WIMU: WEARABLE INERTIAL MONITORING UNIT

A MEMS-based Device for Swimming Performance Analysis

Ana S. Silva, Antonio J. Salazar, Miguel F. Correia

Institute for Systems and Computer Engineering of Porto (INESC Porto), Rua Dr. Roberto Frias, 378, Porto, Portugal

Carla M. Borges

Faculdade de Engenharia da Universidade do Porto (FEUP), Rua Dr. Roberto Frias, s/n, Porto, Portugal

Keywords: Wearable, Monitoring device, Biomechanical parameters, MEMS, Swimming analysis.

Abstract: Advances in sensor technology, electronic textile integration, and integrated circuits have introduced a

paradigm shift in the way most researchers approach signal monitoring. In recent years, devices such as

body sensor networks (BSN) allow for direct on-body physiological and biomechanical parameters

measurements. Such technology allows for a more in depth analysis of an athlete’s performance, without

affecting the results due to awkward wires or uncomfortable carry-on devices. Miniaturization and other

achievements allow a more seamless interaction with the individual, permitting a more natural behaviour

during the monitoring session. The project BIOSWIM (Body Interface System based on Wearable

Integration Monitorization) is a joint multidisciplinary effort of a number of Portuguese universities which

seeks a pervasive monitoring solution for performance, physiological and biomechanical signals from a

swimmer under normal training conditions. In order to achieve such an undertaking a swimsuit prototype

was developed with truly integrated EKG textile sensors; which will work in conjunction with a wearable

inertial monitoring unit (WIMU) and a wearable chemical monitoring unit. This article focuses on the

WIMU, which serves as the biomechanical data processing unit of the system.

1 INTRODUCTION

In order to evaluate an athlete’s performance,

coaches and trainers have been dependent on visual

markers and inferred biomechanical data for years.

Nowadays, the development of miniaturized low-

powered accelerometers and gyroscopes allow their

inclusion on wearable monitoring alternatives,

closing the gap between the athlete’s perception and

that of his/her trainers. In fact recent advances, such

as micro-electromechanical systems (MEMS), have

introduced a viable alternative, which unlike many

of its counterparts; can be made to be wearable and

water friendly, as to be used on swimming analysis.

Electronic textiles, interactive textiles, smart

garments and wearables are terms expressing one of

the most recent research trends in textile science and

technology. From late 1990’s until now investigators

from research centers and companies proposed

several approaches and solutions for many different

problems, from technical applications to leisure.

While the fastest developments started with

multimedia applications, very soon the scientific

community understood the potential of these

technologies on other areas, like in medicine and

sports. Literature describes a large number of

projects intended to remotely measure physiological

parameters on mainly heart failure patients, such as

MyHeart and Wealthy projects (Pacelli et al., 2006),

VTAMN (Noury et. al, 2004) and more recently

BIOTEX (Coyle et al., 2010) and Vital Jacket

(Cunha et al., 2010).

Although the wearable inertial monitoring unit

(WIMU) is still at a prototype stage, preliminary

data reveals the promise of such a device.

2 BIOSWIM PROJECT

The BIOSWIM project is inspired in all these

emerging technologies and intends to apply them in

a smart garment which can allow the measurement

S. Silva A., J. Salazar A., F. Correia M. and M. Borges C..

WIMU: WEARABLE INERTIAL MONITORING UNIT - A MEMS-based Device for Swimming Performance Analysis.

DOI: 10.5220/0003172700870093

In Proceedings of the International Conference on Biomedical Electronics and Devices (BIODEVICES-2011), pages 87-93

ISBN: 978-989-8425-37-9

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

of vital signs and other significant signals both for

health and particularly for sport athletes, thus

contributing for more knowledge in these areas. The

wearable equipment combining comfort properties

together with sensing devices will fit on an

individual in an optimal way without the need of

using unpleasant methods for fixing sensors on the

athletes. Electromyography, cardiac rhythm,

respiratory effort, oxygen consuming taxes, motion

capture, wrist and arm accelerations and rotations,

speed, hydrodynamics’ pressure constitute some of

the parameters of interest of this research project and

intended to be collected at simultaneous time using

the same wearable, thus consisting in a major

breakthrough in the sport sciences (see Figure 1).

