CEILING SENSOR NETWORK FOR SOFT AUTHENTICATION
AND PERSON TRACKING USING EQUILIBRIUM LINE
Hidetoshi Nonaka, Shuai Tao, Jun Toyama and Mineichi Kudo
Graduate School of Information Science and Technology, Hokkaido University, Sapporo, Japan
Keywords: Soft authentication, Person tracking, Ceiling sensor network, Equilibrium line, Electromagnetic interference.
Abstract: In the previous stage of our research, we have developed a soft authentication system using a ceiling sensor
network. Our aim has been to exclude psychological and physical load caused by using strict biometrics,
video camera, and so on. We introduced a notion of distributed personality for authentication and tracking
of several persons. Through experimental results, we confirmed that the system could keep track of up to 5
persons. However, it has been found the performance is not enough for practical use, and then we have
reconstructed and improved the system. In this position paper, we present the design policy, overview of the
network system, and the obtained performance.
1 INTRODUCTION
Soft authentication is one of the important elements
of pervasive computing. Identifying and tracking of
the user are indispensable for appropriate context-
aware services. In general, there is a trade-off
between the accuracy (and security) and the comfort
in the user’s authentication. In the case of context-
aware services, the latter is more important,
considering the user’s privacy and psychological
load.
In our project, we have developed a soft
authentication and tracking system using ceiling
sensor network (Hosokawa, 2009). Our aim has been
to exclude psychological and physical load caused
by using strict biometrics, video camera, and so on.
Through experimental results, we confirmed that the
system could keep track of up to 5 users with a high
probability of correct identification. Now we are
examining the combination of various techniques,
such as individual ambient temperature (Kanda,
2009), and chairs with pressure sensor array
(Yamada, 2009). Complementary integration of
weak evidences is suitable for our purpose
(Koumoto, 2009).
In the previous ceiling sensor network
(Hosokawa, 2009), the sampling rate for 50 sensors
has been relatively slow for effective tracking and
practical use. We decided to reconstruct and
improve the system, especially the network protocol.
In this position paper, we present the design policy,
overview of the network system, and the obtained
performance.
2 SOFT AUTHENTICATION
Recently, various authentication systems have come
into practical use. The authentication factors are
classified into three types: the ownership factors, the
knowledge factors, and the inference factors. The
ownership ones include ID card, cell phone with RF-
ID, USB token, etc. The knowledge ones include
password, pass phrase, PIN number, birthday, etc.
The inference ones are mainly based on biometrics
such as fingerprint, DNA, retinal pattern, iris, face,
finger vein, speech, and so on. Such systems are
effective for maintaining a high level of security and
accuracy. On the other hand, for context-aware
services in daily life at home or at the office, such a
high level of security is not needed. The important
feature required for such purpose is that there is no
mental pressure, physical load, or invasions of
privacy.
We call the traditional authentication for security
“hard authentication,” and we call the authentication
for personalized services “soft authentication.” The
differences between them are summarized in Table 1.
218
Nonaka H., Tao S., Toyama J. and Kudo M..
CEILING SENSOR NETWORK FOR SOFT AUTHENTICATION AND PERSON TRACKING USING EQUILIBRIUM LINE.
DOI: 10.5220/0003398202180223
In Proceedings of the 1st International Conference on Pervasive and Embedded Computing and Communication Systems (PECCS-2011), pages
218-223
ISBN: 978-989-8425-48-5
Copyright
c
2011 SCITEPRESS (Science and Technology Publications, Lda.)
Table 1: Difference between “soft authentication” and “hard authentication” (Hosokawa, 2009).
Item Soft authentication Hard authentication
Goal Personalized service Security
Process/Device Unnoticeable Noticeable
Population size Small (2-20) Large (10-100,000)
Place to use Home or office, anywhere in a room Building or security-controlled room, fixed place
Level of accuracy Low (Very) High
Strength of evidence Weak (behavioural evidence, gait, etc.) Strong (iris, DNA, finger vein, etc.)
User’s cooperation Unnecessary Necessary
Psychological barrier Weak/None String
Sensors Infrared, pressure, etc. Camera, special devices
Necessary environment No special condition, day and night, movable obstacles Controlled condition, no obstacles
Establishment cost Low and flexible High and fixed
Stolen damage Low Crucial
In order to combine authentication and person
tracking, video camera has been used in many
studies. It is used for abnormality detection of single
residents (Aviv, 1997), (Sawai, 2004). Multiple
cameras and multiple omnidirectional cameras are
also utilized for authentication and person tracking
systems (Sogo, 2004), (Zhao, 2004), (Hightower,
2001), (Khan, 2001), (Yam, 2003). However, using
a camera causes psychological load and invasion of
privacy.
