CS-541

Wireless Sensor Networks

Lecture 4: Introduction to Signal Processing and Learning for WSNs

Spring Semester 2016-2017

Prof Panagiotis Tsakalides, Dr Athanasia Panousopoulou, Dr Gregory Tsagkatakis

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 1
University of Crete, Computer Science Department
Today’s Objectives

•WSN as a sensing platform

• Signal sensing in WSN
• Data processing in WSN

•WSN and the Internet-of-Things (IoT)

• Big Data
• Machine Learning

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 2
University of Crete, Computer Science Department
Wireless Sensor Networks (WSNs)

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 3
University of Crete, Computer Science Department
 WSNs motes

Sensors
• Temperature, humidity, light,
• Accelerometer, gyroscope, magnetometer
• Image/Video, audio
Memory
• Internal/External
Radio
• Low-cost CMOS-based RF Radios
Processor
• Microcontroller (ARM7, Atmel AVR)
Battery
• AA, energy harvesting

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 4
University of Crete, Computer Science Department
Energy consumption in WSNs
Processor Transceiver

Modeling of Node Energy Consumption for Wireless Sensor Networks

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 5
University of Crete, Computer Science Department
WSNs from signal processing perspective

Centralized Processing Distributed Processing

• High performance computing • Low capabilities microcontrollers

• Large communications bandwidth • Short range noise channels
• Abundant power supply • Limited battery/harvesting
• Controlled environments • Exposure to outdoors
• Centralized data storage • Distributed data storage

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 6
University of Crete, Computer Science Department
Energy consumption in WSNs

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 7
University of Crete, Computer Science Department
 Sensing in WSN
Sample Quantize Process Transmit
MSE 0.000243
2 2 0.35 2

1.8 1.8
0.3
1.5
1.6 1.6
0.25
1.4 1.4
1
0.2
1.2 1.2

1 1 0.15 0.5

0.8 0.8
0.1
0
0.6 0.6
0.05
0.4 0.4
-0.5
0
0.2 0.2

0 0 -0.05 -1
0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000 0 50 100 150 200 250 300 350 400 450 500

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 8
University of Crete, Computer Science Department
 Sampling

• Discrete-time representation of a continuous signal

  =

Sampling
5

Sensor ID
10
Spatial Field

15

20

25

30

35

40

45

50
10 20 30 40 50 60 70 80 90 100

Reconstruction Sampling Instance

Time

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 9
University of Crete, Computer Science Department
 Sampling

MSE 0.008366
2 2

1.8 1.8

1.6 1.6

1.4 1.4

1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 10
University of Crete, Computer Science Department
 Sampling

MSE 0.003322
2 2

1.8 1.8

1.6 1.6

1.4 1.4

1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 11
University of Crete, Computer Science Department
 Sampling

MSE 0.001008
2 2

1.8 1.8

1.6 1.6

1.4 1.4

1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 12
University of Crete, Computer Science Department
 Sampling

MSE 0.000243
2 2

1.8 1.8

1.6 1.6

1.4 1.4

1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 13
University of Crete, Computer Science Department
 Signal Quantization

Map values to symbols

• Bits per measurements

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 14
University of Crete, Computer Science Department
 Signal Quantization

Map values to symbols

• Bits per measurements
1
Input
0.8 4 bit/sample quantization
2 bit/sample quantization

Dynamic range
0.6

0.4

0.2

-0.2

-0.4
0 20 40 60 80 100

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 15
University of Crete, Computer Science Department
 Data processing

• What do we do with the data?

• Distributed processing
• Robustness, tolerance
• Centralized
• Speed, power
• Hybrid schemes

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 16
University of Crete, Computer Science Department
 Onboard vs off-board processing
To transmit or not to transmit
Cloud computing offers benefits like increased storage, increased processing capacity, flexibility and
mobility, reduced cost etc.
Case study on Face Recognition (Viola-Jones)
Energy (Joules) Execution Duration (ms)
35 21.000

30 Smartphone 18.000 Smartphone

only only
25 MAUI (Wi-Fi, 15.000 MAUI (Wi-Fi,
10ms RTT)
Energy (Joules)

10ms RTT)
20 MAUI (Wi-Fi, 12.000
MAUI (Wi-Fi,
25ms RTT) 25ms RTT)
15 MAUI (Wi-Fi, 9.000
MAUI (Wi-Fi,
50ms RTT)
10 6.000 50ms RTT)
MAUI (Wi-Fi,
100ms RTT) MAUI (Wi-Fi,
5 3.000 100ms RTT)
MAUI* (3G,
220ms RTT) MAUI* (3G,
0 0
Spring Semester 2016-2017
CS-541 Wireless Sensor Networks
University of Crete, Computer Science Department 220ms RTT)
17
 In-network processing
Energy consumption (mAh)
Centralized

