MASSEY UNIVERSITY

SCHOOL OF ENGINEERING AND ADVANCED

TECHNOLOGY

Engineering Project
Submitted as part requirement for B.Eng (Hons)
In
Telecommunications and Network Engineering

Wireless Intelligent Bed Sensing System

By

Usman Awan
04107411

2009

SUPERVISORS
1. Dr. Subhas Mukhopadhyay
2. Dr. Gourab Sen Gupta

0
Acknowledgements:

First of all I would like to thank my supervisor Associate Professor Dr. Subhas
Mukhopadhyay for all his support and help throughout the project. All the credit goes to
Dr. Subhas for giving me the opportunity to work on this project.

I would also like to thank my co-supervisor Senior Lecturer Dr. Gourab Sen Gupta for
helping me through this project. His help during the microcontroller programming process
played a vital role in the success of the project.

I am also very thankful to PhD student Anuroop Gaddam for all his help, his discussions,
problem solving skills and assistance that he provided throughout the project. Without
Anuroop this project would not have been possible.

1
Summary:

The elderly population of the world is growing rapidly, creating the need to increase
geriatric care. More and more old people prefer to live independently alone at home. Due
to the advancement in wireless communications and electronics over the past years, the
development of low cost, low power multi functional sensors is receiving increasing
attention. This sensing technology is playing a vital role in enhancing the quality of life for
elderly. A wireless sensor based smart-home monitoring system is being developed which
detects the usage of household appliances.

One of the main sensors used in this smart home is the bed sensor system which monitors
sleeping pattern of the user. This system is developed for the elderly people living alone at
home or people with disabilities, who require immediate medical attention in case of an
emergency or when an abnormal situation arises. In case of emergency the system sends
a message to the caregiver or the person monitoring the system. This provides monitoring
of the users without invading their privacy.

In this report we discuss the design of an Intelligent Wireless Bed Sensor System to
monitor the elderly. This sensor system not only detects the presence of a person on the
bed, it can also indicate the person’s position on the bed. This system uses four sensors
which are placed under the four legs of the bed. These sensors are connected to a driver
circuit to get the analogue output which is converted into digital using the ADC (Analog to
Digital Converter) of the microcontroller.

Initial test results for this project were conducted using 25 lbs sensors connected to the
legs of a small coffee table. These tests made sure that our algorithm and driving circuit
was working properly. Then this system was used with 100 lbs sensors to get experimental
results by connecting the sensors to the four legs of the bed. The tests were conducted for
various subjects using the bed. All these tests gave us positive results and proved the
potential of the Wireless Intelligent Bed Sensor System.

2
 Table of Contents

1.0 Introduction …………………………………………………………………………………………………………….. 6

2.0 Background …………………………………………………………………………………………………………….. 10

3.0 Need for Bed Monitoring ………………………………………………………………………………………… 13

4.0 System Overview …………………………………………………………………………………………………….. 15

5.0 Hardware ……………………………………………………………………………………………………………….. 17

5.1 Flexi-Force Sensors ……………………………………………………………………………………… 17

5.2 Driver Circuit ………………………………………………………………………………………………. 20

5.3 Microcontroller …………………………………………………………………………………………… 24

5.4 RF Module …………………………………………………………………………………………………… 25

6.0 Software ………………………………………………………………………………………………………….……… 28

7.0 Experiments and Results …………………………………………………………………………………......... 30

7.1 Sensor Characteristics …………………………………………………………………………………. 30

7.2 The four sensor response using a table ……………………………………………………….. 32

7.3 Sensor Test using a Bed ……………………………………………………………………………….. 41

8.0 The Market Potential ………………………………………………………………………………………………. 45

9.0 Problems Faced ………………………………………………………………………………………………………. 47

10.0 Conclusions …………………………………………………………………………………………………………… 48

11.0 Future Work ………………………………………………………………………………………………………….. 50

12.0 References …………………………………………………………………………………………………………….. 51

13.0 Appendices ……………………………………………………………………………………………………………. 53

3
 List of Figures

Figure 1 World Population Growth …….……………………………………………………………………......... 7

Figure 2 Life Expectancy of the World ………………………………………………………………………….…. 8
Figure 3 Configuration of a smart home monitoring system based on limited number
of wireless sensors ………………………………………………………………………………………………………… 11
Figure 4 Functional block diagram of a wireless bed sensor system ………………………………. 16
Figure 5 A Flexi-Force Sensor …………………………………………………………………………………………. 17
Figure 6 Sensor Construction …………………………………………………………………………………………. 18
Figure 7 Resistance Curve & Conductance Curve ………………………………………………………….… 19
Figure 8 Sensor Response Graph ………………………………………………………………………………….… 19
Figure 9 The recommended sensor circuit ……………………………………………………………………… 21
Figure 10 Schematic of the Driving Circuit ……………………………………………………………………… 23
Figure 11 PCB design of the Driving Circuit ………………………………………………………..………….. 23
Figure 12 Linx RF Module ………………………………………………………………………………………………. 26
Figure 13 Schematic of the RF Module ………………………………………………………………………….. 27
Figure 14 PCB of the RF Module …………………………………………………………………………………….. 27
Figure 15 Sensor head sandwiched between disc plates ………………………………………………... 30
Figure 16 Force to Voltage characteristics of the sensor ……………………………………………….. 31
Figure 17 The table and sensor setup ……………………………………………………………………………. 32
Figure 18 Driving Circuit and the Microcontroller with the sensor display …………………..… 33
Figure 19 The table surface divided in to cells along and x and y axes ………………………..… 34
Figure 20 Sensor 1 response ………………………………………………………………………………………….. 35
Figure 21 Sensor 2 response ………………………………………………………………………………………….. 36
Figure 22 Sensor 3 response ………………………………………………………………………….................. 37
Figure 23 Sensor 4 response ……………………………………………………………………………………….... 37
Figure 24 Combined results of all the four sensor outputs ……………………………………………. 38
Figure 25 Zone Allocation …………………………………………………………………………………………….. 38
Figure 26 The placement of a sensor under a leg ………………………………………………………….. 41

4
Figure 27 The position of four sensors under the four legs of a bed …………………………... 42
Figure 28 An elderly person lying on bed is being monitored …………………………………..... 42
Figure 29 A child lying on bed is being monitored ……………………………………………………... 43

List of Tables

Table 1: Experimental observation of locating the weight …………………………………………….. 39

Table 2: Location of the weight …………………………………………………………………………………….. 40

Table 3: Experimental results with inhabitant lying on bed ……………………………………………. 44

5
1. Introduction:

One of the most important features that human beings have is the ability to sense. We
use our senses in our daily life to perform various tasks. Sensors are also playing a vital
role in the development of science and technology. A sensor acquires a physical
parameter and converts it into a signal suitable for processing. Sensors are electrical or
mechanical components that are used to measure a property or behavior of an object or
system. Sensors can be mechanical such as pressure sensors, electrical such as voltmeter,
ammeter, radar etc, thermal such as thermometers or optical sensors such as photo-
detectors, infrared sensors, fiber-optic sensors etc.

Wireless sensors have much more advantages over wired sensors. These sensors are easy
to install as no wiring is needed so they simplify the designs of system. The wireless
sensors are much more flexible, portable and consume less power as compared to wired
sensors. Wireless sensors can be used in locations that are geographically far apart to
monitor activities remotely. These sensors also allow easy expansion or upgrade of
existing systems.

The average human age i.e. the life expectancy of humans has increased over the past
decade. This means that there are more elderly people living in the world today then
there were 10 – 20 years ago. The world population growth can be seen in the figure
below. One problem adding to the aging population is that hospitals aren’t able to coupe
with large number of patients so hospitals are looking to send people back as soon as
possible to recoup at home. Many elderly people prefer to live alone at home rather than
living with their adult children or living in rest homes.

6
 Fig. 1: World Population Growth

From this plot we can see that the population of people above 65 years old is increasing at
a higher percentage as compared to people below 65. For example if we look at the year
2010, the population growth of people below 65 years old is increasing at the rate of 3%
while population growth of people 65 years old or above is increasing at a rate of 9%. This
shows that in the year 2010 the elderly people will have a 6% higher population growth
rate than under 65 year olds. These results prove that the elderly population is increasing
at a greater rate than young people. The main reason for this is the improved healthcare
system over the years.