Pressure sensors have been used for the same
purpose (Murakita, 2004), or auxiliary data for other
methods (Ito, 2004), (Want, 1992), (Okuda, 2005).
However, in general, large number of sensors is
necessary for person tracking (more than 25 sensors
per 1 m
2
).
Pyroelectric infrared sensors have been also
adopted for soft authentication and person tracking.
In some of them, wireless ID badges are used for the
person identification (Want, 1992), (Schulz, 2003),
but some people would hesitate to wear such sensing
devices in a room. Others are the systems for single
person (Okuda, 2005). There is a series of studies
using pyroelectric infrared sensors and Fresnel lens
array (Hao, 2006), (Fang, 2006), (Fang, 2007),
(Shanker, 2006), but they are the tracking systems
for single person, and the resolution is not sufficient
to specify the person’s position in a room.
In the early stage of our study, we constructed a
ceiling sensor network using pyroelectric infrared
sensors (Hosokawa, 2005). We tried to use a
Bayesian network for prediction of the user’s current
position. However, we noticed that it was difficult to
embed into the algorithm several kinds of
knowledge about the difference in trajectories and
the difference in layout of obstacles. To learn the
probability of possible movements depending on the
locations, many training data are necessary.
Therefore we decided to use a simpler method
(Hosokawa, 2009).
We introduced a notion of distributed personality
(Hosokawa, 2009). In principle, two persons cannot
be distinguished when they meet at one place. When
this situation occurs, we divide their personalities.
For example, when persons A and B get together at
the same place, we replace them with two virtual
persons who have multiple personalities consisting
of a half of A and a half of B. By this division of
personality, the total number of persons is kept as
two. In this way, we have ambiguity but can keep
tracking them. The probability of each personality is
updated incrementally according to various weak
evidences, such as walking speed, trajectory of
movement, sitting down specific chair, and so on.
Through experimental results, we confirmed that
the system could keep track of up to five persons
with a high probability, however, precise
identification was difficult and more than six
persons could not be authenticated. We found the
main defect of our system was the sampling rate
from the sensor nodes, which was 2 Hz when all of
the 50 sensors are activated. Therefore, we decided
to reconstruct and improve the system, especially the
network protocol.
3 REQUIREMENTS
In our system, up to 128 pyroelectric infrared
sensors are attached on the ceiling. They are
interconnected by a sensor network. By gathering
the data from sensors, authentication and tracking of
users are achieved. The scope of our system extends
to not only home, office, laboratory, but also nursing
home, elder-care facility, assisted-living
condominium, welfare hospital, and child-care
CEILING SENSOR NETWORK FOR SOFT AUTHENTICATION AND PERSON TRACKING USING EQUILIBRIUM
LINE
219
institution. Consequently, following conditions are
required:
1. Low noise. Medical electronic instruments
such as artificial pacemakers might be used in
the room. Influence of EMI (electromagnetic
interference) should be avoided. Wireless
network is not preferable, and the wire should
be shielded-cable or at least balanced-line.
2. Simple and easy installation and wiring. The
positions of sensor units should be changed
along with the rearrangement of the furniture.
Light-weight wire is preferable.
According to the requirements mentioned above,
we adopted a wired network with balanced
equilibrium line (ANSI EIA-422).
4 SPECIFICATIONS
4.1 Sensor Network Topology
Figure 1 shows the physical topology of the sensor
network. There are 8 segments, and each segment
composes a bus network. In the figure, “T” stands
for the terminator, which can be set by jumper plug.
The bus line can be branched as is shown in the
figure (the lower left segment). The number of
sensor nodes is up to 16 for each segment, and then
the total number of sensor nodes is 128.
Figure 2 illustrates the logical topology of each
segment. Sensor nodes and the controller node are
interconnected with balanced equilibrium line (EIA-
422). The controller node plays the role of bus
master, because only one node is permitted to
transmit packets at each moment.
Figure 1: Physical topology of the sensor network.
Figure 2: Logical topology of the sensor network of each
segment.
4.2 Sensor Node
Passive infrared sensor is also called “pyroelectric
infrared sensor” or “infrared motion sensor”. It
detects change in the motion of a person or an object
with a different temperature from the ambient
temperature. It has been used for many applications,
such as automatic security lights, burglar alarms,
visitor acknowledgement, light switch control and
door opener.
