A decade of Wireless Sensor Network research - Are we there yet?

Transcription

1 A decade of Wireless Sensor Network research - Are we there yet? Prof. Sanjay Jha School of Computer Science and Engineering University Of New South Wales sanjay@cse.unsw.edu.au

2 The Push: Technological Advances Moore s law: # of transistors on a cost effective chip doubles every year or two To continue for next 10/20 years Miniaturised energy capacity In last 20 year AA Nickel Alkaline battery from 0.4 to 1.2 Ah with fast recharging System on a chip (SOC) integration Computing, storage, communications, sensing and energy storage on a cubic millimeter 2

3 History of WSN 1999: Kahn, Katz,Pister: Vision for Smart Dust 2002 Sensys CFP: Wireless Sensor Network research as being composed of distributed systems of numerous smart sensors and actuators connecting computational capabilities to the physical world have the potential to revolutionise a wide array of application areas by providing an unprecedented density and fidelity of instrumentation.

4 ACM Sensys Word Cloud

5 What did we learn? Data: Do scientists need data aggregation? Assumptions: All nodes and their operation are not homogeneous? How many deployments are random? Can we provide a reliable network in the wild? (We do have lots of reliable solutions in papers :)

6 What Next? Smart Dust or Smart Rock? Battery Power Still a problem. Security: Can we take a clean slate approach? Update and Maintenance non CS/EE operators

7 The Pull: Applications Environment and habitat monitoring Precision Agriculture Indoor climate control Military surveillance Intelligent alarms Structural Monitoring Condition based equipment maintenance Medical diagnostics. 7

8 The Australian Context The Tyranny of Sparsity! Biodiversity Monitoring Bushfire management Coastal monitoring 8

9 Sample Research Projects Monitoring Localisation Tracking Participatory Network Newer Areas: Flying Networks Cluttered Environment (Oxford) Outreach What next?

10 Experimental Research Focus Impossible to model real-world phenomena such as: RF propagation vagaries (packet loss, channel delays) sensor calibration issues Measurements always difficult, but always necessary Building prototypes/doing experiments often exposes limitations/serious deficiencies with existing solutions great source of new research problems! 10

11 Environmental Monitoring Study 11

12 Example: Cane Toad Monitoring Cane toads will progressively shell shock an unsuspecting Kakadu environment, in particular they will come close to wiping out native quoll populations, poison large masses of goannas and disturb the food supply of many native animals. 12

13 Detecting Frogs by Their Calls Acoustic features can be used to distinguish the vocalizations of different amphibians. (call rate, call duration, amplitude-time envelope, waveform periodicity, pulserepetition rate, frequency modulation, frequency and spectral patterns.) Frog 1 Frog 2 Frog 3 (Cane toad) Waveform Figures of Three Different Frogs Calls 13

14 Results: Successful Screenshots Detection 14

15 REKF Localization: Localization Using Mobile Robots Localizing static motes using mobile robots combining RSSI readings with acceleration measurements (~0.3m 1m accuracy) 15 NICTA&UNSW

16 Acoustic Chirp Based Localisation Narrowband Signal LFM Chirp Signal Frequency: 20kHz Time-period: 1sec Frequency Span: 20-40kHz Time-period: 1sec Auto-Correlation Adjacent Peaks have Nearly Equal Heights Significant Sidelobe Reduction with respect to the Correlation Peak Difficult Correct Peak Detection Easy Location Accuracy

17 Detection and Tracking 17

18 Tracking Results

19 Secure Network Programming M 1 M 21 M 21 Base station M 21 M 12 Wisec 08. Secure Multi-hop Network Programming with Multiple One-way key chains

20 Participatory Sensing Deploy dedicated sensor networks for specific applications Sensing and communications infrastructure Capital cost Alternate paradigm Exploit existing communication infrastructure , 3G/4G, WiMax, Vehicular communication Exploit existing sensing infrastucture Mobile phones - camera/audio/location (GPS) sensor Humans as sensors - who participate in the sensing process

21 Research Challenges How to effectively sample the environment to reconstruct the ground truth Quality of information Missing data/sparse or insufficient data Redundant data Legality and reliability of data Security and privacy System design

22 Example Application: Consumer sharing of commodity prices

23 HazeWatch Urban air pollution: Causes adverse health effects such as respiratory infections, lung cancer, heart disease Responsible for 2.3% of deaths in Australia each year, and costs NSW $4.7 billion per year Currently measured at only 12 locations in Sydney area Objective: Map urban air pollution at fine spatial resolution Accurately estimate personal exposures Relate personal exposures to health outcomes

24 System Overview

25 Pollution Monitoring Hardware Gas sensors: NO 2, CO, O 3 Microcontroller: ADC, processing Bluetooth: transmission to mobile phone Mobile phone: time/geo-stamp pollution data, upload to database server

26 Real-Time Pollution Map Can be seen live at:

27 Personal Exposure Tool (iphone) Records your location as you move about Displays: Route Plot of exposure to each pollutant (and WHO level) Mean exposure to each pollutant Will aid medical study correlating exposure to health outcome

28 Flying Networks ARC Funded Project to start in 2011at UNSW (Jha, Kanhere and Welsh) developing an understanding of the wireless communication characteristics between mobile aerial vehicles exploiting multiple frequency channels for data collection and ensuring adequate energy efficient coverage of events being monitored.

29 Quadracopter Prototype Various payload capacity (up to 500gm), flying time, motor, wings (hexa..). We plan to put our own sensor nodes as payload for experiments Collaboration with Harvad Robobees Project (Welsh)

30 Underwater Robotic Network Work done at Oxford by former PhD Student Sarfraz Nawaz et al

31 SEEDLING Showcase the use of information and communication technology such as wireless sensor networks (WSNs) to high school students, in order to increase their interest in the field of science and technology, using the left-over equipment from research projects at universities. Seedling.html

32 Outreach: Kings School Deployment Prasant deploying Hop-2 Node

33 Acknowledgement Funding: ARC, CSIRO, DSTO, NICTA, SIT-CRC, UNSW Collaborators: Bulusu (PSU), Chou (UNSW), Corke (QUT), Hu (CSIRO), Kanhere (UNSW), Ostry (CSIRO), Pathirana (Deakin), Savkin & Sivaraman (EET/UNSW), Zic (CSIRO) Details on projects and publications can be found at 33