Synchronized Sensing and Network Scalability of
Low-Cost Wireless Sensor Networks for Monitoring
Civil Infrastructures. 

Scene 1

Synchronized Sensing and 
Network Scalability of
Low-Cost Wireless Sensor Networks for Monitoring
Civil Infrastructures

Supervised by:	Dr. C.S.Lewangamage.                          	Prof. M.T.R.Jayasinghe.     Presented by:	Mr. P.Vishnu. 		Mr. S.Radershan

Introduction to Structural health monitoring and Wireless sensor networks. Objectives. Introduction to the developed low cost Wireless sensor node. Introduction to synchronization. Synchronized sensing method. Application, verification and reliability of the method. Cost aspects of the developed wireless system Conclusion. References.. 

Scene 2

Introduction to Structural health monitoring and Wireless sensor networks. Objectives. Introduction to the developed low cost Wireless sensor node. Introduction to synchronization. Synchronized sensing method. Application, verification and reliability of the method. Cost aspects of the developed wireless system Conclusion. References.

Structural Health Monitoring. The process of implementing a damage detection / health threat and characterization strategy for engineering structures / human is referred to as Structural Health Monitoring (SHM).. 

Scene 3

The process of implementing a damage detection / health threat and characterization strategy for engineering structures / human  is referred to as Structural Health Monitoring (SHM).

Electrocardiogram (ECG)
ECGlead
Holter monitor

Structural Health Monitoring. Cra A. Bridges. Aircraft. 

Scene 4

Importance of SHM. To understand better how civil infrastructures really perform which differs from the calculated or designed behaviour due to, Over-estimated or under-estimated loads. Inability of codifying the exact wind and seismic load for unusual site and geometric conditions. Inability of predicting site conditions such as soil-structure interaction and impact due to close by structures. A structure has should be used as a living laboratory and the realistic behaviors should be captured and used for future designs.. 

Scene 5

To understand better how civil infrastructures really perform which differs from the calculated or designed behaviour due to, Over-estimated or under-estimated loads. Inability of codifying the exact wind and seismic load for unusual site and geometric conditions. Inability of predicting  site conditions such as soil-structure  interaction  and impact due to close by  structures. A structure has should be used as a living laboratory and the realistic behaviors should be captured and used for future designs.

Metric Wired Sensor Networks Wireless Sensor networks(WSN) Number of Sensors Less because of the difficulty in installing Higher due to ease of installation Cost Very high $10,000 to $25000 Low it would only cost around $600 per node Deployment Time Very long and it would take several days Short it would cost only few hours Life span Long, Typically limited by hardware life span Short, Typically limited by node battery life time Connection bandwidth Higher band width due to wired connection Limited band width and unreliable connection Data rate High sensor data rates Lower sensor data rates but higher than conventional Sensor Synchronization Very high due to wired connection Concern due to wireless connections. 

Scene 6

Metric Wired Sensor Networks Wireless Sensor networks(WSN) Number of Sensors  Less because of the difficulty in installing Higher due to ease of installation Cost Very high  $10,000 to $25000 Low it would only cost around $600 per node Deployment Time Very long and it would take several days Short it would cost only few hours  Life span Long, Typically limited by hardware life span Short, Typically limited by node battery life time Connection bandwidth  Higher band width due to wired connection Limited band width and unreliable connection  Data rate  High sensor data rates Lower sensor data rates but higher than conventional Sensor Synchronization Very high due to wired connection Concern due to wireless connections

Tethered vs Wireless Sensor Networks in SHM

Objectives. To develop a low-cost wireless sensor node capable of collecting precise data with high frequency with low cost involved. To incorporating Synchronized sensing technology and synchronize the nodes with acceptable jitter. Establish the developed Wireless sensor network on a target building and verify the performance of WSN using MAC.. 

Scene 7

To develop a low-cost wireless sensor node capable of collecting precise data with high frequency with low cost involved.
 To incorporating Synchronized sensing technology and synchronize the nodes with acceptable jitter.