The proportion of elderly people in any population is growing rapidly thus creating a need
to increase geriatric care. In New Zealand the median age of the population is projected to
rise from 36.6 in 2010 to 43.1 in 2050 [8].

7
 Fig. 2: Life Expectancy of the World

As sensing technology continues to evolve, it is playing an important role in improving the

quality of life for elderly people and their families. A home based on wireless sensor
systems has been developed for monitoring. The current system is based on sensors
which detect the usage of household appliances/devices. It has been realized that there is
a need of incorporating a bed-sensor to the current monitoring system.

The main aim of the bed sensor system discussed in this report is to monitor the sleeping
patterns of an old person or a patient living alone in a house who needs medical attention
in case of emergency or if an abnormal situation arises. This sensor system will also be
used to monitor the general sleeping patterns of any humans. They will be helpful in
determining when a person is restless at night or when he/she is sleeping calmly. This
intelligent sensor system can perform various data collection and processing functions,
including data interpretation and fusion of data from multiple sensors. There are several
sensors readily available in the market, but making them intelligent in context to a specific

8
situation is always a challenging task. In our case this intelligent bed sensor system will not
only detect the usage pattern of the bed but will also have the capability to collate data
and flag out anomalies. This sensor system can be a life saving device.

In our bed sensor system we will be using four flexi force sensors. These sensors will be
placed under the four legs of the bed. These sensors will measure the weight on the bed
and respond accordingly. The sensor system will send the information to a central
controller unit which in turn is connected to a PC. The sensor unit and the PC
communicate wirelessly through radio frequencies (RF).

9
2. Background:

A lot of work has been done over the past years of using technology to monitor and help
out people. Sensor technology is playing a vital role in this field. Sensors are being used to
monitor all kinds of human activities and are making lives much easier. One of the areas
where this technology is playing a key role is the monitoring of elderly people living alone
at home. Another area it is helpful in is monitoring people with disabilities without a
helper being at their side constantly.

Work is being done on creating a smart home. A smart home is a home based on wireless
sensing technologies. This home is based on the integration of different sensor
technologies as shown in figure below. The sensors used communicate wirelessly. The
objective of this home is to monitor appliances/devices of special interest, both electrical
and non electrical. This smart home is basically being created for old people who prefer to
live alone at home. With the help of this smart home these elderly people can be
monitored without invading their privacy. In case of an emergency or unusual activity a
warning message can be sent to the family relatives or the care-giver without them
actually being at the home. It is not required to monitor all the appliances at home but the
daily activities of a person under monitor should be available to the system. The data from
these sensors are collated by a central processor, which saves all the data for processing
as well as to understand the life style of the person under care by the system.

10
Fig. 3: Configuration of a smart home monitoring system based on limited number of
wireless sensors

One of the main sensors used in the smart home is the bed monitoring sensor. This sensor
is used to monitor the sleeping behavior of the user. This sensor is in particular useful to
monitor elderly people living alone at home or people living at homes with disabilities.
These sensors use weight measuring technique to indicate if a person is on the bed or not.
This sensor system can prove to be a life saving device for many users. There have been
many different types of bed sensors made and are readily available in the market. Some of
these sensors are wired and some use wireless technology.

There are bed sensor mats and pads available in the market. The sensors used in these
pads are pressure sensors. These sensor pads/mats are either placed on the mattress
under the bed sheet. These sensor systems alert the care-givers via an alarm that a person
is getting out of the bed. That is, when the weight is taken off the mat, the alarm sounds.
Many of these types of sensors are being used in hospitals for monitoring fall risk patients
and residents. The problem with these sensor mats is that they are placed on the bed and

11
can result in discomfort for the user. Another problem is these sensors are usually wired
so these wires can also be a hindrance and result in discomfort of the user. These sensor
systems are also expensive to run as compared to wireless sensor systems.

A prototype of a bed monitoring device using wireless sensor has been developed, which
has demonstrated its feasibility for monitoring the activity of a user in bed. This system is
different from the sensor pad-alarm systems. This system offers the monitoring of
subjects 24 hours 365 days a year. These sensors offer the monitoring of bed at low cost,
less complexity and good accuracy. The system uses a Flexi-Force sensor to measure the
pressure and give the results according to that. The sensor is connected to a driving circuit
which in turn is connected to a microcontroller to convert the analog readings from the
circuit into digital. The microcontroller can send this information using RF (Radio
Frequency) signals wirelessly to the controller unit. These wireless sensors are much
efficient to use than wired sensors. These sensors are flexible, portable and consume less
power. The sensor is placed under leg of a bed so they provide monitoring without any
disturbance or discomfort to the user. The problem with this sensor system is that it only
uses one sensor which just provides us one result which is either there is someone on the
bed or there is no one on the bed.

The aim of our project was to create a bed monitoring system using four sensors. This
makes our sensor system intelligent. We will not only monitor if a person is on the bed or
not but we can gather a variety of different results based on the four sensors. The four
sensors are also placed under the four legs of the bed and are connected to a driving
circuit and a microcontroller to get the analog and digital results. Our intelligent bed
sensor system can not only be used to monitor elderly people but can also be used to
monitor the sleeping pattern of any person in general.

12
3. Need for Bed Monitoring:

In the process of developing a smart digital monitoring home, it has been realized that
there is a need for incorporating a bed check monitoring system with the existing system.
In the absence of caregivers at night for an elderly person living alone, it is required for a
preventive measure to avoid sudden accidents in the home environment. The main
purpose of this kind of sensor unit is to monitor the bed usage of an old person or a
disable patient who is living alone in a house and needs immediate medical attention in
case of an emergency.

Since, a bed is a place where a person spends most of their time in the house. We
consider three scenarios in which the bed sensor will be useful.

 A person who has a habit of waking up at a particular time in the morning sleeps
beyond his usual wakeup time.
 A person who has a particular sleeping time and he/she is not in bed past their
usual bed time.
 If a person is constantly moving on the bed i.e. they are restless.

The first two situations indicate that the person might have had a serious health issue or
fell down somewhere in the house which made him/her unable to reach for medical
assistance. The third scenario may indicate that the person is unwell or medically
distressed.
This intelligent bed monitoring sensor system cannot only be used for elderly and patients
but it can also be used to monitor the sleeping patterns of anyone. We can monitor
people who have restless sleep at night and gather the results for that. It can also be
helpful in monitoring someone who has a habit of sleep-walking.

13
So, the Bed Monitoring Wireless Sensor monitors the abnormalities in the person’s daily
sleep window period and may act as a lifesaver. Many types of bed monitoring devices
have been developed in the past, but these systems are mainly used to detect monitoring
fall risk-patients and residents in a hospitals scenario. Our bed sensor system will
incorporate intelligent features, it will not only monitor if a person is on or off the bed but
will also be helpful in monitoring the sleeping patterns of patients 24 hours a day 365 days
a year.

Although another option to monitor the bed could be the use of cameras. This would be
efficient but the problem with this approach is that it would incur higher costs to buy
these 24 hours surveillance cameras and to install and monitor them. The use of cameras
will also affect the privacy of the user which is a major feature that the bed sensor system
provides.

14
4. System Overview:

The intelligent bed monitoring system is based on four Flexi-Force sensors. These are
pressure measurement sensors and respond accordingly when a force/load is applied to
the sensing area of the sensor. In the case of bed monitoring, these four sensors are
placed under the four legs of the bed to determine if a force is being applied on the bed
i.e. someone is lying on the bed. The sensors were calibrated in such a way that the
system could confirm that the force on the bed is exerted by a human subject rather than
a pet or an inanimate object. The sensors are then connected to a driving circuit. This
driving circuit is a signal conditioning circuit to reduce noise in the signal. The driving
circuit provides us with a voltage value at the output so we get linear values of
load/weight versus output voltage plot. These sensor output voltages are analog values
so, they are sent to the microcontroller to convert them into digital values. We use the
ADC (Analog to Digital Converter) of the microcontroller to achieve this task. The
measured weight of each sensor is displayed on the LCD of the microcontroller. The
microcontroller is then connected to a RF (Radio Frequency) transmitter which transmits
the data wirelessly to a central processing unit which in turn is connected to a PC
(Personal Computer). The results from the PC can be sent to the care-giver or the person
monitoring the bed sensor system. A block diagram of the system is shown in figure 4.