Sampling
Decentralized Computation
Communication

0 0,05 0,1 0,15 0,2 0,25

Equation… Sampling
Energy Consumption (mAh)
Computation
Coefficien…
Power…
FFT
Raw Data…

0 0,1 0,2 0,3 0,4 0,5

G. Hackmann, O. Chipara, and C. Lu, Robust Topology Control for Indoor Wireless Sensor Networks, SenSys 2008.
CS-541 Wireless Sensor Networks
Spring Semester 2016-2017 18
University of Crete, Computer Science Department
 The case of missing values

• Applications need complete network matrices

• Traffic analysis
1
• Spectrum sensing flow 3
• Localization
• Multi-access channel design link 1 link 2
• Network coding
• Anomaly detection flow 2 router 2
• Data aggregation
•…
flow 1
Time 1 Time 2 Time 3 Time T
link 3
Flow 1 3
Flow 2

Flow F

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 19
University of Crete, Computer Science Department
 The case of missing values

• Applications need complete network matrices

Vacant
• Traffic analysis
• Spectrum sensing
-50
• Localization -60
• Multi-access channel design -70
• Network coding -80

• Anomaly detection

Power (dBm)
-90

• Data aggregation -100

•… -110

-120

-130
Time 1 Time 2 Time 3 Time T -140

-150
Frequency 1
500 520 540 560 580 600 620 640 660 680 700
Frequency (MHz)
Frequency 2

Frequency F

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 20
University of Crete, Computer Science Department
 The case of missing values

• Applications need complete network matrices

• Traffic analysis
• Spectrum sensing
• Localization
• Multi-access channel design
• Network coding
• Anomaly detection
• Data aggregation
•…

Channel 1 Channel 2 Channel 3 Channel T

Location 1

Location 2

Location F

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 21
University of Crete, Computer Science Department
 Network Matrices

Network Date Duration Size (flows/links x #timeslot)

3G traffic 11/2010 1 day 472 x 144
WiFi traffic 1/2013 1 day 50 x 118
Abilene traffic 4/2003 1 week 121 x 1008
GEANT traffic 4/2005 1 week 529 x 672
1 channel CSI 2/2009 15 min. 90 x 9000
Multi. Channel CSI 2/2014 15 min. 270 x 5000
Cister RSSI 11/2010 4 hours 16 x 10000
CU RSSI 8/2007 500 frames 895 x 500
Umich RSS 4/2006 30 min. 182 x 3127
UCSB Meshnet 4/2006 3 days 425 x 1527

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017
University of Crete, Computer Science Department 22
 Distributed Data Storage

Distributed Data Storage (DDS)

• Efficient measurements acquisition
• Minimum bandwidth
• Low power consumption
• HQ Field estimation

Distributed MC
• Local field estimation
• Reliable data recovery
• Distributed processing

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 2323
University of Crete, Computer Science Department
 Link Quality Maps

Link Quality Map (LQM)

• Characterization of channel quality among WSN nodes
• Adaptive communications rate → Reduced power consumption
• Efficient data routing
LQM from few samples
• Received Signal Strength Indicator (RSSI)
• Exploit spatio correlations → Low Rank Recovery

Low rank

Singular Values

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 24
University of Crete, Computer Science Department
 The Internet of Things (IoT)

• By the end of 2014, the number of mobile-connected devices

will exceed the number of people on earth, and by 2019 there
will be nearly 1.5 mobile devices per capita.
• Mobile video traffic exceeded 50%of total mobile data traffic
for the first time in 2012
• Almost half a billion (497 million) mobile devices and
connections were added in 2014
• Globally, there were nearly 109 million wearable devices in
2014 generating 15 petabytes of monthly traffic.
• Globally, 46% of total mobile data traffic was offloaded onto
the fixed network through Wi-Fi or femtocell in 2014
• Applications
• Smart appliances
• In-home medical sensors
• Smart sensor tags (farm animals)
• Smart cars
• Smart weather umbrellas
• Beacons

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 25
University of Crete, Computer Science Department
 Enablers of the IoT

• Cheap sensors – Sensor prices have dropped to an average 60 cents from $1.30 in the past 10
years.
• Cheap bandwidth – The cost of bandwidth has also declined precipitously, by a factor of nearly
40X over the past 10 years.
• Cheap processing – Similarly, processing costs have declined by nearly 60X over the past 10
years, enabling more devices to be not just connected, but smart enough to know what to do
with all the new data they are generating or receiving.
• Smartphones – Smartphones are now becoming the personal gateway to the IoT, serving as a
remote control or hub for the connected home, connected car, or the health and fitness devices
consumers are increasingly starting to wear.
• Ubiquitous wireless coverage – With Wi-Fi coverage now ubiquitous, wireless connectivity is
available for free or at a very low cost, given Wi-Fi utilizes unlicensed spectrum and thus does
not require monthly access fees to a carrier.
• Big data – As the IoT will by definition generate voluminous amounts of unstructured data, the
availability of big data analytics is a key enabler.
• IPv6 – Most networking equipment now supports IPv6, the newest version of the Internet
Protocol (IP) standard that is intended to replace IPv4.