 Establish the developed Wireless sensor network on a target building and verify the performance of WSN using MAC.

Developed Sensor Node. A close up of a device Description automatically generated. 

Scene 8

A close up of a device

Description automatically generated

CC1310 Ultra-Low-Power wireless MCU Sub-1 GHz:  	Processor:	48MHz Arm Cortex M3.  	Flash:	 	32KB, 64KB and 128KB. 	SRAM:		8KB. Raspberry Pi 3+:  	Processor:	Quad Core 1.2GHz ARM Cortex-A53.  	RAM: 		1GB LPDDR2 (900 MHz). 	Networking:	10/100 Ethernet, 2.4GHz 802.11n wireless 	Storage: 		microSD

Analog to digital conversion Buffering the data Conversion of data into suitable format Establishing the network connection Synchronization Sending the data

Functioning of Sensor Node. The master node at the proximal center of the network will synchronize all the other nodes with respect to its clock, which is done by CC1310 Launchpad which uses Sub 1 GigaHertz (800m range in open air) After all the nodes have been synchronized, CC1310 sends a signal to the Raspberry Pi through its General Purpose Input Output pins (GPIO), which will trigger the Raspberry Pi to start collecting data Using Inter-Integrated Circuit (I2C) protocol, MMA8451 collects the data and sends it to Raspberry Pi which will be stored in a 32GB microSD card. After 40 minutes worth of data is collected, the node will start this process again.. 

Scene 9

The master node at the proximal center of the network will synchronize all the other nodes with respect to its clock, which is done by CC1310 Launchpad which uses Sub 1 GigaHertz (800m range in open air)   After all the nodes have been synchronized, CC1310 sends a signal to the Raspberry Pi through its General Purpose Input Output pins (GPIO), which will trigger the Raspberry Pi to start collecting data
 Using Inter-Integrated Circuit (I2C) protocol, MMA8451 collects the data and sends it to Raspberry Pi which will be stored in a 32GB microSD card.
  After 40 minutes worth of data is collected, the node will start this process again.

Synchronization. A clock on each of it s sides Description automatically generated. 

Scene 10

A clock on each of it s sides

Description automatically generated

A clock hanging from the side

Description automatically generated

Scene 11

Synchronization. Synchronization error (Jitter) needs to be less than 120 μs for precise phase angle calculations. The primary reason for synchronization is that each collected data by a particular node has to be matched with the data collected at the same time by another node of interest. If the jitter exceeds beyond 120 μs, the calculation of phase angle calculations will be erroneous thus affecting the following. damage detection Localization error. 

Scene 12

Synchronization error (Jitter) needs to be less than 120 μs for precise phase angle calculations.              The primary reason for synchronization is that each collected data by a particular node has to be matched with the data collected at the same time by another node of interest. If the jitter exceeds beyond 120 μs, the calculation of phase angle calculations will be erroneous thus affecting the following. damage detection Localization error

Peak Matching Method. A close up of a piece of paper Description automatically generated. 

Scene 13

A close up of a piece of paper

Description automatically generated

A screenshot of a cell phone

Description automatically generated

Peak matching method consists of synchronization by matching the peaks of the acceleration time graph drawn by two accelerometers from two nodes.

Acceleration vs time graph drawn for one of the sensor nodes and a calibrated accelerometer similar to the accelerometer within the sensor node. Source: [1]

State of the art synchronization protocols that is available at present for WSNs which has achieved jitter at the range of 0.5 µs. Due to having hardware specific requirements in mote devices such as having an SFD pin and separate radio and CPU cores for the implementation, the cost of a single node increases, which increases the cost of the whole network drastically.. 

Scene 14

State of the art synchronization protocols that is available at present for WSNs which has achieved jitter at the range of 0.5 µs. Due to having hardware specific requirements in mote devices such as having an SFD pin and separate radio and CPU cores for the implementation, the cost of a single node increases, which increases the cost of the whole network drastically.