15
 LCD Display
Bed Amplifier /
Sensor 1 Signal conditioning

Bed Amplifier / ADC

Sensor 2 Signal conditioning RF
Transmitter
Bed Amplifier / µController
Sensor 3 Signal conditioning

Bed Amplifier /
Sensor 4 Signal conditioning

Fig. 4: Functional block diagram of a wireless bed sensor system

16
5. Hardware:

The hardware that we use in our system plays an important role in making the system
efficient and effective. The hardware components are the basic parts that make up our
bed sensor system. The main hardware components are Flexi-Force sensors, the driving
circuit, and the microcontroller used for converting analog values to digital and the RF
module for transmitting and receiving data wirelessly.

5.1 Flexi-Force Sensors:

The bed sensor system uses the Tekscan’s Flexi-Force Sensors. The Flexi-Force
sensors were chosen because of their reliability, cost-effectiveness and ease of
integration.

Fig. 5: A Flexi-Force Sensor

The Flexi-Force sensor is ultra-thin (0.008”), flexible printed circuit that senses contact
force i.e. the force or load being applied to the sensor. The force sensor is constructed of
two layers of substrate (polyester) film. On each layer, a conductive material (silver) is
applied, followed by a layer of pressure-sensitive ink. Adhesive is then used to laminate
the two layers of substrate together to form the force sensor. The active sensing area is
defined by the silver circle on top of the pressure-sensitive ink. Silver extends from the
sensing area to the connectors at the other end of the sensor, forming the conductive
leads. The sensors are terminated with male square pins, allowing them to be easily
incorporated into a circuit. The two outer pins of the connector are active and the center
pin is inactive.

17
Fig. 6: Sensor Construction

Flexi-Force sensors are strategically placed underneath the bed legs to determine if a
force is being exerted on the bed. The Flexi-Force single element force sensor acts as a
force sensing resistor in an electrical circuit. When there is no force applied on the sensor
the resistance of the sensor is very high. When a force is applied to the sensor, this
resistance decreases. This resistance is inversely proportional to the force applied on the
sensor. The conductance curve of the force-conductance graph shows linear relationship.
The conductance and resistance curves for a force sensor are shown below.

18
Fig. 7: Resistance Curve Conductance Curve

One way in which the sensor can be used is to use it in an application to incorporate it into
a force-to-voltage circuit. The output voltage varies linearly with the exerted force. This is
shown in the figure below.

Fig. 8: Sensor Response Graph

19
The pressure sensor that we are using for our bed sensor is the 100 lbs (pounds) sensor.
This sensor can measure up to 444N of force (45.5 kg). To differentiate between
inanimate objects and human beings, we calibrate the sensor unit so that the force
applied by each extremity should produce a significant voltage output.

Some of the physical parameters of flexi force sensor are:

1. The operating range of temperature is from -9°C to 60°C,
2. Repeatability is +/- 2.5% of full scale (conditioned sensor when 80% force applied),
3. Linearity is <+/- 5%,
4. Hysteresis is <4.5% of full scale

5.2 Driver Circuit:

The four sensors of our bed sensor system are all connected to a driving circuit to
get the output voltage to force relationship. The recommended driving circuit for the flexi
force sensor is shown below. This circuit is driven by a 5V DC excitation voltage. The
circuit uses an inverting amplifier to produce analogue output based on the sensor
resistance and a fixed reference resistance (RF). The dynamic force range (sensitivity) of
the sensor can be changed by changing the value of RF or by changing the drive voltage VT.
A lower reference resistance and/or a drive voltage will make the sensor less sensitive and
increase its active force range. A -5V is also connected to drive the sensors.

20
 Fig. 9: The recommended sensor circuit

Our driver circuit used in the bed sensor system is based on this recommended Flexi-Force
sensor circuit. In our driver circuit we use the IC (Integrated Circuit) chip LM324N. This
consists of four low power quad operational amplifiers. The chip also has an external
power supply which is 5V and a ground pin. The four sensors are connected to each of the
four amplifiers. These inverting Op-Amps provide us with analogue voltage signals. The
sensor input is connected to the inverting input of the amplifier and the non-inverting
input is connected to ground. We use a variable resistor (potentiometer) to change the
sensitivity of the sensors according to our needs.

We use a 3 pin voltage regulator LM7805 at the start of our driving circuit. This voltage
regulator is a fixed 5V output regulator. This regulator has an input pin, an output pin and
a ground pin. The input pin is connected to the external input voltage of 9V. The regulator
converts this voltage and gives 5V at its output pin. The purpose of using this regulator is
so that we can use our sensor system anywhere by connecting a 9V or 12V adapter as the
voltage supply. The regulator converts this voltage into 5V and gives that to the rest of the
driver circuit to work appropriately.

21
We can see from the recommended circuit that we need a positive as well as negative 5V
supply to connect to the sensor. This negative 5V is achieved by using a voltage inverter
chip. The chip that we use in our driving circuit is MAX828 chip. This chip is used to
generate a -5V supply from a +5V logic supply to power analogue circuitry. The MAX828
are ultra-small monolithic, CMOS charge-pump inverters. This IC comes in a 5-pin package
and can deliver 25mA with a voltage drop of 500mV. This chip only requires two external
10uF capacitors to get the desired output.

Since MAX828 is a Switched-Capacitor voltage inverter so it causes a 12 kHz noise at the

output. This noise can considerably affect the output of the sensors, resulting in an
unstable output. To eliminate this noise we have used a low-pass filter at the output
terminal of each of the sensors. This filter suppresses the high frequency noise in the
output of the amplifier. The waveform output from the driver circuit before and after the
filter is shown in Appendix A.

The circuitry for one sensor is shown on the next page. Rs represents the force sensor. The
same is required for all the four sensors. The complete schematic of the system is shown
in Appendix A.

22
 Fig. 10: Schematic of the Driving Circuit

4 13 12

8 7
5
3 2 1 3 2 1 3 2 1 3 2 1

17
0 1 9
14

1 2
1 2 1 2 1 2
1 1 1

1
2 2 2

3 3 3 3

2 2 2
2

1 2 1 1 2 1 2 1 1 2 1
1 2

1
2 1 10

14 13 12 11 10 9 8

1
3 4

0 2

1 5 9
2

1 2 1 3
1 2 3 4 5 6 7

7 3

2 2 2

2 3 2 1

8 4

1 1 1

Fig. 11: PCB design of the Driving Circuit

23
 5.3 Microcontroller:
The main purpose of using a microcontroller in our bed sensor system is to read
and display the results digitally. The microcontroller used in our system is Silabs
C8051F020. This microcontroller was chosen because of its rich resources. We used the
ADC (Analog to Digital Converter) of the microcontroller to convert our analogue outputs
from the driver circuit. The ADC has 12-bit resolution and 8-channel multiplexed single-
ended inputs. We used four of these inputs to connect the four analogue outputs from
our sensors. The digital values are displayed on the LCD of the microcontroller. Digital I/O
ports (port6 and 7) are used to interface the microcontroller to the LCD module.

The reference voltage of the ADC converter of the microcontroller is 2.4V. Therefore the
maximum output response of the sensor is made to be below or equal to 2.4V. This is
done by adjusting the Rf1 potentiometer as the drive voltage is fixed at -5V. The ADC is
capable of running at a sampling frequency of 100 KBPS which is adequate for this
application. The conversion clock frequency of the SAR (Successive Approximation
Register) can be programmed to be up to 2.5 MHz.

The microcontroller runs off a crystal of 22.1184 MHz and is programmed in C. The
microcontroller has two UART (Universal Asynchronous Receiver Transmitter) modules of
which one is used to interface with the RF transmitter. The UART outputs RS232 protocol
serial data at TTL level; hence interfacing with the RF module does not require any
additional translator. The UART is programmed to transmit serial data at 38400 baud rate,
8-bit data, 1 stop bit and no parity.

