AUTOMATIC FALL DETECTION AND ALERT SYSTEM

A Compact GPS/GSM Enabled Unit based on Accelerometry

Hugo Silva, Filipe Silva, Hugo Gamboa

PLUX – Biosensor Engineering, Av. 5 de Outubro nº 70 – 8º, Lisbon, Portugal

Vítor Viegas

EST – IPS, Setúbal, Portugal

Keywords: Healthcare, Quality of Life, Accelerometry, Biomechanics, Kinematics.

Abstract: Accidental falls are among one of the main causes of death and disability on elderly people. This stands

both as a healthcare problem, in the sense that, upon falling, if individuals are not assisted in an early stage

severe long term consequences may arise; and as a limitation for the individual’s daily life, in the sense that

they generally deprive themselves of regular routines as a preventive measure to avoid falling. In this paper

we describe a hardware unit conceived to automatically detect fall events, and trigger a set of alert actions

which allow the remote detection of the occurrence and facilitate rapid assistance.

1 INTRODUCTION

Population aging is nowadays growing to a

worldwide concern. In the U.S., projections for 2000

pointed a 14,3% population share for senior citizens

of 65+; projections indicate that this number will

grow by 2050 to an impressive 22,6% (U.S. Census

Bureau, 2004), disturbing the balance in the

population structure.

The leading cause of death by injuries, and one

of the main causes of disability in elderly citizens,

are accidental falls (CDC, 2006). Statistics show

that, in the U.S., more than one third of the senior

population with 65+ fall every year with dramatic

outcomes (Hausdorff et al, 2001; Hornbrook et al,

1994). Besides the increase in the number of deaths

resulting from involuntary falls (Stevens, 2006), fall-

related injuries usually have severe mid- and long-

term consequences (Jager et al, 2000; Bell et al,

2000).

Necessary financial and logistics resources to

provide for appropriate nursing services are highly

demanding on healthcare systems; in 2000 direct

medical costs resulting from fatal falls ascended to

$0.2 billion, and $19 billion for the non-fatal case

(Stevens et al., 2006). It is estimated that by 2020

the overall direct and indirect fall-related costs will

surpass $43 billion (Englander et al, 1996).

Furthermore, fall consequences diminish the

individual’s quality of life, standing in many cases

as a long-term limitation to a regular daily life. Fear

of falling leads, among others, to mobility reduction

and fitness degradation, further potentiating the risk

of involuntary falls (British Columbia Ministry of

Health Planning Office of the Provincial Health

Officer, 2004).

Figure 1: Fall detector system architecture.

186

Silva H., Silva F., Gamboa H. and Viegas V. (2008).

AUTOMATIC FALL DETECTION AND ALERT SYSTEM - A Compact GPS/GSM Enabled Unit based on Accelerometry.

In Proceedings of the First International Conference on Biomedical Electronics and Devices, pages 186-189

DOI: 10.5220/0001056401860189

 SciTePress

This is a trend in most developing countries, and

critical issues like the quality of life and healthcare

for senior citizens are rapidly arising, thus making

way for the development of novel, efficient and cost-

effective methods and technologies.

In this paper we present a GPS/GSM enabled

portable hardware unit, designed with the purpose of

automatically detecting involuntary fall events to

alert emergency or nursing teams upon fall detection

(Figure 1). The system facilitates rapid assistance

allowing remote detection of the occurrence, as well

as the GPS coordinates to track the user location.

2 HARDWARE

Recent advances in hardware design and integration

provide appropriate conditions for the development

of compact, miniaturized systems with a wide

application range in real-time signal acquisition and

processing (Silva et al, 2005).

For the fall detector hardware unit, three main

vectors guided the development process: (a)

automation; (b) autonomy; and (c) miniaturization.

With a small form factor, the unit was desgined to

be worn at waist level, measuring 5.2x5.2x1.6cm,

and weighting aproximately 55g.

Figure 2: Fall detector block diagram.

Figure 2 depicts the block diagram for the fall

detector hardware unit, in which there are three main

components which will be detailed next: (a)

accelerometry sensor; (b) MCU; and (c) GPS/GSM

module.

2.1 Accelerometry Sensor

Fall events can be detected using a number of

heuristics; we recur to accelerometry in order to

detect sudden changes in magnitude, and angular

position.

