Design and implementation of a home embedded surveillance system with ultra-low alert power

Abstract

In this paper we design and implement a home embedded surveillance system with ultra-low alert power. Traditional surveillance systems suffer from an unnecessary waste of power and the shortcomings of memory conditions in the absence of invasion. In this design we use Pyroelectric Infrared sensors (PIR) and pressure sensors as the alert group in windows and doors where an intruder must pass through. These low-power alert sensors wake up the MCU (Micro Controller Unit) which has power management for the ultrasonic sensors and PIR sensors indoors. This state transition method saves a large number of sensors required for the alert power. We also use the Majority Voting Mechanism (MVM) to manage the sensor groups to enhance the probability of multiple sensors sensing. After the MCU sends the sensor signals to the embedded system, the program starts the Web camera. Our sensing experiment shows that we reduce the system's power consumption.

Full text

Page 1

Y.-W. Bai et al.: Design and Implementation of a Home Embedded Surveillance System with Ultra-Low Alert Power 153

Manuscript received 12/26/10
Current version published 02/21/11
Electronic version published 02/21/11. 0098 3063/11/$20.00 © 2011 IEEE

Design and Implementation of a Home Embedded
Surveillance System with Ultra-Low Alert Power
Ying-Wen Bai, Zi-Li Xie and Zong-Han Li

Abstract — In this paper we design and implement a home
embedded surveillance system with ultra-low alert power.
Traditional surveillance systems suffer from an unnecessary
waste of power and the shortcomings of memory conditions in
the absence of invasion. In this design we use Pyroelectric
Infrared sensors (PIR) and pressure sensors as the alert
group in windows and doors where an intruder must pass
through. These low-power alert sensors wake up the MCU
(Micro Controller Unit) which has power management for the
ultrasonic sensors and PIR sensors indoors. This state
transition method saves a large number of sensors required
for the alert power. We also use the Majority Voting
Mechanism (MVM) to manage the sensor groups to enhance
the probability of multiple sensors sensing. After the MCU
sends the sensor signals to the embedded system, the program
starts the Web camera. Our sensing experiment shows that we
reduce the system’s power consumption1.

Index Terms — Embedded Surveillance System, Majority
Voting Mechanism, PIR Sensor, Ultrasonic Sensor, Low-Power
State.
I. INTRODUCTION
The traditional surveillance systems take a long time to
detect whether there is any intruder. If there is no intruder,
the sensing device which continuous to work and consumes
much power [1-5]. To meet the increased requirements of
the IEA we have to reduce the standby power of each
electrical apparatus to less than 1 Watt [6-8]. A recently
published survey shows that various attempts have been
made to reduce such power loss by to making the adapters
more efficient [9-12]. Another way to improve power
efficiency is accurate control of the apparatus by both
software and microcontroller [13-15].
In this paper the alerting sensors with low-power
consumption are placed near those home windows and doors
where an intruder must pass through. When an intruder
enters the sensing area, the sensors wake up the sleeping
MCU (Micro Controller Unit) which starts the power supply
for the indoor sensors and for the sensor signal transmission
to the embedded system. For the indoor sensors we use the

1 Ying-Wen Bai is with the Department of Electrical Engineering and
Graduate Institute of Applied Science and Engineering, Fu-Jen Catholic
University, Taipei, Taiwan, 242, R.O.C. (e-mail: bai@ee.fju.edu.tw).
Zi-Li Xie is a graduate student of Fu-Jen Catholic University, Taipei,
Taiwan, 242, R.O.C. (e-mail: 498506022@mail.fju.edu.tw).
Zong-Han Li is a graduate student of Fu-Jen Catholic University, Taipei,
Taiwan, 242, R.O.C. (e-mail: 497506207@mail.fju.edu.tw).
MVM to improve the sensing reliability [16-17]. The
embedded surveillance system determines the sensor results
and then decides whether to start the Web camera to both
capture images and upload these captured images to the Web
page through the Internet [18-20]. We use the MCU’s sleep
mode to reduce the alert power consumption for our home
embedded surveillance system when there is no intruder so
as to improve the traditional surveillance system without
wasting the power.
II. SYSTEM ARCHITECTURE
Fig. 1 shows the home embedded surveillance system
which has two groups of sensors, indoor and outdoor. The
outdoor sensor group contains a number of PIR and pressure
sensors placed near windows and doors of a home. When the
outdoor sensors sense an intruder, the MCU is woken up and
turns on the power for the indoor PIR and ultrasonic sensors
for the Majority Voting Mechanism. When this is completed,
the decision signal passes to the embedded board GPIO
(General purpose input and output). The software module of
the embedded board turns on the Web camera to capture
images, and the embedded system uploads them by means of
the Web server through the Internet. The user can view the
images captured by the home surveillance system through the
Internet. Fig. 2 shows the system architecture.