24
 5.4 RF Module:
The RF (Radio Frequency) module is used to transmit the data between the
microcontroller connected to the driving circuit and the central control unit. The RF
module is connected to output port of the microcontroller. It takes the readings from the
microcontroller and transmits them via the RF circuit to the controller unit using radio
frequencies.

The RF module consists of three main components, a Linx® TRM-418-LT chip, a multiplexer
chip MC14016BCL and a hex inverter from the chip SN74HC04D.

The Linx chip is a transceiver which is used for bidirectional wireless transfer of data,
control or command information in 418 MHz band. This chip is a surface mount device and
it has long range, low cost and low power consumption. The chip has 12 pins out of which
we are using 9 in our RF module circuit. Since this chip is a low power consumer we supply
it power directly from the microcontroller port. So, we don’t need an external power
supply for this chip. Pin 1 is connected to the antenna which is used for RF transmission of
data. Pin 2 and 10 are connected to the ground of the microcontroller port and pin 11 is
connected to its power supply. Pin 5 is the Analog RMS (Average) Voltage Reference which
is connected to Vcc (Port 11). Pin 12 is the level adjust line which can be used to adjust the
output power level of the transmitter. This pin is also connected to Vcc to get the highest
output, while placing a resistor to Vcc lowers the output level. Pin 9 is the Power Down
which is used to pull this line to low or leave floating to place the receiver into a low-
current state. The module does not send or receive a signal in this state so we pull high to
activate the transceiver. Pin 8 is the Transmit/Receive which is used to pull the line low to
place the transceiver into receive mode and to pull it high in the transmit mode. Pin 7 is
the digital data line which will output the received data when in receive mode and it is the
data input in the Transmit Mode. The diagram of all these pins connected to the module
circuit is shown in figure 13.

25
Fig. 12 Linx RF Module

The MC14016BCL chip is a quad analog/quad multiplexer. This chip acts like a switch and
consists of four independent switches capable of controlling either digital or analog
signals. We use two of these switches in our module circuit to transmit and receive data.
This chip is used to multiplex the single data line available on Linx RF module, in order
transmit and receive the data. To make the switch work we set the control line which
comes from the microcontroller port to high (1).

We also use an inverter to control the output from the T/R select pin of the Linx chip. So if
our module is in transmit mode and control line is high, then switch 1 will work. Since the
control line is high it will set switch 2 to receive mode as well. In order to avoid the
problem of having both transmit and receive switches working at the same time, we use
an inverter in front of the control input to switch 2. This sets switch 2 to low, thus no data
is received. When the control line is set to low switch 1 does not transmit anything, so
switch 2 starts to receive data.

The microcontroller Port 1 is used to transmit and receive data. Pin 1 is used to transmit
and Pin 2 to receive data. Pin 5 of this port is used as a control pin which is set to either hi
(1) or lo (0). When it is hi it allows the data to go through and stops it when it is lo. This
control pin is connected to the two switches to make them work.

26
 Fig. 13: Schematic of the RF Module

2 4 6 8 10

1 2
1 7
1 3 5 7 9

13
2 8

6 1 3 5
7

3 9 12

14 13 12 11 10 9 8
4 10

5 11

6 12

1 2 3 4 5 6 7

10 8 9
11

14 13 12 11 10 9 8
1 2

1 2 3 4 5 6 7

Fig. 14: PCB of the RF Module

27
6. Software:

There are basically two main types of software that we have used in the implementation
of the Intelligent Bed Sensor System. We have used Altium Designer and Silabs IDE
(Integrated Development Environment).

Altium designer was used to create the circuit diagrams of our driver circuit. Then these
schematics were converted into PCB (Printed Circuit Board) diagrams and the physical PCB
was constructed based on these diagrams. Altium provided a huge library selection from
which all the components and their footprints were easily found. The projects created in
Altium were all PCB projects.

The other software which was used as part of the project was Silabs IDE. This software
was used to program the microcontroller. We used C language to program the
microcontroller. Silabs has its own development studio which provides an environment to
develop the software and then load the software onto the microcontroller. The software
initializes the microcontroller by setting values in registers in the microcontroller. This
initializes the input and output ports and enables and disables the microcontroller.

In our microcontroller program we use the 12-bit Analog to Digital Converter ADC0. This
consists of 9-channel, configurable analog multiplexer (AMUX0), a programmable gain
amplifier (PGA0) and a 12-bit Successive Approximation Register (SAR) ADC. ADC0 is
enabled by setting AD0EN to 1. AMUX0, PGA0 and the ADC data conversion modes are all
configurable via SFRs. The ADC0 can be programmed through the following sequence:

1) Configure the voltage reference (REF0CN)

2) Set the SAR0 conversion clock frequency and PGA0 gain (ADC0CF)
3) Configure the multiplexer input channels (AMX0CF)
4) Select the desired multiplexer input channel (AMX0SL)

28
 5) Set the appropriate control bits and start of conversion mode and turn on ADC0
(ADC0CN)

The ADC conversion takes place when the timer flows. In our program we are using Timer
3. Timer 3 is always configured as an auto reload timer, with the 16-bit reload value held
in TMR3LL (low byte) and TMR3RLH (high byte) registers. The Timer 3 is programmed in
the following way:

1) Write the auto-reload value into the auto-reload registers (TMR3RLL and
TMR3RLH)
2) Write the starting value for count up sequence into the count registers (TMR3L and
TMR3H)
3) Select the desired clock source (T3XCLK) and frequency (T3M), set the control bits
(TR3) and turn on Timer 3 (TMR3CN)

TMR3CN is used to select the clock source and is the only SFR used to configure Timer 3.

29
7. Experiments and Results:

In this section we present the results and tests carried out during the project. There were
three main phases of experimentation conducted for this project. First we tested the
individual sensor characteristics, then we tested the four sensors using a table and then
the sensor system was tested using a bed. All these phases are described in detail below.

7.1 Sensor Characteristics:

The first step of the experimentation process was testing the characteristics of the
Flexi-Force sensor. The sensor used for this experiment was 25 lbs sensor. First thing we
did was make small metal discs which had the same diameter as the active sensing area of
the sensor. The sensing area of the sensor was sandwiched between two of these metal
discs when taking measurements. The purpose of this was to ensure that the load applied
to the sensor is evenly distributed across its sensing surface.

Fig. 15: Sensor head sandwiched between disc plates

30
This sensor was tested by applying load to it and plotting the Force versus Voltage graph.
The sensor was connected to the driving circuit to get the analogue output thorough a
multi-meter and digital output through the microcontroller. For our experiment we used
sand in a pot as weight. A small solid cylindrical base with the same diameter as the
sensor head was used at the bottom of the pot to concentrate the load directly on top of
the sensor. The load applied to the sensor was from 0.5kg up to 11.5 kg in incremental
steps of 0.5 kg. Three consecutive readings were taken for every weight added to the
sensor to check its repeatability. The average of these three readings was taken as the
final output. Beyond 11.5 kg the response of the sensor was saturated and for any
additional weight to that we got the same reading as at 11.5 kg. This was because we
were using a 25 lbs (11.4kg) sensor. The Force to Voltage characteristics of the sensor are
shown in the figure below.

Sensor Test 2 y = 0.1939x + 0.0848

R2 = 0.9872
2.6

2.4

2.2

1.8

1.6
Voltage (V)

1.4

1.2

0.8

0.6

0.4

0.2

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10 10.5 11 11.5 12 12.5
Force (kg)

Fig. 16: Force to Voltage characteristics of the sensor

31
 7.2 The four sensor response using a table:
The next step of experimentation involved testing all the four sensors connected
together and reading and comparing their outputs. At the time of this experiment the 100
lbs sensors (which are required for a bed) were not available. So, this experiment was
conducted using the 25 lbs sensors placed under the four legs of a small coffee table as
shown in the figure below. This sensor experiment was done to test the accuracy and the
usefulness of the proposed algorithm.