The fall detector hardware unit integrates one

ADXL330 ±3G MEMS

tri-axial accelerometer

which measures the acceleration in the cartesian tri-

dimensional coordinate space.

To maximize the unit’s autonomy, the system is

maintained in an idle mode, only being activated if

the accelerometer readings exceed a pre-defined

magnitude level.

A set of comparators establishes the threshold,

and a hardware interrupt is triggered in the MCU

whenever the threshold is superseded by one of the

axis, switching it to active mode. This works as a

pre-filter for the fall detection algorithm discarding

standard daily life actions as walking, trunk

rotations, among others.

2.2 Microcontroller Unit

An ATMEGA168-20PU MCU is used to command

the hardware unit, maintain runtime data, and

implement the alert and working logic. It has 16KB

program memory, one 16 bit and two 8 bit timers,

serial interface, a built-in 6 channel Analog-to-

Digital Converter (ADC), and two externally

triggered interrupts.

The MCU is kept in idle mode, switching to

active mode from time to time as established by the

watchdog mode previously described in Section 2.1.

Once activated by the accelerometer, the MCU is

switched to active mode starting the real-time

accelerometer output signal acquisition and

processing routine.

If the fall detection software algorithm identifies

a fall event, the MCU enters in alert mode, waking

up the GPS/GSM module to conduct the procedure

described in Section 3.2.

2.3 GPS/GSM Module

GPS/GSM services are assured by a low power,

small form factor, Telit GM862 module. It is

controlled via AT commands from the MCU through

the serial interface, and it has a built-in GPS receiver

and a fully functional GSM device.

Two antennas are necessary in order to guarantee

the Radio Frequency (RF) signal reception for each

function. For this hardware unit we used two low

power consumption, small footprint, PCB antennas,

AUTOMATIC FALL DETECTION AND ALERT SYSTEM - A Compact GPS/GSM Enabled Unit based on

Accelerometry

187

of 13.4x13.4x5.5mm and 42x16x1.6mm, for the

GPS and GSM components respectively.

2.4 Power

The unit is powered by a Li-Ion rechargeable battery

with 3.7V nominal tension, and 2000mAh nominal

current. Table 1 presents the mean power

consumption analysis for the hardware unit. From

Table 1, and taking into account the operating modes

of the unit, an estimated battery lifetime of 72h is

achievable.

The GPS/GSM module described in Section 2.3

has a built-in battery charger, and charge level

indicator. Since we are using a rechargeable battery,

the battery status and charger functions from the

GM862 are also used. Without introducing

additional circuitry we are able to charge the battery

on-board, and assess the battery level during the

watchdog mode previously described in Section 3.1.

Table 1: Fall detector mean power consumption.

Item Mode Consumption

(mA)

Accelerometry

Sensor

Active 0.32

MCU Idle 5.20

Active 9.70

GPS/GSM Idle 4.00

GPS Search 60

Antenna - 13

3 OPERATING MODES

When first connected from a complete power down

state, the hardware unit has a cold start time of <60s.

The Microcontroller Unit (MCU) starts in active

mode and performs the following initialization tasks

(Figure 3): (a) connect the GPS unit; (b) register the

GSM in the carrier network; (c) search the SIM card

memory for the predefined emergency phone

number that should be used while in alert mode; (d)

and store this number in the runtime memory.

During this sequence, a red led is used to

indicate that the unit is in start up mode. When the

initialization tasks are completed, the MCU puts the

GPS/GSM module into sleep mode.

While connected, the hardware unit has two

possible operating states: (a) watchdog mode; and

(b) alert mode. The watchdog mode is the state in

which the unit is normally working, while the alert

mode is the state in which the unit will enter upon

detection of a fall event.

Figure 3: Cold start state diagram.

3.1 Watchdog

This is the mode in which the unit operates until a

fall event is detected.

While in watchdog mode, every 4s the MCU

wakes up to signal the battery status; the activity

indicator led blinks with the colour varying

according to the battery charge level: (a) green,

>50%; (b) yellow, =50%; and (c) red, <50%.

Every 10m the MCU communicates with the

GPS/GSM module in order to update the battery

status and to determine the current GPS location of

the hardware unit. The last known GPS location is

stored in the runtime memory of the MCU.

3.2 Alert

If a potential fall detected by the pre-filter (described

ahead in Section 2.1), the MCU is activated entering

the alert mode.