Fixed timeslots of Glossy protocol . Source: [2]

Existing Methods Used for Synchronization

Synchronization implemented in the current system. 

Scene 15

Synchronization implemented in the current system

The chronological order of events of the synchronization in the developed system. 

Scene 16

The chronological order of events of the synchronization in the developed system

aster
Node
ower On
Timestamp 1
Next increment
time - Delay
Synchronize
other nodes
Olle
Data
ave
Node
ower On
Counter = nl
Timestamp 3
E.
Counter =
Wait until nx
Counter
..........I Counter = nx
Ollec
Data

Node. Master. Timestamp 1. Timestamp 2. Process. Timestamp 3. 

Scene 17

Sync Message Timestamp 2, 3 Schedule Message

Repeats the same  process for  n number of nodes

Identification of Jitter at the Laboratory Level. 

Scene 18

Identification of Jitter at the Laboratory Level

Session
1
2
3
4
5
7
8
9
10
11
12
13
1
15
16
17
18
19
20
21
22
23
24
25
Mean
50.8
61.7
56.9
58.2
43.9
40.15
38.4
38.5
40.5
42.25
38.75
.9
33.45
41.2
32.55
35.65
38.1
35.95
40.1
43.4
35.85
37.25
36.65
Std deviation
13.83
16.61
23.98
26.59
22.48
19.1
21.24
15.81
18.04
17.64
16.85
19.72
19.36
1.
18.38
17.24
16.4
17.34
17.92
17.23
18.92
17.8
19.49
15.68
18.8

A total of 25 synchronization sessions were conducted using an oscilloscope where synchronization was achieved to the level of 10 µs with maximum average of 56.9 µs with maximum standard deviation 26.59 µs.(Max-83.49 µs)

Scaling of the WSN. Hop Hop Hop. Hop Hop Star Topology. 

Scene 19

Laboratory level test showed that synchronization jitter is NOT increasing linearly when the WSN was expanded to multi hop system. This is an encouraging results which leads to scaling of developed WSN to a multi hop system without compromising the synchronisation. Further experiments are under progress at insitu level to check the pragmatic way of establishment.. 

Scene 20

Laboratory level test showed that synchronization jitter is NOT increasing linearly when the WSN was expanded to multi hop system.
 This is an encouraging results which leads to scaling of developed WSN to a multi hop system without compromising the synchronisation.
 Further experiments are under progress at insitu level to check the pragmatic way of establishment.

Application on Target Building. øøø øøø gur. 

Scene 21

Scene 22

Mode 1: T = 5.893s f = 0.17Hz. Mode 4: T = 1.935s f = 0.517Hz. 

Scene 23

Mode 5: T=1.797s f=0.557 Hz. Mode Natural frequency (Hz) Mode shape 1 0.17 Translational 2 0.178 Translational 3 0.335 Torsional 4 0.517 Torsional 5 0.557 Torsional. 

Scene 24

Mode Natural frequency (Hz) Mode shape 1 0.17 Translational 2 0.178 Translational 3 0.335 Torsional 4 0.517 Torsional 5 0.557 Torsional

Establishment of WSN at various phases in the target building. 

Scene 25

Establishment of WSN at various phases in the target building

Phase 1. 3. 5. Level 47. Level 48. i!i'iiiiiiliiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiliiiiiil . l; l; 1.11.11. 

Scene 26

i!i'iiiiiiliiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiliiiiiil
. l; l; 1.11.11

Phase 6. 3. Level 36. Level 38. 1. li- i? i? iiiili:i iä. 

Scene 27

lil E iG
lii i? iiiiiliili- iT iä i? iiii.'lii i -

Phase 11. Level 24. 1. Level 23. 3. li l: II: II:II: j:. 