This project is a multidisciplinary effort of three

major Portuguese institutions: Centre of Textile

Science and Technology of the University of Minho,

INESC Porto (Institute for Systems and Computers

Engineering of Porto) and the Faculty of Sports of

the University of Porto.

Figure 1: BIOSWIM project's monitoring objectives.

The main goals of the BIOSWIM are the

development of wearable prototypes for measuring

several biological parameters, critical for sport

performance and health purposes, oriented for

regular sport and for in-pool (or underwater

depending of the sports) activity. Moreover, the

wearable must be design with maximum comfort,

comprehending the best combination of textile

material with embedded microelectronics for signal

acquisition and data transmission. Also, an universal

electronic system for power supply, data acquisition,

processing, management, and communication

working for both inside and outside water

environments and embedded on textile substrate is

expected to be developed.

An immediate application of this wearable is in

sports, for competition athletes, sport amateurs, and

body maintenance. While for the less demanding

activities some of the actual systems provide the

essential information, the same may not apply when

it is intended to evaluate performance on athletes.

The main goal in this situation is the improvement of

their performance in order to achieve better results.

Proper movement inside water will definitely

contribute for higher speed. Beyond vital signs,

other signals are crucial for optimizing the athlete

effort and thus reduce the human fatigue when

subjected to intensive training or competition

charges.

The development of wearable monitoring

devices for sports practitioners can present several

noticeable repercussions in sportive community,

both with reference to the optimization of the

training process of elite athletes, as well as to the

promotion of safety in rehabilitation and leisure

sports. The idea behind the BIOSWIM project is to

produce an elite training evaluation wearable station,

allowing the most innovative monitoring,

ambulatory registration, real-time visualization and

post-exercise display of both physiological and

biomechanical relevant data for training (heart-rate,

respiratory frequency, oesophageal temperature,

sweat, arm tri-axial acceleration, body vertical

acceleration).

Important repercussions are expected both for

practitioners, coaches and scientists, allowing an

increased safety in physical activity, an augmented

objectivity – and efficiency - of the elite training

process, and an easy data collection for scientific

research in neuro-physiology and biomechanics of

sport.

3 ARCHITECTURE OVERVIEW

Within this article, monitoring systems will be

considered divided in three main sections:

• Sensing section.

• Processing section.

• Transmitting section.

BIODEVICES 2011 - International Conference on Biomedical Electronics and Devices

The mentioned sections can be found separate,

intermixed or integrated depending on the design,

but the objectives of their functions can be readily

separated if needs be.

3.1 Sensing Section

Advances in micro-fluidics, material science, nano-

structures, micro-electromechanical devices,

bioelectrical interfaces, and others; have contributed

to a new generation of wearable and implantable

sensors and monitoring devices. Healthcare has

greatly benefitted from the development of

biosensors (also referred to as chemical sensors)

(Patel et al., 2006) and physiological sensors. Such

achievements have paved the way for truly pervasive

monitoring strategies, which will benefit patients

and reduce the load to health-care facilities. From a

sport monitoring perspective, non-invasive,

minimally intrusive sensors are the preferred choice,

and consideration of their positioning, calibration,

noise, offset, deviation, etc., are concerns (Yang and

Shouqian, 2007). There exist a wide array of

commercially available sensors and even more

experimental devices and concepts waiting their

turn.

3.2 Processing Section

In today’s market, the competition to claim to be the

lowest powered microcontroller is fierce. Depending

of the complexity required by the application and the

feature extraction methods to be applied, an array of

Reduced Instruction Set Computer (RISC) or

Advanced RISC Machine (ARM) architecture based

microcontrollers offer different features which

accommodate varying solutions. Based on the

popular “motes” designs and further research in the

area (unpublished work from the authors), there

seems to be a preference for Texas Instrument

MSP430 ultra-low power, Atmel’s ATMEGA ultra-

low power, and the Microchip’s extreme-low-power

(XLP) PIC microcontrollers. Using the power

specifications, indicated on the datasheet of each

microcontroller, as a base for comparison, can

sometimes lead to problems and confusion; careful

attention must be paid to the conditions in which

each manufacturer measures their devices power

consumption.