Fig. 1. The state transition for the home embedded surveillance system to
save power.

Page 2

154 IEEE Transactions on Consumer Electronics, Vol. 57, No. 1, February 2011


Fig. 2. The home embedded surveillance system with ultra-low alert
power.
A. Software Modules
Our design shown in Fig. 3 consists of the internal MCU
software module and the home embedded system software
module. In the MCU software module, when an intruder has
been detected, the MCU wakes up and decide whether to test
the threshold, and then turn on the power supply for the
indoor sensors. If the indoor sensors detect no intruder when
the outdoor sensors are misjudging, the MCU turns off the
power for the indoor sensors and goes back to the alert state.

Fig. 3. The software flowchart of the MCU.
We use Embedded Linux in the embedded board software.
The software flowchart shown in Fig. 4 scans the embedded
board GPIO pin, and when any input pin is high, this
represents a sensor sensing an intruder. The embedded system
turns on the Web camera to capture images.

Fig. 4. The software flowchart of the embedded board.
B. Hardware Modules
To reduce the power consumption of the alert state we
utilize both pressure sensors and PIR sensors. We use the PIR
sensors to detect the change in environment temperature
through human temperature in our experiment. The sensing
distance of a PIR sensor can reach 4m, and the sensing angle
is 110°. We use a common PIR sensing IC. The standby
power is only 1.05mW if there is no intruder. Fig. 5 shows a
system block diagram of the PIR module.


Fig. 5. System block diagram of the PIR module.

We place the PIR sensor groups on the ceiling, and the
sensing area is a cone-shaped projection area. Covering the

Page 3

Y.-W. Bai et al.: Design and Implementation of a Home Embedded Surveillance System with Ultra-Low Alert Power 155
sensing area with multiple PIR sensor groups extends the
sensing area, and by using MVM we enhance the overall
sensing probability. But the PIR sensor cannot sense a low-
speed or heat-insulated object, because obviously these
won’t change the environment temperature. The pressure
sensors are designed by linear potentiometers. The pressure
sensors used are thin and placed on the ground. When an
intruder invades, the pressure on the linear potentiometer
increases the resistance value, and the MCU determines
according to the resistance value whether someone is
passing through. Fig. 6 shows the system block diagram of
the pressure sensor module.


Fig. 6. System block diagram of the pressure sensor module.

For the detection state we use PIR sensors and multi-
frequency ultrasonic sensors. Fig. 7 shows the ultrasonic
sensor module which uses a typical oscillator chip to design
a square waveform generator and to adjust the resistance and
capacitance so as to generate a multi-frequency ultrasonic
transmission. The ultrasonic transducer transforms the
voltage waveform into an ultrasonic transmission and the
transducer of the receiver transforms the ultrasonic
transmission into the voltage waveform. Since the receiver
may experience external interference at different frequencies,
it is necessary to screen both the filter signals outside the
receiving frequency and the signal input to the amplifier and
the comparator; as other ultrasonic sensors are also
susceptible to refractive interference, so we use several
ultrasonic sensors at the receiving end. The total number of
ultrasonic sensors as the majority of the sensors triggered is
then sent to the MCU.



Fig. 7. A system block diagram of the ultrasonic sensor module.

Table I shows the sensing characteristics of the PIR sensor,
the ultrasonic sensor and the pressure sensor.
TABLE I
COMPARISON OF THE CHARACTERISTICS OF PIR SENSOR, ULTRASONIC
SENSOR AND PRESSURE SENSOR
Influence of
environment
temperature
PIR
sensor
changed
Ultrasonic
sensor
Pressure
sensor
change

Sensor
Condition
for trigger
Sensing
type
Disadvantages
Temperature
Dependence
Projection
area
Line
direction
Line
direction
Slow speed and
heat insulation
Noise
interference
Moving block Independence
Pressure
Independence
Placement
III. MAJORITY VOTING MECHANISM
We first discuss the definition and the probability equation
of a majority voting mechanism. According to our MVM the
resolution count must be greater than
number of sensors. To fit the extreme value of n we use
to deduce the relation between
n

5 . 0
, n being the total
nw
in
mmultiple
PPn

)(
and
ssingle
PP

the extreme value of n [8].