Scene 28

11
ii iiiii iiiii iii'
iiGiiii ,iiiii iii
lii¯

Mode shapes –Using Response time history. 3 3 3 s 5 s 7 7 9 9 9 9 17 19 21 23 2s 27 29 31 3s 37 39 a: 43 as 47 49 19 23 25 29 37 39 41 43 45 47 49 11 17 19 21 23 2S 27 29 33 39 41 as as 49 15 17 21 23 25 27 33 3 17 19 21 23 27 29 31 33 •s 3 41 43 as 47. 

Scene 29

Mode shapes –Using Response time history

3
3
3
s
5
s
7
7
9
9
9
9
17
19 21 23 2s 27 29 31 3s 37 39 a: 43 as 47 49
19 23 25 29 37 39 41 43 45 47 49
11 17 19 21 23 2S 27 29 33 39 41 as as 49
15 17 21 23 25 27 33 3
17 19 21 23 27 29 31 33 •s 3
41 43 as 47

Modal Assurance Criteria(MAC). Modal assurance criteria (MAC) takes value from 0 to 1, where 0 represents no consistent correspondence and 1 represents a good correspondence. Values larger than 0.9 indicates a satisfactory level of correspondence. All five modes considered were compared using this method and it could be observed that the modal assurance values were closed to 0.9.. 

Scene 30

Modal assurance criteria (MAC) takes value from 0 to 1, where 0 represents no consistent correspondence and 1 represents a good correspondence.  Values larger than 0.9 indicates a satisfactory level of correspondence.  All five modes considered were compared using this method and it could be observed that the modal assurance values were closed to 0.9.

A picture containing bird

Description automatically generated

A close up of a logo

Description automatically generated

Cost of the Node. Component of the node Cost of a unit (in USD) Quantity Total cost (in USD) Raspberry pi 3B+ 35.00 1 35.00 CC1310 Launchpad 29.00 1 29.00 MMA8451 Accelerometer 7.95 1 7.95 81-MXFR01JA1000 cable 21.83 1 21.83 712-ANT-868-CWHWRRPS 11.17 1 11.17 32GB microSD card 9.34 1 9.34. 

Scene 31

Component of the node  Cost of a unit (in USD)  Quantity  Total cost (in USD) Raspberry pi 3B+ 35.00 1 35.00 CC1310 Launchpad 29.00 1 29.00 MMA8451 Accelerometer 7.95 1 7.95 81-MXFR01JA1000 cable 21.83 1 21.83 712-ANT-868-CWHWRRPS 11.17  1 11.17  32GB microSD card  9.34 1 9.34

Cost Comparison. Project Cost of a Sensor node (in USD) Node of this system 114.29 Straser and Kiremidjian et al. (1998) [16] 598.50 Bennett et al. (1999) [17] 590.61 Lynch et al. (2001, 2002b) [18] [19] [20] 389.06. 

Scene 32

Project Cost of a Sensor node (in USD) Node of this system 114.29 Straser and Kiremidjian et al. (1998) [16] 598.50 Bennett et al. (1999) [17] 590.61 Lynch et al. (2001, 2002b) [18] [19] [20]  389.06

The sensor node developed is 70.62% cheaper than the cheapest of the compared nodes.

Conclusion. A Low cost WSN was successfully established in target building with jitter within allowable limit. Using Synchronised sensing method the maximum jitter was kept below 90 µs. The mode shapes calculated using time history response captured using developed low cost WSN resulted in MAC values close to 1. The cost of the developed node yielded 70 % of cost saving compared to the sensor nodes currently used.. 

Scene 33

A Low cost WSN was successfully established in target building with jitter within allowable limit. Using Synchronised sensing method the maximum jitter was kept below 90 µs. The mode shapes calculated using time history response  captured using developed low cost WSN resulted in MAC values close to 1.  The cost of the developed node yielded 70 % of cost saving compared to the sensor nodes currently used.