3.3 Transmitting Section

When referring to WMS, it is unavoidable to

consider a wireless component for interfacing with

the system; either be it for real-time (or continuous)

or sporadic updating to a remote processing node, or

for downloading the collected stored data, or even

for transmitting the data from a sensor node to the

on-body or remote processing unit. The presence of

cables or the need for physical removal of the device

for data download represents an alternative that

while permissible at prototyping and troubleshooting

stages, is impractical at more advance stages of

design and implementation.

A number of alternatives exist for mid-range

wireless communication including common

protocols (GSM, WiMAX, UMTS, WLAN, etc.) and

upcoming 4G mobile communication solutions.

From a more local point of view the IEEE 802.15

Workgroup has introduced and arrays of solutions.

Among the favorite standards one counts with the

IEEE 802.15.1, known as Bluetooth, and the IEEE

802.15.4, also referred to as Zigbee. The number of

low-power short-range transceivers in the market

today is enough to overwhelm even experienced

researchers. It seems every brand offers their

particular RF solution, claiming low-power

transmission; companies such as Texas Instrument,

Atmel, Semtech, Maxim and Microchip (to mention

a few), offer interesting and varying solutions.

4 WIMU

For years, numerous devices and setups have been

implemented in order to assist on swimming

performance analysis. Many of these devices were

based on video analysis, while others made direct

measurement and signal capturing through awkward

setups, generally uncomfortable for the swimmer

and thus affecting hers/his performance. Advances

in a number of fields have allowed for compact

wearable monitoring devices, greatly improving the

data gathering process and closing the gap for a truly

seamless biomechanical signal monitoring solution.

Although there is a relatively reduced number of

biomechanical signal monitoring systems being used

for swimming performance analysis today

(particularly when compared to the number of

wearable monitoring devices for healthcare or even

for land based sports), a shift on the approaches for

swimming analysis is being noted. Different

strategies have been applied by the mentioned

systems, however a common element seem to be

their dependence on accelerometers. Some systems

worth mentioning are the ETH Zurich Wearable

Computing Laboratory’s SwimMaster (Bächlin et

WIMU: WEARABLE INERTIAL MONITORING UNIT - A MEMS-based Device for Swimming Performance Analysis

al., 2009) and Imperial College BSN device (Pansiot

et al., 2010).

4.1 Design Overview

The WIMU – Wearable Inertial Measurement Unit –

was designed in the scope of the BIOSWIM project.

The goal was to develop a MEMS-based wearable

device for assessing biomechanical parameters of a

swimming athlete. The inertial unit comprises a tri-

axial accelerometer for linear acceleration and a

gyroscope for angular velocity measurements. In

addition, a microcontroller for signal acquisition,

conversion and wireless transmission is used. A 3D

schematic and a picture of the overall system are

depicted in Figure 2.

(a)

(b)

Figure 2: (a) Tri-dimensional view of the WIMU. (b) Top

view of the system.

Both inertial sensors used are MEMS devices.

Micromachined inertial sensors have a huge

potential for applications in the biomedical field.

They fulfill the requirements of small volume,

portability, low power consumption and they are

available at low cost

To extract the acceleration value, the

accelerometer has a movable mass which is

connected to a fixed frame via spring structures. An

external acceleration will displace the mass from its

rest position, proportionately to the input movement.

The micromachined gyroscope used relies on a

mechanical structure that is driven into resonance

and excites a secondary oscillation in either the same

structure or in a second one, due to Coriolis force.

The amplitude of this secondary oscillation is

directly proportional to the angular rate signal to be

measured.

Commercially available evaluations boards were

used to integrate the MEMS sensors into the WIMU.