r












nw
r
s
s
nw
s
nw
sm
P
P
nwr
n
PPP
)1 (
0
)1 (
)])
1
([1 ()(
(1)
We define


}])1[(, ] 1

)1[(, , 2 , 1 , 0{,)
1
()(nwnwkand
P
P

nwk
n
kf
k
s
s








As we expect that


k

nw
kf
)1 (
0
)(
will converge, we need to
determine whether
ratio test for the convergence function we learn that increasing
the n value gradually decreases the ratio for each
relationship is as follows:
1 ( [
)2( ) 1 ()0(

ffff
Let

n
, so
1
(
)0(

n
wf
We let
12/1
 w
and
wPs

kfPPP
r
If we let
m
that
mm
PP
1
, and
m
We rewrite (1) by letting
wPs
by the ratio test.
) 1 (
)1 (


r
According to (2) and (3) the sensing probability of multiple
sensors must be greater than the sensing probability of any
single sensor. We know that when
)(kf
is a decreasing function. From the
)( kf
. The
] 1

)1 ( [
])

1 ( [
]2)1 ( [
] 1

)) 3()2() 1 (
1


w




nf
nwf
nw
nwffff


1)
1
()
)1(





s
s
P
P
wf
lim


1
)(
)()()1 (

)(
)1 (

0
)1 (




n
lim



wP
P
s
m
nw
nw
s
nw
sm
(2)
P represent the miss rate of sensors, we can learn
P =0 in this case.
and thus deduce
mm
PP
1

1
)(
)()()1 ()(
0



n
lim



wP
PkgPPP
s
m
nw
nw
s
nw
sm
(3)
single
P
is greater than 0.5, the
)(nPmultiple
improvement of the sensing probability of multiple sensors
through majority voting. We obtain Table II by given
will be greater than 0.5. Fig. 8 shows the
single
P
.
If the sensing probability of a single sensor is 0.7, the sensing
probability of seven sensors will be 0.874.

Page 4

156 IEEE Transactions on Consumer Electronics, Vol. 57, No. 1, February 2011
020406080100
0
10
20
30
40
50
60
70
80
90
100
Single Individual Module Recognition Probability ( % )
Combined System Overall Recognition Probability ( % )
n = 1
n = 3
n = 5
n = 7
n = 9
n = 39
n = 99
n = 159
n near to infinity
Combined Probability in Any
Odd Number Modules Will
Always be Equal to 50%
-Multiple(7)=87.4%
-Multiple(5)=83.7%
-Multiple(3)=78.7%
-Multiple(1)=70.0%
-Multiple(9)=90.1%
-Multiple(1)=30.0%
-Multiple(3)=21.6%
-Multiple(5)=16.3%
-Multiple(7)=12.6%
-Multiple(9)=9.88%

Fig. 8. Sensing probability of a single sensor and multiple sensors.

TABLE II
RELATIONSHIP BETWEEN
single
P
AND
)(nPmultiple


IV. MAJORITY VOTING MECHANISM AND
COMPLEMENTS
In this Section we continue to deduce the majority voting
mechanism and look at the complementing of multiple groups
of sensors. We choose three sensors in a sensor group because
three is the smallest number possible for the majority voting
mechanism, and we use three as an example to show the
process. If we assume that the sensing probabilities of each
single sensor are
PPsingle
and the miss rates are
probability of three sensors which are all in favor is
PPPP

, and the probability of two sensors in favor and
one sensor not in favor is
PP

probability of a majority vote in favor is the sum of two of the
items mentioned above, which is
P

1
, the
3
)1 (33 )(1
2
PPP

. The entire
)1 (3
23
sensors 3


PPPPmultiple

(4)
We deduce further the sensing probability of a two-set or a
three-set sensor group. As mentioned above, the sensing
probability of a single group is (4) and that of three groups
covering the sensing areas is (5).



1
 ) 1 (3)1 (33
)1 (3
)1 (

3

23
2
23
3
23
23
sensors group 3


PPPPPP
PPP
PPP
P
multiplemultiplemultiple
group




(5)
Table III shows the overall sensing probability with respect
to the different counts of the sensors in a sensor group.
TABLE III
OVERALL SENSING PROBABILITIES AND COUNTS
)(nPmultiple


single
P

n=3
Single
0.2160
0.3520
0.5000
0.6480
0.7840
n=5
Single
0.1631
0.3174
0.5000
0.6826
0.8369
Three
0.1198
0.2845
0.5000
0.7155
0.8802
Five
0.0709
0.2382
0.5000
0.7617
0.9290
Three
0.0711
0.2383
0.5000
0.7617
0.9289
Five
0.0335
0.1868
0.5000
0.8132
0.9665
0.30
0.40
0.50
0.60
0.70

We add the pressure sensors to the PIR sensors in a
combined design whose overall sensing probability is shown
in (6).
) (
Complement
UP
groupgroup
P
groupgroup
PUU

(6)
) (
groupgroup
P
group
UU
pressure sensors and PIR sensors to increase the sensing
probability. Table IV shows the overall probability by adding
the pressure sensors for sensing the intruder in the alert state.

