Documente Academic
Documente Profesional
Documente Cultură
I. I NTRODUCTION
Continuously monitoring seismic vibration events generated
by earthquakes, drilling activities, and moving heavy machinery or vehicles, and thoroughly analyzing the measurements
to detect potentially harmful vibrations for buildings, bridges,
dams, and pipelines are important to prevent disasters or environmental catastrophes. To this end, low-cost battery-powered
sensor devices are now ready to be used thanks to the advances
made in MEMS-based sensor technology and programmable
low-power signal-processing hardware components. In the
application scenario envisioned, sensor devices are positioned
at several geographic locations close to the foundation of the
building to measure the vibration acceleration. Each device
autonomously computes the maximum amplitude and dominant frequency of the vibration velocity at its location. If the
computed values exceed pre-defined threshold values, an alarm
is generated. To take the right action in case of an alarm and to
continuously trace parameter values of non-harmful vibrations,
the alarm signals as well as the vibration data retrieved are
forwarded from the sensors via a basestation (BS) to backend
applications for processing and control purposes. This can be
achieved by deploying an advanced low-power WVSN in the
field surrounding the building that comprises sensor devices,
additional relay nodes, and a gateway (GW) to connect the
nodes to backend applications that typically reside on servers
in an enterprise network.
Several challenging requirements have to be taken into
account when developing a wireless vibration sensor network:
37
38
Fig. 2.
39
V. N ETWORK A RCHITECTURE
40
Fig. 4.
Protocol architecture of WVSN solution. Clusters of WVSN are controlled by the network cluster control (NCC) and a global network control.
B. Synchronized operation
The NCC starts synchronizing the WVSN nodes by requesting the BS to execute a simple, but reliable time synchronization protocol that is derived from [13]. The BS, which
obtains a global reference time from a GPS receiver, acts as
a timing source for all nodes. It distributes the reference time
by broadcasting a synchronization beacon to its children. To
reduce the risk of losing beacons due to transmission errors,
it is transmitted several times. After receiving a beacon, the
nodes readjust their time reference by taking into account the
arrival time of the beacon and the processing delay between
the nodes. Each node then forwards its adjusted timing information to its children. The synchronization procedure is
iterated until all nodes in the network have been reached. After
having been synchronzied, the nodes turn their radio on and
off according to the TDMA schedule provided by the BS.
C. Medium access strategy and TDMA superframe structure
In synchronized mode, the access to the radio medium
and the timely execution of the various network tasks are
controlled according to the TDMA superframe structure. Each
superframe is divided into frames of different types and
41
Fig. 5.
Fig. 6.
corresponds to the one specified in [3], whereas the vibrationsensing application executed on the sensor nodes uses a
modified API that is optimized to work with the WVSN stack
and satisfies the low-power and timing requirements of the
application. This is reflected in Fig. 4 by adding a to the
MQTT-S client.
Before a MQTT-S client can communicate with the broker, it has to set up a session with the broker, inform the
broker about its identity and the topics it will use during
the session, and maintain the session by regularly sending a
keep-alive message to the broker. Instead of sending MQTT-S
messages through the network, we perform a cross-layer
optimization by making use of information already gathered
by the gateway during WVSN setup and configuration. The
message processor, which is part of the gateway, generates
the actual subscriptions as it already knows the identity and
the subscription topics of the sensor nodes after setting up the
underlying network. In addition, keep-alive messages are not
generated as long as the corresponding sensor nodes operate
according to the TDMA schedule.
When receiving messages destined for a certain sensor node
over the MQTT-S client, the message processor adds the
required source routing information so that the relay nodes
can route them to the correct receiver. It also applies a similar
routing function to the control and management messages
exchanged between the NCC and the wireless nodes.