The table used was an evenly balanced table with equal amount of force being exerted on
each of the legs. The sensors were sandwiched between flat metal plates and connected
(taped) to the bottom of the legs of the table. These plates were used to ensure that equal
amount of force was distributed to every point in the sensing head of the sensor. It was
also made sure that the table is being tested on an even floor so that equal amount of
force is applied on each of the legs.

Fig. 17: The table and sensor setup

32
The potentiometers (Rf) on the driving circuit were also adjusted to show the same output
value for all the sensors in idle condition. This was useful as the force was applied on the
sensors, all the values increased with the same rate and showed the same value when
weight was applied at the center point of the table.

The sensors were connected to the driving circuit which gives the analog output voltage as
the force is applied on a sensor. This analog input was then sent to the microcontroller in
which the ADC converts it to a digital output and displays it on the LCD as shown in the
figure.

Fig. 18: Driving Circuit and the Microcontroller with the sensor display

The surface area of the table was calibrated and divided into four zones known as Zone1,
Zone2, Zone3 and Zone4. Each of these zones was further divided into cells along
horizontal x axis (width) and vertical y axis (height). The cells were 10x14 that is 10 cells
along the x axis and 14 cells along the y axis as shown in figure below. The table was
calibrated in this way to get the exact reaction of the sensors and to observe how force
being applied in each of the cells affected the output. A fixed amount of weight (6kg) was

33
used to test the sensors. We used a thin solid cylindrical base at the bottom of the weight
to make sure that the entire force was acting directly on the particular cell on the table.
The force versus voltage results were taken by placing equal amount of weight in each of
the cells and recording the output from the microcontroller. This gave us a good idea of
how the output reacts to force applied at every point on the table. These results
determined the reaction of our sensors when applied with force.

Fig. 19: The table surface divided in to cells along and x and y axes

After the set-up was complete for the experiments we started taking the results. A fixed
weight of 6kg was used to apply force at each of the cells and take the corresponding
output from the microcontroller.

34
The following surface plots show the result of our experiments done on the table. The
plots were plotted as the reaction force output which results as the 6kg weight is applied
on each of the cell. The plots show exactly where the maximum force is being applied. The
force is plotted on the vertical z axis and the horizontal y and x axes represent the cell
where the force is being applied. The surface colour distribution represents the amount of
weight that is being applied on the sensors.

From this first plot we can see the response of Sensor 1. The highest value for this sensor
is 884 which is equal to 8.84 kg. This value is at the cell x1y1 which is the position right
above the leg of the table where sensor1 is attached. The minimum value for this sensor is
at cell x10y14 which is above the leg where sensor4 is attached. This is to be expected as
the sensor will show the highest value when the weight is directly above it and show the
minimum value when the weight is at the farthest point from the sensor.

Fig. 20: Sensor 1 response

35
The following plot is the result for Sensor 2. From this plot we can see that the highest
value is around 700 which is at the cell x1y14 which is located right above the leg where
sensor2 is connected. The minimum value is 225 at cell x10y1 which is the farthest part
from sensor2.

Fig. 21: Sensor 2 response

Similarly it can be seen from the surface plot for Sensor 3 that the maximum value is at
cell x10y1 which is 845. This cell is located on top of the leg of the table where sensor3 is
placed. The minum value for this sensor will be at the cell located above the sensor2.

36
 Fig. 22: Sensor 3 response

Sensor4 also shows similar response with the highest value 888 being at the cell x10y14
which is above the leg connected to sensor4. The minimum value for this sensor is at the
cell located above the Sensor1.

Fig. 23: Sensor 4 response

37
The following plot shows the combined results of all the four sensor outputs. It can be
seen from the U-shaped curve that more force is being applied to the corners. These are
the four corners of the table under which the sensors are placed. This is true with the fact
that sensors will experince the maximum force when the weight is placed directly on top
of the snesor.

Fig. 24: Combined results of all the four sensor outputs

The table was calibrated and divided into four zones known as Zone1, Zone2, Zone3 and
Zone4 as shown in figure below.

Fig. 25: Zone Allocation

38
Ideally if the weight is in the centre point of the table, the response of sensors,
S1  S 2  S 3  S 4  Savg

where Savg  (S1  S 2  S 3  S 4) / 4

To detect the position of the weight with respect to the zone on the table, the
microcontroller program is made to continuously monitor the ADC inputs from the four
sensors. The weight position is calculated based on the following conditions:

If (S1  S 2)  (S 3  S 4) , then the weight is positioned in the upper half (UH)

If (S 3  S 4)  (S1 S 2) , then the weight is positioned in the lower half (LoH)
If (S1  S 3)  (S 2  S 4) , then the weight is positioned in the left half (LeH)
If (S 2  S 4)  (S1 S 3) , then the weight is positioned in the right half (RH)

Based on the above observation, it can be concluded that-

ZONE 1: If UH and LeH both are true,

ZONE 2: If UH and RH both are true,
ZONE 3: If LeH and LoH both are true,
ZONE 4: If LoH and RH both are true.

Actual Sensors Sensors Sensor Sensor Calculated

Position 1 2 3 4 Position

ZONE 1 890 441 498 370 ZONE 1

ZONE 2 470 791 385 485 ZONE 2
ZONE 3 480 405 815 460 ZONE 3
ZONE 4 385 452 472 857 ZONE 4

Table 1: Experimental observation of locating the weight

39
Once the above observations were done further investigations were carried out to
calculate/estimate, at which location the weight is predominant. The weight is positioned
at various points along the line joining the middle point and the edge of the location of
Sensors1 (dotted line in figure 22). By using the following interpolation technique the
position of the weight is calculated and is shown in table 2.

S1max : Maximum output of the Sensor1 corresponding to the weight located on leg 1

Savg : Average signal of the four sensors

S1 : Output of Sensor1
d : Distance between the center of the table to the edge corresponding to Sensor1
p: Position of the weight along the line

S1  Savg
p d (1)
S1max  Savg

Actual position of the

weight from the Calculated position (cm) Error (%)
centre (cm)

0 4 -
10 9 10
20 17 15
30 27 10
40 39 2.5
50 49 2

Table 2: Location of the weight

From table 2 it is seen that the system can be used to estimate the location of the weight
with a fair amount of accuracy. The error can be reduced further if the signals from the
other sensors are also used in the interpolation technique.

40
 7.3 Sensor tests using a bed:
The last step of the experimentation was to test the sensor system for the purpose
that it had been built for. In this part, the sensor system was tested using a bed and real
life scenarios. The sensors used in this experiment were the 100 lbs (45.5 kg) sensors.

Fig. 26: The placement of a sensor under a leg

The Flexi-Force sensors were strategically placed under the legs of the bed to determine if
a force is being exerted on the bed i.e. someone is lying on the bed. In this experiment the
sensors were calibrated to make sure that the force being exerted on the bed was from a
human body rather than some inanimate object. The schematic representation of the four
sensors placed under the bed is shown below. S1, S2, S3 and S4 represent the four sensors
Sensor 1, Sensor 2, Sensor 3 and Sensor 4 respectively.

41
 S2 S1

S4 S3

Fig. 27: The position of four sensors under the four legs of a bed

The output signals from the four sensors are interfaced to four different channels of the
12-bit ADC of the Silabs microcontroller C8051F020. The signals are measured by reading
the ADC channels and the values are stored in the microcontroller. These values are
displayed on the LCD of the microcontroller.

The output signals from the microcontroller are used to determine all the information
related to person’s posture, quality of sleep and so on.
The experiments to test the bed sensor system were conducted with three inhabitants,
one elderly person of 74 years of age and around 42 kg of weight, one 5 year old child
with 20 kg of weight and one adult with 68.5 kg of weight.

Fig. 28: An elderly person lying on bed is being monitored

42
Fig. 29: A child lying on bed is being monitored

The results that we concluded from the experiment are shown in table below. We can see
from these results that under normal situation all the four sensors read very similar
weights. The sensors only measure the weight of the bed in this situation, so each sensor
shows roughly one-fourth of the weight of the bed. When a person lies on the bed the
sensors read different readings depending on the amount of weight shared by that
particular leg of the bed. The total weight measured by the four sensors is roughly equal
to the weight of the bed and the weight of the person. From table 3 it can be easily said
whether the bed is occupied by an elderly or a child from weight consideration. The
position of the head whether the head is at (S1, S2) side or (S3, S4) side can easily be said
by checking the difference of the sensors signals. The pair of sensors with the head side on
them would show higher reading as compared to the other sensors.