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 26
University of Crete, Computer Science Department
Some history
1st Phase – Stand Alone Computer 2nd Phase – The Web

Data Web Sites

Data
Data
Data
App.

Data

App.
Browser
Data

3rd Phase – The Cloud 4th Phase – Cloud + IoT

Data Data Data

Data
Data App. Data
App. App. App.
App.
App.
Data
Data

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 27
University of Crete, Computer Science Department
IoT Meets Big Data

Spring Semester 2016-2017 28

 Big Data in WSN/IoT

Big Data
• Volume: size of data such as terabytes (TB), petabytes (PB), zettabytes (ZB),
• Variety: types of data from difference sources (sensors, devices, social networks, the web, mobile
phones)
• Velocity: how frequently the data is generated (every millisecond, second, minute, hour, day,
week, month, year.) Processing frequency may also differ from the user requirements.
• Challenges
• High volume processing using low power processing architectures.
• Discovery of real-time data-adaptive Machine learning techniques.
• Design scalable data storages that provide efficient data mining.

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 29
University of Crete, Computer Science Department
 What to do with all this data

Machine Learning
• Checkers (1995)
• Chess (1997)
• jeopardy (2011)
• Go (2015)
• Poker (2017)

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 30
University of Crete, Computer Science Department
 Key ML approaches

Support Vector Machines

Regression Models

K-means
Spectral Clustering

Association rules
Distance learning

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 31
University of Crete, Computer Science Department
Applications

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 32
University of Crete, Computer Science Department
Machine Learning in WSN/IoT

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 33
University of Crete, Computer Science Department
 Case study: ageing well

• Activity recognition: sitting,

standing, coughing, walking,
jumping etc. to measure
health fitness.
• Detection of disease:
symptoms + Research in
identification of new
symptoms from big data.
• Disease risk prediction.
• Alert doctors/family in cases
of emergencies.

http://datascience.blog.uhuru.co.jp/machine-learning/safe-aging-with-iot-and-machine-learning/

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 34
University of Crete, Computer Science Department
 Material from Deep Learning tutorial 2016 by Russ Salakhutdinov

CS-541 Wireless Sensor Networks

Spring Semester 2015-2016 35
University of Crete, Computer Science Department
 Material from Deep Learning tutorial 2016 by Russ Salakhutdinov

CS-541 Wireless Sensor Networks

Spring Semester 2015-2016 36
University of Crete, Computer Science Department
 Material from Deep Learning tutorial 2016 by Russ Salakhutdinov

CS-541 Wireless Sensor Networks

Spring Semester 2015-2016 37
University of Crete, Computer Science Department
Deep Learning

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 38
University of Crete, Computer Science Department
 Rest of the course

• How can we estimate missing values?

• How often should each node sample
the field?
• How can we reduce power
consumption without losing sensing
quality?
• How much should we compress the
data?
• Where should the processing be
performed?
• How can we localize nodes from RF
signals

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 39
University of Crete, Computer Science Department
 Reading list

• Swami, Ananthram, et al., eds. Wireless Sensor Networks: Signal Processing and
Communications. John Wiley & Sons, 2007.
• Zaslavsky, Arkady, Charith Perera, and Dimitrios Georgakopoulos. "Sensing as a service
and big data." arXiv preprint arXiv:1301.0159 (2013).
• Hackmann, Gregory, Octav Chipara, and Chenyang Lu. "Robust topology control for
indoor wireless sensor networks." Proceedings of the 6th ACM conference on
Embedded network sensor systems. ACM, 2008.
• Gubbi, Jayavardhana, et al. "Internet of Things (IoT): A vision, architectural elements,
and future directions." Future Generation Computer Systems 29.7 (2013): 1645-1660.
• Aggarwal, Charu C., Naveen Ashish, and Amit Sheth. "The internet of things: A survey
from the data-centric perspective." Managing and mining sensor data. Springer US,
2013. 383-428.
• Tsai, Chun-Wei, et al. "Data mining for internet of things: A survey.“ Communications
Surveys & Tutorials, IEEE 16.1 (2014): 77-97.

CS-541 Wireless Sensor Networks

Spring Semester 2016-2017 40
University of Crete, Computer Science Department