D. Network recovery
Node failures and varying link qualities are common in lowpower wireless sensor networks and often lead to dynamic
changes of the network topology. A node operated in the
synchronized mode detects a synchronization loss after having
not received any synchronization beacon from its parent for a
certain time. In this case, the node switches to management
mode and continuously listens on the radio channel. If the
node again receives beacons from its parent with the original configuration ID, it immediately returns to synchronized
operation. If the node receives unknown beacons, it waits for
a listening frame and announces itself to the network. The
announcements are forwarded to the NCC, which wakes up
parts of the network in a subsequent superframe for performing
a local node discovery and link probing. With the information
provided by these functions, the NCC generates a new routing
tree and TDMA schedule, which are imposed on the WVSN by
first stopping its operation, reconfiguring the nodes, and then
restarting synchronized operation with the new configuration.
42
Fig. 7.
with 68% needed for the signal-processing units digital lowpass filter, integrator, and FFT processing as shown in Fig. 7.
As the NVRAM has a size of 8 Mb, it can store continuous
data packets for 6.6 h and exception data packets for 64 s.
The vibration-sensing application, the MQTT-S client, the
WVSN stack, and the operating system TinyOS v2.1 are
executed on the microcontroller, and require 6 kB of RAM
and 45 kB of flash memory. Note that all network nodes are
programmed with the same firmware and can detect their
network role based on the available hardware components.
B. Sensing module
The performance of the sensing module implementation was
evaluated in a certification test conducted at an independent
calibration service laboratory. The module and a reference
system were mounted onto a shaking table and excited along
the x-, y-, and z-axes by sinusoidal vibrations with frequencies
from 1 to 100 Hz and velocity amplitudes from 5 to 50 mm/s,
respectively. In addition, the table was excited by chirp modulated sinewaves with a frequency sweep from 1 to 10 Hz in
10 s and velocity amplitudes between 1 and 60 mm/s. The test
results confirm that the sensing module fullfills the DIN 4150-3
requirements for frequencies between 1 and 40 Hz and does
not introduce nonlinear signal distortions.
C. Time synchronization
The performance of the synchronization procedure was
evaluated in our laboratory by measuring synchronization
errors in a prototype WVSN consisting of a chain of six relay
nodes. The errors were determined by measuring the time
offsets between the start of the superframe as monitored on the
micrcontrollers of the network nodes and the BS. Fig. 8 gives
the mean and jitter of the synchronization error between the
global reference time provided by the BS and the local time
of the network nodes at several hop distances. The mean and
the variance of the error increase with the hop distance, but
remain within acceptable bounds. It can also be seen that the
fluctations of the errors remain smaller than 1.5 ms as long as
the nodes are fewer than 6 hops away from the BS. Moreover,
the results indicate that the duration of the TDMA slots can be
reduced to 20 ms without significantly degrading the network
performance.
43
100
99.9
Latency [ms]
99.8
99.7
99.6
99.5
99.4
6
Fig. 11. Latency and base station packet delivery ratio for different hop
distances. Note that the axis for PDR starts at 99.4 %
Fig. 9. Placement of nodes in the field tests. Squares represent sensor nodes,
circles are relay nodes. Only links are shown that have been established with
a success rate of > 95%.
TABLE I
VALUES CHOSEN FOR VARIOUS NETWORKING PARAMETERS
Parameters
Number of sensor nodes
Number of relay nodes
Distances between nodes
Transmit power
Number of hops
Number of acknowledgments
Number of link probe messages
Superframe duration
Slot duration
Number of slots for Synchronization frame
Number of slots for Collection frame
Number of slots for Listening frame
Number of slots for Exception frame
Values
9
29
30 - 300 m
2 mW
4-6
3
100
10 s
20 ms
16 - 20
80 - 115
2
40
Fig. 10.
D. Network lifetime
To determine the lifetime of the network, the power consumption of the sensor and relay nodes was computed based
on the TDMA schedule monitored in the field tests. According
to this schedule, both types of nodes are alternatingly operated
in the awake and sleep state, in which they consume power as
listed in Table II. Their average power consumption is thus a
function of their duty cycle, which is defined as the fraction
of time of the superframe during which a node is not in sleep
state divided by the duration of the superframe and which can
be calculated from the scheduling information.