The accelerometer evaluation board is a small circuit

board intended to be used for evaluating the

MMA7260QT accelerometer and developing

prototypes quickly without requiring a PCB to be

designed. It also provides means for understanding

the best mounting position and location of this

accelerometer. Similarly, the IDG-300 gyroscope is

integrated in an evaluation board along with the

electronics necessary for application-ready

functionality

In order to integrate both the accelerometer and

the gyroscope into the inertial unit, a microcontroller

was selected from the many commercially available

devices. The one chosen is the eZ430-RF2500

development tool from Texas Instrument. The

eZ430-RF2500 uses the MSP430F2274

microcontroller which combines 16-MIPS

performance with a 200-ksps 10-bit ADC and 2 op-

amps and is paired with the CC2500 multi-channel

radio-frequency (RF) transceiver designed for low-

power wireless applications. This board is a

complete wireless development tool that includes all

the hardware and software required to develop an

entire wireless project. One reason for choosing this

board is its low-power consumption characteristic.

The wireless communication uses the SimpliciTI

protocol which belongs to Texas Instruments and is

a low-power RF (2.4 GHz) protocol aimed for

simple and small RF networks. It was designed for

easy implementation with minimal microcontroller

resource requirements.

The WIMU behaves like an End Device in a

sensor network topology. It is responsible for

acquisition and conversion of the sensor signals and

posterior wireless transmission to a remote station

for post-processing and analysis. It was designed so

that multiple WIMUs could be used in different

body segments in a truly body sensor network.

4.2 WIMU Architecture

As mentioned above, the WIMU consists of a 3-

axial accelerometer, a gyroscope, a microcontroller

and a power supply unit. The architecture of the

WIMU is represented in Figure 3.

The acceleration and angular velocity signals are

acquired and converted sequentially by the 10-bit

ADC integrated in the MSP430F2274 micro-

controller at a sampling frequency of approximately

50 Ksps. A time-stamp is then retrieved from the

BIODEVICES 2011 - International Conference on Biomedical Electronics and Devices

system and added to a sensor packet containing the

3-axis accelerations and angular velocity. Therefore,

the final packet frame consists of sensor signals

together with a time-stamp, which is then ready to

be sent by the CC2500 radio frequency to a remote

station where the sensor data can be analysed and

processed.

Figure 3: WIMU architecture.

The WIMU weights approximately 65.6 grams

and measures 57x90.5x24 mm. This way the device

can unobtrusively be worn by the swimmer without

affecting his/her performance during swimming.

With two AA batteries the system can operate

continuously up to four hours.

5 EXPERIMENTAL RESULTS

In this study, the sensor was placed on the upper

back of the swimmer as can be seen on Figure 4.

Before entering the pool, the swimmer was asked to

perform a number of flexibility related movements

routines in order to determine if she felt movement

constraints. Once in the pool, she was asking to

swim, submerge and perform various movements in

order to asses if the WIMU’s presence represented

an obstruction to her movements. In both cases

(outside and inside the pool) the swimmer reported

that the unit did not affect her movements in any

manner. Several more fast submersions and from

wall impulses were performed as to determine the

waterproofness of the package, proving to function

adequately. Finally, the swimmer was told to

complete several sets of laps using free-style (i.e.

crawl) technique, then a number of laps with

breaststroke, and finally butterfly. For all styles

indicated the end of pool turn was perform through a

vertical turnaround (i.e., stop-touch wall-

turnaround), reversing direction without flipping

under water (in order to avoid signal loss).

(a)

(b)

Figure 4: (a) WIMU positioned at the upper back of the

athlete. (b) The athlete swimming with the WIMU.

The signals are captured and buffered, and

although there is the potential for signal conditioning

at this point, the raw data is sent to the base station

as is. Synchronization between the base station and

the WIMU allow for missed data marking, in order

to apply signal restoration strategies. Such data is

compensated through cubic interpolation when

necessary and then smoothed by a convolution

approach. At this point, the crawl technique requires

of no compensation (i.e., interpolation) for missed

data points since no link loss occurs for a significant

period (except for horizontal flip at the end of a pool

lap). Although the butterfly and breaststroke

technique do present gaps due to momentary loss of

signal they do not represent a significant number of

points as to deteriorate the signals core; some

resulting data can be observed on Figures 5 and 6.

Observing the graphs concerning each swimming

technique, it was possible to differentiate between

styles, each stroke, and end of pool. For example,

WIMU: WEARABLE INERTIAL MONITORING UNIT - A MEMS-based Device for Swimming Performance Analysis

Figure 5: Two laps using Crawl technique signals.