References. S. Kim, G. Fenves, and S. Glaser, “Health monitoring of civil infras-tructures using wireless sensor networks,” in Proceedings of the 6thInternational Conference on Information Processing in Sensor Networks,2007, pp. 254–263. J. P. Lynch, “A summary review of wireless Sensors and SensorNetworks for Structural Health Monitoring,” Shock Vib. Dig., vol. 38,no. 2, pp. 91–128, 2006. A. B. Noel, A. Abdaoui, T. Elfouly, M. H. Ahmed, A. Badawy, and M. S.Shehata, “Structural health monitoring using Wireless sensor networks:A Comprehensive Survey,” IEEE Commun. Surv. Tutorials, vol. 19, no.3, pp. 1403–1423, 2017. E. Sazonov, V. Krishnamurthy, and R. Schilling, “Wireless intelligentsensor and actuator network - A scalable platform for time-synchronousapplications of structural health monitoring,” Struct. Heal. Monit., vol.9, no. 5, pp. 465–476, 2010.. 

Scene 34

S. Kim, G. Fenves, and S. Glaser, “Health monitoring of civil infras-tructures using wireless sensor networks,” in Proceedings of the 6thInternational Conference on Information Processing in Sensor Networks,2007, pp. 254–263.
 J. P. Lynch, “A summary review of wireless Sensors and SensorNetworks for Structural Health Monitoring,” Shock Vib. Dig., vol. 38,no. 2, pp. 91–128, 2006.
  A. B. Noel, A. Abdaoui, T. Elfouly, M. H. Ahmed, A. Badawy, and M. S.Shehata, “Structural health monitoring using Wireless sensor networks:A Comprehensive Survey,” IEEE Commun. Surv. Tutorials, vol. 19, no.3, pp. 1403–1423, 2017.
  E. Sazonov, V. Krishnamurthy, and R. Schilling, “Wireless intelligentsensor and actuator network - A scalable platform for time-synchronousapplications of structural health monitoring,” Struct. Heal. Monit., vol.9, no. 5, pp. 465–476, 2010.

References. S. Beskhyroun and Q. Ma, “Low-cost accelerometers for experimentalmodal analysis,” in World Conference on Earthquake Engineering,Lisbon, Portugal, 2012, pp. 1–10. S. Jeong, R. Hou, J. P. Lynch, H. Sohn, and K. H. Law, “An informationmodeling framework for bridge monitoring,” Adv. Eng. Softw., vol. 114,pp. 11–31, 2017, doi: 10.1016/j.advengsoft.2017.05.009. F. Ferrari, M. Zimmerling, L. Thiele, and O. Saukh, “Efficient networkflooding and time synchronization with Glossy,” Proc. 10th Int. Conf.Inf. Process. Sens. Networks - IPSN ’11,pp. 73–84,May.2011. M. L. Sichitiu and C. Veerarittiphan, “Simple, accurate time synchro-nization for wireless sensor networks,” IEEE Wirel. Commun. Netw.Conf. WCNC, vol. 2, no. C, pp. 1266–1273, 2003.. 

Scene 35

S. Beskhyroun and Q. Ma, “Low-cost accelerometers for experimentalmodal analysis,” in World Conference on Earthquake Engineering,Lisbon, Portugal, 2012, pp. 1–10.
 S. Jeong, R. Hou, J. P. Lynch, H. Sohn, and K. H. Law, “An informationmodeling framework for bridge monitoring,” Adv. Eng. Softw., vol. 114,pp. 11–31, 2017, doi: 10.1016/j.advengsoft.2017.05.009.
 F. Ferrari, M. Zimmerling, L. Thiele, and O. Saukh, “Efficient networkflooding and time synchronization with Glossy,” Proc. 10th Int. Conf.Inf. Process. Sens. Networks - IPSN ’11,pp. 73–84,May.2011.
 M. L. Sichitiu and C. Veerarittiphan, “Simple, accurate time synchro-nization for wireless sensor networks,” IEEE Wirel. Commun. Netw.Conf. WCNC, vol. 2, no. C, pp. 1266–1273, 2003.

Synchronized Sensing and Network Scalability of Low-Cost Wireless Sensor Networks for Monitoring Civil Infrastructures