43
Test Reading Reading Reading Reading Total
condition S1 S2 S3 S4 weight
(kg) (kg) (kg) (kg) (kg)
Only Bed 10.3 10.4 10.3 10.5 41.2
Elderly in the 18.4 15.2 23.6 26.9 84.1
middle
Elderly on one 22.2 16.5 26.2 19.3 84.2
side
Elderly on 14.0 21.3 18.5 30.5 84.3
another side
Child in the 13.5 14.3 16.9 17.0 61.7
middle
Child on one 14.6 10.5 21.6 15.1 61.8
side
Child on 12.2 13.4 13.7 22.4 61.7
another side
Adult in the 18.7 20.8 36.7 33.8 110
middle
Adult on one 25.2 14.2 45.5 25.3 110.2
side
Adult on 13.7 22.5 25.8 48.1 110.1
another side
Adult lying 17.2 20.5 28.2 44.3 110.2
diagonally
(S4-S1)
Adult lying 21.5 17.4 44.8 26.5 110.2
diagonally
(S3-S2)
Table 3: Experimental results with inhabitant lying on bed

44
The following observations are made with reference to figure 24:

S1 : Signal from Sensors#1.

S2 : Signal from Sensors#2.
S3 : Signal from Sensors#3.
S4 : Signal from Sensors#4.

Savg = (S1 + S2 + S3 +S4)/4; Savg is the average signal.

If (S1 + S2) > 2* Savg; the head is at (S1, S2) side.

If (S3 + S4) > 2* Savg; the head is at (S3, S4) side.

If (S1 + S3) = (S2 + S4); the person is sleeping in the middle of the bed.

If (S1 + S3) > (S2 + S4); the person is sleeping in the right side of the bed.

If (S1 + S3) < (S2 + S4); the person is sleeping in the left side of the bed.

If the sleep quality is good, the four sensors will provide signals which are quite steady. If
the person is suffering from lack of sleep, the sensors will not provide steady signals. The
signals from the sensors can be studied and conclusions can be drawn based on sensors’
information. If the person comes out of the bed at night, the time duration for which the
bed is not used is monitored. The time information can be used for making some decision.
Since the bed sensing system has been configured around a microcontroller, the processor
can do some amount of processing and the conclusions along with the sensors output is
sent to the central controller. The central controller can do another level of processing.

45
8. The Market Potential:

The bed sensor system was designed and fabricated keeping in mind that the end product
should be simple, flexible, reliable, efficient and cost-effective so that and old person
living alone on his pension can afford to buy the system and use it as a stand-alone device.

The Wireless Intelligent Bed Sensor System has a bright future in the market as this device
can prove to be a life saver in some cases. This system will be of great importance to
elderly people living alone at home or people living with disabilities. The system also helps
in avoiding the use of having a care giver or helper constantly monitoring the patient by
being by their side. The caregivers can monitor the subject without actually being at the
house as the system works using wireless technology for transmitting and receiving
information. The system will alarm the caregiver or the person monitoring the sensor
system in case of an emergency or any unusual activity. This feature of the system
provides safety to its users without invading their privacy.

The major components used in the system are Flexi-Force sensors, the driving circuit and
the microcontroller. These sensors are flexible, cost-effective and can easily be integrated.
Our driving circuit does not use any major components so it is also cost-effective to build.
The microcontroller we have used has a rich list of functions and is in an affordable range
for the user.

So, the end product that we have for our Bed Sensor System has all the necessary
elements for it to be successful in the market. The product developed is simple and easy
to use for the user.

46
9. Problems Faced:

Although we reached the goal of having a working prototype of Bed Sensor System at the
end but there were some problems that we faced along the way. One of the things that
took us the most time was the selection of the right components for the driving circuit.
We overcame this problem through trial and error by using various circuit configurations.
We had minor problems like some of our components blew up due to high voltages
provided to the circuit. Since we were using the MAX828 micro chip, we had some
problems in soldering of our components as the circuit used to get shorted.

One of the main problems that we had with our initial circuit was that the output signal
had a lot of noise in it. This noise was affecting our microcontroller readings which weren’t
stable. We overcame this problem by using a low pass filter at the output of the sensors.

Another problem that we had with the initial testing of the characteristics of the Flexi-
Force sensor was that the sensor was very sensitive so the weight that we used for testing
the sensor had to be placed exactly on the same spot on top of the sensor. Slight
unbalance of the weight would result in the sensor readings changing drastically. This
problem was overcome by taking three readings for each weight and taking the average
reading as the final output. We also made a mark on the weight to make sure that the
load is placed on exactly the same spot for all the weights.

Other than these problems we did not have any major problems in this project. The
experimentation was mostly based on trial and error. This was to ensure that we get the
best results possible.

47
10. Conclusions:

To conclude, we can say overall the project has been a success. The main aim of the
project was to have a prototype working Bed Sensor System which has been developed.
This sensor system accurately measures the forces that are applied on the bed. This
intelligent bed sensing system has been designed and fabricated as a part of the smart
home based on wireless sensors.

We first tested our algorithm using a table and the four 25 lbs sensors as we did not have
the 100 lbs sensors available at the time. The results from our table test were satisfying.
The driving circuit gave us good output values from the sensors and the microcontroller
program was also accurate in reading these output values and converting and displaying
them in digital format. The actual results from this table test show that the system was
working fine as the sensor with the most weight on top of it showed the highest reading
on the microcontroller.

After the success of the experiment using the four sensors to measure the weight on a
table we used our circuit and program to conduct tests on an actual bed and a subject. In
this experiment the 100 lbs sensors were used. The results achieved from this experiment
were also as expected. These results gave us a good idea of how our system works. All the
sensors measured the right readings. When the head of the human subject was on the
side with sensor 1 and sensor 2 connected these sensors showed a high value and vice
versa for sensor 3 and sensor 4. Also sensors 1 and 3 showed high readings when a person
was towards the right side of the bed and sensors 2 and 4 read high when the person was
leaning towards the left side of the bed.

From these results we can conclude that the system can estimate the position of the
actual loading point. The response from the sensor system can be used to report normal
sleep, uncomfortable sleep, sitting on the bed at night time or no one being on the bed for

48
a long time. Any abnormality of sleep can be detected and transmitted to the central
controller for generating the warning message to the person monitoring the system. So
this system will be useful in monitoring the bed using the intelligent wireless sensor
system.

49
11. Future Work:
We have been working on further improving our Bed Sensor System in order to make it
more efficient and cost-effective. We have developed a new driving circuit for the system.
This new circuit uses a differential amplifier circuit for the sensor output using a single
supply (0-5V) only. This helps a great deal as we no longer require the use of MAX828
switched capacitor voltage inverter to generate -5V. This also helps us save costs on the
circuit and reduces the number of IC chips used. The new circuit is only based on SMD
(Surface Mounted Device) components which are ultra small components. These reduce
the size of our PCB (Printed Circuit Board). We have made the PCB for this circuit and have
tested it to some extent using the sensors. The results from this circuit are promising. Due
to the lack of time we haven’t been able to test this circuit using a table or an actual bed.
Future work can be done using this circuit in order to achieve a more effective and more
compact driving circuit for our Bed Sensor System.

Improvements can also be made in the ADC output by averaging the output over several
samples to smooth our minor fluctuations in the sensor output voltage (which is generally
due to noise).

The interpolation technique could be extended to incorporate the rate of output from all
the sensors to calculate the position of the load.

The microcontroller can be connected to the PC using UART (Universal Asynchronous

Receiver Transmitter) module (RS232). The microcontroller will pass the data pertaining to
the load position on the bed to a program running on the PC which will depict the
position/orientation of the person (on a simulated bed on the computer screen) in real-
time mode. The data will be logged over an extended period and an inference engine will
deduce the quality of sleep or generate alarm in case of abnormal usage of the bed.