Figure 6: One lap captured signals for each performed technique.

free style can be readily distinguish from

breaststroke and butterfly by the signal provided by

the accelerometer in Y-axis. The referred signal

presents large variations for the crawl technique,

while for the other presented styles these variations

are comparatively small. Alternatively, the butterfly

and breaststroke technique can be differentiate from

the data produced by the accelerometers in the X and

Z axes; which in the case of the butterfly samples

they are both almost double the frequency when

compared to the angular velocity. In the meanwhile

the breaststroke samples show all these three signals

at nearly equal frequency.

BIODEVICES 2011 - International Conference on Biomedical Electronics and Devices

It is still early in this project to produce

conclusive results regarding feature extraction for

performance analysis, and there is a clear need for

accumulating data of diverse swimmers of different

genders and competitive level, however these first

steps are quite promising. A more comprehensive

analysis of the data gathered up to now and future

collections, as well as feature extraction strategies

for performance analysis will be presented in future

works.

6 CONCLUSIONS AND FUTURE

WORK

In this paper, we present a prototype for swimming

performance analysis monitoring based on

accelerometers and gyroscopes, referred to as

WIMU. The WIMU is one of the objectives of a

much more ambitious multi-disciplinary effort of a

group of Portuguese universities known as the

BIOSWIM project. Such project seeks to

characterize swimming through physiological and

biomechanical signal capturing at points distributed

throughout the athlete’s body. Although the device

presented is still at the initial stages of development

it was capable of providing promising data under in-

pool normal conditions. The current version of the

WIMU serves as a basis for future implementation

that will focus on wearability, energy harvesting,

and integration within the BIOSWIM’s project

swimsuits.

ACKNOWLEDGEMENTS

The authors would like to acknowledge the

Foundation for Science and Technology of Portugal

for their support of the BIOSWIM project

(PTDC/EEA-ELC/70803/2006) and of some of the

PhD students involved in this article

(SFRH/BD/61396/2009 and SFRH/BD/60929

/2009). Additionally the authors would like to

acknowledge the contribution of Barbara Mota, main

testing swimmer.

REFERENCES

Bächlin, M., Förster, K. and Tröster, G. (2009).

SwimMaster: a wearable assistant for swimmer. In

International Conference on Ubiquitous

Computing. ACM. pp. 215-224.

Coyle, S et al. (2010). BIOTEX - Biosensing Textiles for

Personalised Healthcare Management. In IEEE

Transactions on Information Technology in

Biomedicine 14(2): 364-370.

Cunha, J. P. et al. (2010). Vital-Jacket: A wearable

wireless vital signs monitor for patients' mobility in

cardiology and sports. In 4

International Conference

on Pervasive Computing Technologies for Healthcare

(PervasiveHealth).

Noury, N. et al. (2004). VTAMN - A Smart Clothe for

Ambulatory Remote Monitoring of Physiological

Parameters and Activity. In 26th Annual International

Conference of the IEEE Engineering in Medicine and

Biology Society, IEMBS '04. IEEE Conference.

Pacelli, M. et al. (2006). Sensing Fabrics for Monitoring

Physiological and Biomechanical Variables: E-textile

solutions. In 3

IEEE/EMBS International Summer

School on Medical Devices and Biosensors. IEEE

Conference.

Pansiot, J., Lo, B. and Yang, G. Z. (2010). Swimming

Stroke Kinematic Analysis with BSN. In 2010

International Conference on Body Sensor Networks.

IEEE Conference. pp. 153–158.

Patel, B., Anastassiou, C. and O’Hare, D. (2006).

Biosensor Design and Interfacing. In Body Sensor

Networks. G. Z. Yang, Ed. London: Springer. pp. 41-

87.

Yang, L. and Shouqian, S. (2007). Smart Sport Underwear

Design. In 7th International Conference on Computer

Aided Industrial Design & Conceptual Design. pp. 1-

WIMU: WEARABLE INERTIAL MONITORING UNIT - A MEMS-based Device for Swimming Performance Analysis