We have used NaPiOn (AMN11111, Panasonic
Denko Co. Ltd.) as the sensor module. It includes 16
lenses for gathering infrared radiation to 4 quadrants
on the surface of the pyroelectric infrared detector.
Then, 64 detection zones are formed in front of the
sensor module. The detection area is up to 7.42 m ×
5.66 m on a plane at a distance of 2.5 m from the
sensor. In our system, the detection area of each
sensor was narrowed by a hand-made cylindrical
lens hood. Specifications of the sensor module are
shown in Table 2.
Figure 3 shows the circuit diagram of sensor
node attached on the ceiling.
Figure 3: Circuit diagram of sensor node.
Table 2: Detection performance of the sensor module.
Rated detection distance 5 m (max)
Horizontal detection range 100°
Vertical detection range 82°
Detection zone 64 zones
Movement speed 0.5 m/s (min) 1.5 m/s (max)
Object = human body H 700 mm × W 250 mm
Controller
USB
PC
T
T
T
T
T
TTT
T
T
MCU
Sensor
MCU
Sensor
MCU
Sensor
MCU
Sensor
MCU
USB
Controller
PC
Balanced Equilibrium Bus
Controller
A
B
Z
Y
REN
MAX489ECPD
TE
12
11
10
9
14
2
3
4
5
6,7
R
T
+12V
RS422-P
RS422-N
Vcc Vcc Vcc
1K
jumper for
terminator
A1
A0A3
A2
OSCI
OSCO
Vdd
B4
MCLR
B0
B1
B2
B3
Vss
22p
20MHz
22p
1
2
4
5
6
7
8
910
14
15
16
17
18
LED
Vcc
PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems
220
4.3 Controller Node
The controller consists of MCU (PIC16F873), USB
controller (FT232BM), and eight RS-422 (EIA-422)
transceivers (MAX489), which correspond to 8
segments. Some of these devices are slightly
obsolete, but obtainable by low prices at such as
junk shop. Figure 4 illustrates the circuit diagram of
the controller node. In the diagram, only the signal
lines are indicated.
Figure 4: Circuit diagram of controller node.
Figure 5: The format of packet framing.
Figure 6: Interconnection of sensor nodes with cables.
4.4 Cables for Equilibrium-lines
Figure 6 shows the physical connection of sensor
nodes. We adopted general ready-made phone cable
with modular connectors 6P4C (RJ14) without
shield. The reason is easiness of installation and
wiring. This type of cable is easily obtainable and
light enough to wiring only by connecting to the
plugs and leaving the cable drooping. It is expected
that EMI noise is suppressed by using equilibrium-
line, but further test is necessary.
4.5 Packet
The request from the controller node and the answer
from the sensor nodes are transmitted by packets.
The packet consists of only 8 bits (Figure 5). The
least significant 4 bits contain the address of the
sensor node, which is set using the dip switch
beforehand. If a node detects human motion, Bit 4 is
set when the node transmits the answer packet to the
controller node. Bit 5 toggles the flag of LED on the
sensor node, which blinks along with the human
motion. Bit 7 is usually unset. If time-out occurs, the
controller sets this bit internally.
4.6 Timing of Transmission
Data rate is suppressed to 250 kbps because of the
EMI matter (Figure 7). It takes 36 us to transfer each
packet with a start bit. In order to switch the tri-state
buffer of the equilibrium line, additional 8 us is
required. Consequently, time required to collect the
sensor data is within 90 us per node. If the number
of nodes is 128, the total time is about 12 ms, and
the sampling rate is up to 80 Hz.
Table 3: Command set of the controller.
Command Function
0x01
The controller is reset and all the bits of 8 mask registers
corresponding to the 8 segments are set. One byte is
returned only as acknowledge whose value is 0x00.
0x02
The controller is reset and all the bits of 8 mask registers
corresponding to the 8 segments are set. One byte is
returned only as acknowledge whose value is 0x00.
0x03
The controller sends the “LED ON” commands to all the
sensor nodes. One byte is returned only for acknowledge
whose value is 0x00.
0x04
The controller sends the “LED OFF” commands to all
the sensor nodes. One byte is returned only for
acknowledge whose value is 0x00.
0x05
The controller scans the sensor nodes without
acknowledge.
0x06
8 words (16 bytes) of mask registers are returned.