is the probability of the incorporation of
TABLE IV
OVERALL SENSING PROBABILITIES AND INDIVIDUAL SENSING
PROBABILITY
Pgroup＋＋Ugroup –( Pgroup ×Ugroup )
Psingle Usingle 0.3 0.4
0.3
0.2253 0.3702
0.4
0.3702 0.4880
0.5
0.5599 0.6422
0.6
0.7496 0.7964
0.7
0.8945 0.9143

V. IMPLEMENTATION RESULTS
Table V shows the sensors used in different operating states.
We use PIR and pressure sensors as the sensors in the alert
state since these sensors consume little power. Because the
ultrasonic sensor has advantages in the sensing range, it
consumes, of course, more power. We use it and the PIR
sensors in the detection state.

TABLE V
THE SENSORS USED IN DIFFERENT OPERATING STATES
0.5
0.5599
0.6422
0.7500
0.8578
0.9401
0.6
0.7496
0.7964
0.8578
0.9191
0.9659
0.7
0.8945
0.9143
0.9401
0.9659
0.9856

System State Alert State Detection State
PIR sensor
Use Use
Pressure sensor
Use Not in use
Ultrasonic
sensor
Not in use Use

To enhance the sensing reliability, we use multiple PIR
sensors and pressure sensors in alert sensor groups to detect
an intruder. Each group uses the MVM management to
improve the sensing probability. We also install another alert
sensor group indoors to avoid any miss-sensing outdoors. Fig.
single
P

)(nPmultiple

n = 3
0.216
0.352
0.500
0.648
0.784
n = 5
0.163
0.317
0.500
0.682
0.836
n = 7
0.126
0.289
0.500
0.710
0.874
n = 9
0.098
0.266
0.500
0.733
0.901
n = …


0.500


n ∞
0
0
0.500
1.000
1.000
0.30
0.40
0.50
0.60
0.70

Page 5

Y.-W. Bai et al.: Design and Implementation of a Home Embedded Surveillance System with Ultra-Low Alert Power 157
9 shows that many alert sensor groups are placed both indoors
and outdoors in the alert state. When the alert sensor group
senses an intruder, the MCU is woken up and turns on the
power supply for the detection sensor group.
3583mm

Fig. 9. Our sensor placement in the alert state.

Fig. 10 shows our sensor placement in the detection state
where we use ultrasonic sensors and PIR sensors to sense
an intruder. The ultrasonic transmission is blocked when an
intruder enters the transmission path of the sensing area.
But as the ultrasonic transmission spreads a beam angle
which is likely to cause interference, we use multiple
receivers, and the MVM enhances the overall sensing
probability. The PIR sensor circuit has a least three groups
of PIR sensors placed on the ceiling of the specific room
environment.


Fig. 10. Our sensor placement in the detection state.

To find the average power consumption of the overall
operation we make Vdd the working voltage. IDetection is the
current value in the detection state. DutyCycleDetection is the
detection period, IAlert is the current value in the alert state and
DutyCycleAlert is the alert period. The average power
consumption is as shown in (7).
Poweraverage = Vdd ×(IDetection×DutyCycleDetection)
＋(IAlert×DutyCycleAlert) (7)
To reduce the average power consumption one can reduce
the operating voltage. However, if we reduce the voltage of
the sensor, the sensing probability of the PIR sensors may be
reduced as shown in Fig. 11. It shows the sensing distance as
about 4 to 5 meters when the working voltage is 3V. Since
the ceiling height is about 3 m, this is a suitable place for the
PIR sensor group. Hence the PIR sensor can have a lower
voltage and still enough sensing probability. For the working
voltage of the pressure sensors and the MCU we choose 3 V
for our prototype.
1 1.52 2.533.54 4.555.56
40
50
60
70
80
90
100
Sensing Distance (m)
Sensing Probability (%)


3V
4V
5V

Fig. 11. The sensing probability of different working voltages.

Table VI shows the power consumption of a single sensor
group if there is no intruder. In this case the power
consumption of a single PIR sensor group is about 1 mW,
which increases to 7.3 mW when a PIR sensor detects an
intruder. The pressure sensor consumes 0.8 mW, which
increases to about 1.5 mW when it senses an intruder. The
ultrasonic sensor group needs to constantly transmit and
receive signals, so it consumes around 256 mW.

TABLE VI
THE POWER CONSUMPTION OF A SINGLE SENSOR GROUP
Alert State
Power Consumption
mW
System State
Detection State
Power Consumption
mW
PIR sensor
1.05 7.35
Pressure sensor
0.8 1.5
Ultrasonic
sensor
Not in use 256.08

We use the MCU which has a sleep mode to reduce the
alert power, and we reduce the working clock of the MCU to
its minimum clock rate to reduce the power consumption. At