50
12. References:

[1] A. Gaddam, S. C. Mukhopadhyay and G. Sen Gupta, “Wireless Sensors Based Smart
Home to care Elder People”, Proceedings of 2008 Digital signal Processing Creative Design
Contest, pp. 21-28, November 29, 2008, Southern Technological University, Tainan,
Taiwan.

[2] A. Gaddam, S.C. Mukhopadhyay and G. Sen Gupta, Necessity of a Bed Sensor in a
Smart Digital Home to Care for Elder-people, Proceedings of the 2008 IEEE Sensors
conference, Lecce, Italy, October 26-28, 2008, page 1340-1343.

[3] Bowman, R. (2009). Doctors are where patient’s are. Retrieved October 23, 2009, from
http://www.dailyyonder.com/doctors-are-where-patients-arent/2009/02/12/1924

[4] G. Sen Gupta, A. Gaddam and S.C. Mukhopadhyay, Implementation of Wireless

Sensors Based Home Monitoring System, Proceedings of the 2008 IEEE Sensors
conference, Lecce, Italy, October 26-28, 2008, page 1332-1335.

[5] Independent Living Centres Australia. (2009). Bed checking monitoring systems.
Retrieved October 25, 2009, from
http://www.ilcaustralia.org/home/search4.asp?state=WA&page=2&MC=61&MinC=59

[6] Koontz, D. (2009). Life expectancy. Retrieved October 24, 2009, from
http://en.wikipedia.org/wiki/Life_expectancy

[7] Kristoff, S. (2008). An intorduction to sensors. Retrieved October 22, 2009, from
http://engineering.suite101.com/article.cfm/an_introduction_to_sensors

51
[8] Population Division of the Department of Economic and Social Affairs of the United
Nations Secretariat, World Population Prospects: The 2008 Revision,
http://esa.un.org/unpp

[9] S.C.Mukhopadhyay, A. Gaddam and G. Sen Gupta, “A safe, sound and Secured Living
Environment for Elder People: Wireless Sensors based Digital Home”, Proceedings of 2008
IEEE TENCON, October 18-21, 2008, Hyderabad, India.

[10] Tekscan. (2009). FlexiForce force sensors. Retrieved October 25, 2009, from
http://www.tekscan.com/Flexi-Force/Flexi-Force.html

[11] Chew, M. T., & Sen Gupta, G. (2005). Embedded Programming with Field-
Programmable Mixed-Signal uControllers.

52
13. Appendices:

Appendix A:
The schematic of the Driving circuit for all four sensors

53
The waveform output of the driving circuit with and without filter

This is the waveform that we get at the output from the sensors. The bottom line shows
the output before the low pass filter and the top line shows the ouput after the filter has
been used. We can see from this the affect of using a filter to reduce noise in our signal.

54
Appendix B: Data Sheets

55
56
57
58
Appendix C: Microcontroller Code for the Bed Sensor System

//-- Filename : BED_SENSOR_4_FINAL(2).c

//-- Uses Timer 3 Overflow to initiate ADC conversion
//-- Uses the External Crystal oscillator at 22MHz
//-- Measures the Force (analog output) from a Sensor cirucit and
Dsiplays in Digital
//-- format on the LCD
//-- ADC0 interrupt is enabled.
//-- Flashes green LED when SW2 (P3.7) is pressed

#include <c8051f020.h>
#include <stdio.h>

//-----------------------------------------------------------------------
-------------
// Global Defines
//-----------------------------------------------------------------------
-------------

#define LCD_DAT_PORT P6 // LCD is in 8 bit mode

#define LCD_CTRL_PORT P7 // 3 control pins on P7
#define RS_MASK 0x01 // for assessing LCD_CTRL_PORT
#define RW_MASK 0x02
#define E_MASK 0x04

//-----------------------------------------------------------------------
-------------
// Global MACROS
//-----------------------------------------------------------------------
-------------
#define pulse_E();\
small_delay(1);\
LCD_CTRL_PORT = LCD_CTRL_PORT | E_MASK;\
small_delay(1);\
LCD_CTRL_PORT = LCD_CTRL_PORT & ~E_MASK;\

//-- function prototypes ------------------------------------------------

----
void lcd_init (void); // initialize the lcd to 8 bit mode
void lcd_busy_wait (void); // wait until the lcd is no longer
busy
char putchar (char c); // replaces standard function and
uses LCD
void lcd_cmd (char cmd); // write a command to the lcd
controller
void lcd_home (void); // home curser
void lcd_clear (void); // clear display
void lcd_goto (char addr); // move to address addr
void lcd_move_curser(char dist); // moves curser forward or back by
dist
void lcd_curser (bit on); // 1 displays curser, 0 hides it

59
void lcd_puts (char string[]); // send string to lcd at current
curser location
void small_delay (char d); // 8 bit, about 0.34us per count
@22.1MHz
void large_delay (char d); // 16 bit, about 82us per count
@22.1MHz
void huge_delay (char d); // 24 bit, about 22ms per count
@22.1MHz
//-----------------------------------------------------------------------
----

//-----------------------------------------------------------------------
-------------
// 16-bit SFR Definitions for 'F02x
//-----------------------------------------------------------------------
-------------

sfr16 DP = 0x82; // data pointer

sfr16 TMR3RL = 0x92; // Timer3 reload value
sfr16 TMR3 = 0x94; // Timer3 counter
sfr16 ADC0 = 0xbe; // ADC0 data
sfr16 ADC0GT = 0xc4; // ADC0 greater than window
sfr16 ADC0LT = 0xc6; // ADC0 less than window
sfr16 RCAP2 = 0xca; // Timer2 capture/reload
sfr16 T2 = 0xcc; // Timer2
sfr16 RCAP4 = 0xe4; // Timer4 capture/reload
sfr16 T4 = 0xf4; // Timer4
sfr16 DAC0 = 0xd2; // DAC0 data
sfr16 DAC1 = 0xd5; // DAC1 data

//-----------------------------------------------------------------------
-------------
// Global CONSTANTS
//-----------------------------------------------------------------------
-------------

#define SYSCLK 22118450 //-- External Crystal Oscillator @22MHz

unsigned char LCD_count;

int Sensor1, Sensor2, Sensor3, Sensor4; //-- variables to store
ADC0 value
int channel;

//-- function prototypes ------------------------------------------------

----
void Init_Clock(void); //-- initialise the clock to use external crystal
oscillator
void Init_Port(void); //-- Configures the Crossbar and GPIO ports
void Init_ADC0(void); //-- Initialise the ADC1
void Init_Timer3(unsigned int counts);
void Timer3_ISR(void); //-- ISR for Timer 3
void ADC0_ISR(void); //-- ADC0 end-of-conversion ISR
//-----------------------------------------------------------------------
----

60
void main(void)
{
LCD_count = 0;

EA = 0; //-- disable global interrupts

//-- It is a good idea to disable interrupts
before all the initialisation
//-- is complete
//-- disable watchdog timer
WDTCN = 0xDE;
WDTCN = 0xAD;

Init_Clock();
Init_Port();
Init_ADC0();
Init_Timer3(10237); //-- Initialise Timer3 to generate
interrupts every 30ms
lcd_init();
lcd_curser(0);
lcd_clear();

EA = 1; //-- enable global interrupts

P5 = 0x0F; //-- Turn the 4 green LEDs off

while(1) //-- go on forever

{
LCD_count++;
if ( (LCD_count % 100) == 0) //-- do every 100th count
{
LCD_count = 0;

lcd_goto(0x00); //-- First row of LCD

printf("S1:%4d", Sensor1);
printf(" S3:%4d", Sensor3);
lcd_goto(0x40); //-- Second Row of LCD
printf("S2:%4d", Sensor2);
printf(" S4:%4d", Sensor4);
huge_delay(3);
} //-- do every 100th count

} //-- end of while loop

}

void Init_Clock(void)
{
OSCXCN = 0x67; //-- 0110 0111b
//-- External Osc Freq Control Bits (XFCN2-0) set to 111 because
crystal frequency > 6.7 MHz
//-- Crystal Oscillator Mode (XOSCMD2-0) set to 110