0x07
8 words (16 bytes) of sensor data are returned.
to PC via USB controller
16
24
26 252728
PIC16F873-20/SPRC5
RB4
RB0 RB1 RB2 RB3
RB7 RB5RB6
21 22 23
6P4C-Modular
+12V
6P4C-Modular
+12V
6P4C-Modular
+12V
6P4C-Modular
+12V
6P4C-Modular 6P4C-Modular
+12V
6P4C-Modular 6P4C-Modular
+12V +12V +12V
RX
TX
18
17
0: Detect: OFF
1: Detect: ON
0: LED: OFF
1: LED: ON
Address
01234567
Reserved
0: Node is active
1: Node is inactive or not exist
CEILING SENSOR NETWORK FOR SOFT AUTHENTICATION AND PERSON TRACKING USING EQUILIBRIUM
LINE
221
Figure 7: Waveforms of packet transmission.
4.7 Gathering the Sensor Data
All the data of 128 nodes in the network can be
gathered within 12 ms, but all the addresses are not
necessarily used. The number of nodes may be less
than 128, and there are possibilities that some nodes
get out of order or a node is detached while
operating. We adopted a time-out checking. The
time-out period is set to 50 us. When a time-out
occurs, the corresponding bit of the 16 bit mask
register for the specific segment is unset, and after
that, the request to the node is skipped. On the other
hand, each mask bit is set compulsorily once every
16 times. It allows the hot-plugging of sensor nodes
without shutting down the network. A newly
plugged node can be detected within 200 ms. Table
3 shows the command set of the controller node.
5 DISCUSSION
If electromagnetic interference can be disregarded,
the wireless sensor network seems to be applicable.
However, usually it is used for discrete data
acquisition, for example in several samples per
second with power saving mode, because of the
limitation of power supply. There have been many
researches, for example, (Bandyopadhyay, 2004),
(Liu, 2006), (Jafari, 2009), and so on. To our
purpose, much higher rate sampling of 80 Hz from
up to 128 nodes are necessary. It means
80×128=10240 samples per second, thus it is
difficult to make use of wireless network.
As is mentioned in the introduction, we are now
examining the integration of various kinds of weak
evidence of soft authentication and person tracking.
Wireless sensor network will be effectively used for
gathering and processing such heterogeneous sensor
data. Also, it will be suitable for tracking inter-room
or ambulant sensor nodes in nursing home, child-
care institution, and so on.
There have been following items to be studied
empirically in current solution of our system:
1. The performance of the authentication and the
tracking increase with the data rate.
2. EMI increases with the data rate.
3. Using shielded cable reduces EMI.
4. Using shielded cable diminishes the easiness
of wiring and installation.
It is necessary to examine above items
experimentally and quantitatively as the near future
work.
6 CONCLUSIONS
We have presented an improved ceiling sensor
network for soft authentication and person tracking.
Sampling rate of 80 Hz for up to 128 nodes using
250 kbps equilibrium line have been realized. The
total sampling rate is up to 10k samples per sec.
Now, the development and improvement of the soft
authentication system using our network system is in
progress. The experimental evaluation of the
performance of tracking persons is also needed.
REFERENCES
Hosokawa, T., Kudo, M., Nonaka, H., Toyama, J., 2009.
Soft Authentication Using an Infrared Ceiling Sensor
Network, Pattern Analysis and Applications, Vol. 12,
No. 3, pp. 237-250.
4us8us 4us
4us
8us4us32us
Port B
Tris B
PIN_C5
Unit R
Unit D
Unit T
Unit T
6us
6us
6us
0.4us
0.2us
2us
8us
4us
4us
0.4us
3us8us
4us
Unit R
Unit D
6us
6us 4us 6us 4us
6us
Controller
node
Target
sensor
node
Other
sensor
node
PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems
222
Kanda, M., Yoshikawa, T., Nonaka, H., 2009. Family
Communication Support Tool Using Time-Series
Visualization of Individual Ambient Temperature,
Proceedings of IEEE Industrial Symposium on
Consumer Electronics, Vol. 13, pp. 984-987.
Yamada, M., Kamiya, K., Kudo, M., Nonaka, H., Toyama,
J., 2009. Soft Authentication and Behavior Analysis
Using a Chair with Sensors Attached: Hipprint
Authentication, Pattern Analysis and Applications,
Vol. 12, No. 3, pp. 251-260.