Fig. 12 shows the calculated histograms of the duty cycle
of the nodes obtained for two operational scenarios and
44
TABLE II
C URRENT DRAWN BY THE NETWORK NODES IN DIFFERENT STATES AT
3.6 V SUPPLY VOLTAGE
State
sleep
awake
sensor node
[mA]
10.31
28.66
relay node
[mA]
0.012
19.52
Fig. 12.
in terms of packet delivery ratio and latency: The high endto-end PDR of above 99% and the low latency of less than
2.5 s indicate that data packets can be transferred with a high
reliability to the destination and that near real-time monitoring
of vibration events on a remote server is possible.
R EFERENCES
[1] Structural Vibration - Effects of Vibration on Structures, DIN 4150-3,
1999.
[2] ANSI/IEEE, IEEE LAN MAN Standards, Part 15.4: Wireless Medium
Access Control (MAC) and Physical Layer (PHY) Specifications for
Low-Rate Wireless Personal Area Networks (WPANs), 2006.
[3] U. Hunkeler, H. Truong, and A. Stanford-Clark, MQTT-S: A Publish/Subscribe Protocol for Wireless Sensor Networks, in Proc. Workshop on Information Assurance for Middleware Communications (IAMCOM), 2008.
[4] J. P. Lynch and K. J. Loh, A Summary Review of Wireless Sensors
and Sensor Networks for Structural Health Monitoring, The Shock and
Vibration Digest, vol. 38, no. 2, pp. 91128, 2006.
[5] S. Kim, S. Pakzad, D. Culler, J. Demmel, G. Fenves, S. Glaser, and
M. Turon, Health Monitoring of Civil Infrastructures using Wireless
Sensor Networks, Apr. 2007, pp. 254263.
[6] N. Xu, S. Rangwala, K. K. Chintalapudi, D. Ganesan, A. Broad,
R. Govindan, and D. Estrin, A Wireless Sensor Network for Structural Monitoring, in Proc. 2nd International Conference on Embedded
Networked Sensor Systems (SenSys), New York, USA, 2004, pp. 1324.
[7] G. Hackmann, W. Guo, G. Yan, C. Lu, and S. Dyke, Cyber-Physical
Codesign of Distributed Structural Health Monitoring with Wireless
Sensor Networks, in Proc. 1st ACM/IEEE International Conference on
Cyber-Physical Systems (ICCPS), 2010, p. 119.
[8] K. Chebrolu, B. Raman, N. Mishra, P. K. Valiveti, and R. Kumar,
BriMon: A Sensor Network System for Railway Bridge Monitoring,
in Proc. 6th International Conference on Mobile Systems, Applications,
and Services (MobiSys), 2008, p. 2.
[9] L. Liu and F. G. Yuan, Wireless Sensors with Dual-controller Architecture for Active Diagnosis in Structural Health Monitoring, Smart
Materials and Structures, vol. 17, no. 2, 2008.
[10] C. Kapoor, T. L. Graves-Abe, and J.-S. Pei, Development of a Smart
Wireless Sensing Unit using off-the-shelf FPGA Hardware and Programming Products, Smart Structures and Systems, vol. 3, no. 1, pp. 6988,
2007.
[11] D. D. Mascarenas, E. B. Flynn, M. D. Todd, T. G. Overly, K. M.
Farinholt, G. Park, and C. R. Farrar, Development of Capacitance-based
and Impedance-based Wireless Sensors and Sensor Nodes for Structural
Health Monitoring Applications, Journal of Sound and Vibration, vol.
329, no. 12, pp. 24102420, 2010.
[12] M. Picozzi, C. Milkereit, S. Parolai, K.-H. Jaeckel, I. Veit, J. Fischer,
and J. Zschau, GFZ Wireless Seismic Array (GFZ-WISE), a Wireless
Mesh Network of Seismic Sensors: New Perspectives for Seismic Noise
Array Investigations and Site Monitoring, IEEE Sensors, vol. 10, no. 4,
pp. 32803304, 2010.
[13] M. Maroti, B. Kusy, G. Simon, and A. Ledeczi, The Flooding Time
Synchronization Protocol, in Proc. 2nd International Conference on
Embedded Networked Sensor Systems (SenSys), 2004, pp. 3949.
45