// OSCXCN = 0x77; //-- if you want Crsytal Osc. Mode with

divide by 2 stage

//-- wait till XTLVLD pin is set

while ( !(OSCXCN & 0x80) );

61
 OSCICN = 0x88; //-- 1000 1000b
//-- Bit 2 : Internal Osc. disabled (IOSCEN = 0)
//-- Bit 3 : Uses External Oscillator as System Clock (CLKSL = 1)
//-- Bit 7 : Missing Clock Detector Enabled (MSCLKE = 1)
}
void Init_Port(void) //-- Configures the Crossbar and GPIO ports
{
XBR1 = 0x00;
XBR2 = 0x40; //-- Enable Crossbar and weak pull-ups
(globally)
P1MDOUT |= 0x40; //-- Enable P1.6 (LED) as push-pull output

//-- Configure P3.7 for input

P3MDOUT &= 0x7F; //-- Write a logic 0 to set Open-Drain Output
mode
P3 |= 0x80; //-- write a logic 1 to P3.7

//-- Port 7-4 I/O Lines

P74OUT = 0x48; // Output configuration for
P4-7
// (P7[0:3] Push Pull)
- Control Lines for LCD
// (P6 Open-Drain)-
Data Lines for LCD
// (P5[7:4] Push Pull)
- 4 LEDs
// (P5[3:0] Open Drain)
- 4 Push-Button Switches (input)
// (P4 Open Drain) - 8
DIP Switches (input)

//-- Write a logic 1 to those pins which are to be used for input
P5 |= 0x0F;
P4 = 0xFF;
}

//-- Configure Timer3 to auto-reload and generate an interrupt at

interval
//-- specified by <counts> using SYSCLK/12 as its time base.
void Init_Timer3 (unsigned int counts)
{
// TMR3CN = 0x00; //-- Stop Timer3; Clear TF3;
//-- use SYSCLK/12 as timebase
TMR3CN |= 0x02; //-- uses SYSCLK (NOT SYSCLK/12)

TMR3RL = counts; //-- Init reload values

TMR3 = 0xffff; //-- set to reload immediately
// EIE2 &= ~0x01; //-- disable Timer3 interrupts
EIE2 |= 0x01; //-- enable Timer3 interrupts
TMR3CN |= 0x04; //-- start Timer3 by setting TR3 (TMR3CN.2) to
1
}

void Init_ADC0(void)
{
REF0CN = 0x07; //-- Enable internal bias generator and
internal reference buffer

62
 // Select ADC0 reference from VREF0
pin
// ADC0CF = 0x86; //-- SAR0 conversion clock=1.3MHz approx.,
Gain=0.5 (default)
// ADC0CF = 0x80; //-- SAR0 conversion clock=1.3MHz approx.,
Gain=1
ADC0CF = 0x81; //-- SAR0 conversion clock=1.3MHz approx.,
Gain=2

AMX0CF = 0x00; //-- 8 single-ended inputs

AMX0SL = 0x01; //-- Select Sensor1
channel = 0;
ADC0CN = 0x84; //-- enable ADC0, Continuous Tracking Mode
// Conversion initiated on Timer 3
overflow, ADC0 data is right justified

EIE2 |= 0x02; //-- enable ADC Interrupts

}

//-- Interrupt Service Routine

//-- This routine changes the state of the LED whenever Timer3 overflows
every 10th time.
void Timer3_ISR (void) interrupt 14
{
TMR3CN &= ~(0x80); // clear TF3
}

void ADC0_ISR(void) interrupt 15

{
AD0INT = 0; //-- clear ADC0 conversion complete interrupt
flag
if (channel == 0) //-- Sensor1 Reading
{
Sensor1 = ADC0;
AMX0SL = 0x02; //-- Select AIN0.1
channel = 1;
}

else if (channel == 1) //-- Sensor2 Reading

{
Sensor2 = ADC0;
AMX0SL = 0x03; //-- Select AIN0.2
channel = 2;
}
else if (channel == 2) //-- Sensor3 Reading
{
Sensor3 = ADC0;
AMX0SL = 0x04; //-- Select AIN0.3
channel = 3;
}
else if (channel == 3) //-- Sensor4 Reading
{
Sensor4 = ADC0;
AMX0SL = 0x01; //-- Select AIN0.4
channel = 0;
}

63
}

//----------------------------- LCD related Functions -------------------

----------
#pragma OPTIMIZE (7)
void lcd_init(void)
{
LCD_CTRL_PORT = LCD_CTRL_PORT & ~RS_MASK; // RS = 0
LCD_CTRL_PORT = LCD_CTRL_PORT & ~RW_MASK; // RW = 0
LCD_CTRL_PORT = LCD_CTRL_PORT & ~E_MASK; // E = 0
large_delay(200); // 16ms delay

LCD_DAT_PORT = 0x38; // set 8-bit mode

pulse_E();
large_delay(50); // 4.1ms delay

LCD_DAT_PORT = 0x38; // set 8-bit mode

pulse_E();
large_delay(2); // 1.5ms delay

LCD_DAT_PORT = 0x38; // set 8-bit mode

pulse_E();
large_delay(2); // 1.5ms delay

lcd_cmd(0x06); // curser moves right

lcd_clear();
lcd_cmd(0x0E); // display and curser
on
}
#pragma OPTIMIZE (9)

//-----------------------------------------------------------------------
-------------
// lcd_busy_wait
//-----------------------------------------------------------------------
-------------
//
// wait for the busy bit to drop
//
void lcd_busy_wait(void)
{
LCD_DAT_PORT = 0xFF;
LCD_CTRL_PORT = LCD_CTRL_PORT & ~RS_MASK; // RS = 0
LCD_CTRL_PORT = LCD_CTRL_PORT | RW_MASK; // RW = 1
small_delay(1);
LCD_CTRL_PORT = LCD_CTRL_PORT | E_MASK; // E = 1
// TB_GREEN_LED = 1;
do
{ // wait for busy
flag to drop
small_delay(1);
} while ((LCD_DAT_PORT & 0x80) != 0);
// TB_GREEN_LED = 0;
}

64
//-----------------------------------------------------------------------
-------------
// lcd_dat (putchar)
//-----------------------------------------------------------------------
-------------
//
// write a character to the lcd screen
//
char putchar(char dat)
{
lcd_busy_wait();
LCD_CTRL_PORT = LCD_CTRL_PORT | RS_MASK; // RS = 1
LCD_CTRL_PORT = LCD_CTRL_PORT & ~RW_MASK; // RW = 0
LCD_DAT_PORT = dat;
pulse_E();
return 1;
}

//-----------------------------------------------------------------------
-------------
// lcd_cmd
//-----------------------------------------------------------------------
-------------
//
// write a command to the lcd controller
//
void lcd_cmd(char cmd)
{
lcd_busy_wait();
LCD_CTRL_PORT = LCD_CTRL_PORT & ~RS_MASK; // RS = 0
LCD_CTRL_PORT = LCD_CTRL_PORT & ~RW_MASK; // RW = 0
LCD_DAT_PORT = cmd;
pulse_E();
}

//-----------------------------------------------------------------------
-------------
// lcd_goto
//-----------------------------------------------------------------------
-------------
void lcd_goto(char addr)
{
lcd_cmd(addr | 0x80);
}

//-----------------------------------------------------------------------
-------------
// lcd_clear
//-----------------------------------------------------------------------
-------------
void lcd_clear(void)
{
lcd_cmd(0x01); //-- clear LCD display

65
 lcd_cmd(0x80); //-- curser go to 0x00
}

//-----------------------------------------------------------------------
-------------
// lcd_curser
//-----------------------------------------------------------------------
-------------
void lcd_curser(bit on) // 1 displays curser, 0 hides it
{
if (on)
lcd_cmd(0x0E);
else
lcd_cmd(0x0C);
}

//-----------------------------------------------------------------------
-------------
// delay routines
//-----------------------------------------------------------------------
-------------
//
// delay using spin wait
//
void small_delay(char d)
{
while (d--);
}

void large_delay(char d)
{
while (d--)
small_delay(255);
}

void huge_delay(char d)
{
while (d--)
large_delay(255);
}

66