Koumoto, Y., Nonaka, H., Yanagida, T., 2009. A Proposal
of Context-Aware Service Composition Method Based
on Analytic Hierarchy Process, Studies in
Computational Intelligence, Vol. 199, pp. 65-71,
Springer, 2009.
Aviv, D. G., 1997. Abnormality Detection and
Surveillance System, United States Patent 6028626.
Sawai, K., Yoshida, M., 2004. Algorithms for Detect of
Elderly People Abnormal State for the Monitoring of
Home, The Institute of Electronics, Information and
Communication Engineers, Vol. J87-D-II, No. 11, pp.
2054-2061 (in Japanese).
Sogo, T., Ishiguro, H., Trivedi, M., 2004. Real-time
Human Tracking System with Multiple
Omnidirectional Vision Sensors, Systems and
Computers in Japan, Vol. 25, pp. 79-90.
Zhao T., Nevatia, R., 2004. Tracking Multiple Humans in
Complex Situations, IEEE Transactions on Pattern
Analysis and Machine Intelligence. Vol. 26, pp. 1208-
1221.
Hightower, J., Borriella, G., 2001. Location Systems for
Ubiquitous Computing, IEEE Computer, Vol 34, pp.
57-66.
Khan, S., Javed, O., Rasheed, Z., Shah, M., 2001. Human
Tracking in Multiple Cameras, Proceedings of the 8th
International Conference on Computer Vision, pp.
331-336.
Yam, C., Nixon, M. S., Cater, J. N., 2003. Automated
Person Recognition by Walking and Running via
Model-based Approaches, Pattern Recognition, Vol.
37, pp. 1057-1072.
Murakita, T., Ikeda, T., Ishiguro, H., 2004. Human
Tracking Using Floor Sensors Based on the Markov
Chain Monte Carlo Method, Proceedings of
International Conference on Pattern Recognition, pp.
917-920.
Ito, T., Oguri, K., Matsuo, T., 2004. A Location
Information System Based on Real-time Probabilistic
Position Inference, Proceedings of International
Conference on Industrial, Engineering and Other
Applications of Applied Intelligent Systems, pp. 797-
806.
Want, R., Hopper, A., Falcao, V., Gibbons, J., 1992. The
Active Badge Location System, ACM Transactions on
Information Systems, Vol. 10, pp. 91-102.
Okuda, S, Kaneda, S, Haga, H., 2005. Human
Position/Height Detection Using Analog Type
Pyroelectric Sensors, Proceedings of EUC Workshops
2005, pp. 306-315.
Schulz, D., Fox, D., Hightower, J., 2003. People Tracking
with Anonymous and ID-Sensors Using Rao-
Blackwellised Particle Filters, Proceedings of
International Joint Conference on Artificial
Intelligence.
Hao, Q. et al., 2006. Human Tracking with Wireless
Distributed Pyroelectric Sensors, IEEE Sensors
Journal, Vol. 6, No. 6, pp. 1683-1696.
Fang, J. S. et al., 2006. Real-time Human Identification
Using a Pyroelectric Infrared Detector Array and
Hidden Markov Models, Optics Express, Vol. 14, pp.
6643-6658.
Fang J. S. et al., 2007. A Pyroelectric Infrared Biometric
System for Real-time Walker Recognition by Use of a
Maximum Likelihood Principal Components
Estimation (MLPCE) Method, Optics Express, Vol. 15,
pp. 3271-3284.
Shankar, M. et al., 2006. Human-tracking Systems Using
Pyroelectric Infrared Detectors,
Optical Engineering,
Vol. 45, No. 10, pp. 106401-106410.
Hosokawa, T., Kudo, M., 2005. Person Tracking with
Infrared Sensors, Knowledge-Based Intelligent
Information and Engineering Systems, Lecture Notes
in Computer Science, LNCS-3684, pp. 682-688.
Bandyopadhyay, S., Coyle, E. J., 2004. Spatio-Temporal
Sampling Rates and Energy Efficiency in Wireless
Sensor Networks, Proceedings of IEEE INFOCOM.
Liu, X., et al., 2006. Optimal real-time sampling rate
assignment for wireless sensor networks, ACM
Transactions on Sensor Networks, Vol. 2, No. 2, pp.
263-295.
Jafari, A. M., Lang, W., 2009. Optimal Sample Rate for
Wireless Sensor Actuator Network, IAENG
International Journal of Computer Science, Vol. 26,
No. 4.
CEILING SENSOR NETWORK FOR SOFT AUTHENTICATION AND PERSON TRACKING USING EQUILIBRIUM
LINE
223