In this stage the fall detection software algorithm

traces the input signal in order to check if it is a real

fall or a false alarm. If a false alarm is detected, the

unit goes back into watchdog mode. If it is validated

as a real fall, the MCU wakes up the GPS/GSM unit,

a text string is formed containing the date, time,

alarm notice, and last known GPS position, and sent

as a Short Message Service (SMS) text message to

the predefined emergency number loaded to the

runtime memory during the initialization tasks.

Upon failure in sending the alert SMS, the

hardware unit maintains the alert mode. If at first the

message is not sent, the process is retried for three

times. If the message fails sending in all consecutive

tries, the GPS/GSM module is restarted, and the

MCU initiates the alert message procedure once

more. This process is repeated until the hardware

BIODEVICES 2008 - International Conference on Biomedical Electronics and Devices

188

unit is disconnected or until the message is

successfully sent.

4 CONCLUSIONS

This paper describes the implementation of an

automatic fall detection hardware unit (Brown,

2005). Due to its compact size, it is easily worn, and

it does not limit the actions of its bearer.

With the integrated GPS/GSM feature, it allows

the remote detection of fall events and indicates the

last known GPS location of the unit’s bearer,

therefore facilitating the rapid intervention of family

members, emergency or nursing teams in case of fall

(Figure 1).

Most accidental falls occur in contexts in which

subjects are often alone and without means of

calling for aid upon falling; also, involuntary falls

have severe consequences, both in terms of

healthcare and quality of life. Portable fall detection

units are therefore a useful tool which play a major

role, in minimizing the adverse health consequences

of falls and in improving the confidence of fall

victims so that they do not deprive themselves of

their regular activities.

REFERENCES

Bell A., Talbot-Stern J., Hennessy A., 2000.

Characteristics and outcomes of older patients

presenting to the emergency department after a fall: a

retrospective analysis. Medical Journal of Australia;

173(4):176–7.

British Columbia Ministry of Health Planning Office of

the Provincial Health Officer, 2004. Prevention of falls

and injuries among the elderly: A special report from

the office of the provincial health officer. (online;

accessed 2007 Sep 10). Available from URL:

http://www.healthservices.gov.bc.ca/pho/special.html

Brown G., 2005. An accelerometer based fall detector:

Development, experimentation, and analysis.

University of California, Berkeley, SUPERB Technical

Report.

Centers for Disease Control and Prevention, National

Center for Injury Prevention and Control, 2006. Web-

based injury statistics query and reporting system

(WISQARS) (online; accessed 2007 Sep 10).

Available from URL: www.cdc.gov/ncipc/wisqars

Englander F, Hodson T., Terregrossa R., 1996. Economic

dimensions of slip and fall injuries. Journal of

Forensic Science; 41(5):733–46.

Hausdorff J., Rios D., Edelber H., 2001. Gait variability

and fall risk in community-living older adults: a 1-year

prospective study. Archives of Physical Medicine and

Rehabilitation 82(8):1050–6.

Hornbrook M., Stevens V., Wingfield D., Hollis J.,

Greenlick M., Ory M., 1994. Preventing falls among

community-dwelling older persons: results from a

randomized trial. The Gerontologist 34(1):16–23.

Jager T., Weiss H., Coben J., Pepe P., 2000. Traumatic

brain injuries evaluated in U.S. emergency

departments, 1992–1994. Academic Emergency

Medicine; 7(2): 134–40.

Silva H., Gamboa H., Viegas V., and Fred A., 2005.

Wireless physiologic data acquisition platform. In

Proceedings of the 2005 Conference on

Telecommunications.

Stevens J., 2006. Fatalities and injuries from falls among

older adults — United States, 1993–2003 and 2001–

2005. MMWR; 55(45).

Stevens J., Corso P., Finkelstein E., Miller T., 2006. The

costs of fatal and nonfatal falls among older adults.

Injury Prevention; 12:290–5.

U.S. Census Bureau, 2004. U.S. Web-based interim

projections by age, sex, race, and hispanic origin.

(online; accessed 2007 Sep 10). Available from URL:

http://www.census.gov/ipc/www/usinterimproj

AUTOMATIC FALL DETECTION AND ALERT SYSTEM - A Compact GPS/GSM Enabled Unit based on

Accelerometry

189