimation in pipelines
  • 16.
    • The proposedtechnique was investigated for performance and capabilities by a series of trials on a field deployed test bed, with regard to performance of leakage detection and size estimation in pipelines
  • 17.
    LEAKAGE SYSTEM ARCHITECTURE •Data Collection and Communication Module: For communication, the system utilizes ZigBee modules that can be connected to a standard UART connector. ZigBee standard compliant transceiver can provide an outdoor range of 3200 m (2 miles), indoor range of 90 m (300 ft), transmit power of +18 dBm and receiver sensitivity of -102 dBm. A 2-cell 7.4 V Lithium Polymer Battery Pack is used with the high capacity of 13,500 mAh. • Learning module: A few learning approaches for WML leakage detection system are considered KNN, GMM, SVM and Naive bayes. Details of these techniques are not included due to limited space • Inference module: The knowledge gained in learning module is applied in this module to make real time decisions. This is the last module in the WML architecture for leakage detection as shown in Fig. Every senor node in the network is allowed to interfere from the features it collects based on the knowledge of training during the learning phase.
  • 18.
    FEATURE SELECTION AND DIMENSIONALITYREDUCTION • Pre-processing: Wavelet analysis is used to remove signal noise as well as provide insight into frequency content of the signal 20. The signal of NPW is denoised using daubechies wavelets and low pass filter . • Feature selection and reduction: Many features are extracted to detect leakage in the pipelines. Time domain features selected are Expected value ( f1) , Variance ( f2) , Gradient ( f3) and Kurtosis ( f4) and frequency domain features selected are Pseudo spectrum ( f5) Entropy ( f6), Power Spectral Density (PSD) ( f7), Percentage of energy ( f8) and Entropy ( f9). The features are selected from a set of nine features by performing two tests
  • 19.
    • first testis Wilcoxon rank-sum test, which checks classes based on median values. Second test used is Ansari-Bradley test which tests the dispersions of benign or non-benign class‘s features with median values. • The final features selected are f1, f2, f7, and f9. Closer view of sensor nodes placed
  • 20.
    PERFORMANCE ANALYSIS • Artificialleaks are created by setting up 5 valves in the layout. • Pressure transducers are used for sending data to wasp motes • There are five valves with diameters 0.25, 0.5, 0.75, 1.0, 1.5 inches to create different sized leaks in the pipelines. As first step, a third level decomposition is performed using daubechies wavelet for noise removal. • With an increase in the leakage size, the deflection in the pressure curve becomes prominent that is why large leaks are comparatively easy to detect but we focus on detection of small leakages here. Performance analysis for different dimensions of leak.
  • 21.
    CONCLUSION • Wireless sensornetwork can leverage the existing machine learning algorithms to provide robust, reliable and autonomous monitoring. • The overall performance comparison of the classifiers has been tabulated in Table. • The system is developed indigenously and provides capability of reporting pipeline health, structure and condition related statistics stretched over large geographic areas.
  • 22.