Ambient Intelligence Systems for Personalized Sport …ait.upct.es/deportes/publicaciones/Ambient Intelligence Systems for... · Ambient Intelligence Systems for Personalized Sport

Submitted to Sensors. Pages 1 - 21.

OPEN ACCESS

sensors

ISSN 1424-8220www.mdpi.com/journal/sensors

Article

Ambient Intelligence Systems for Personalized Sport Training

J. Vales-Alonso1,�, P. Lopez-Matencio1, F. J. Gonzalez-Castano2, H. Navarro-Hellın1, P. J.

Banos-Guirao1, F. J. Perez-Martınez1, R.P. Martınez-Alvarez3, D. Gonzalez-Jimenez3, F.Gil-Castineira2, R. Duro-Fernandez4

1 Universidad Politecnica de Cartagena, Campus Muralla del Mar. Cartagena, Spain2 Universidad de Vigo, Campus Lagoas-Marcosende. Vigo, Spain3 Gradiant, Campus Lagoas-Marcosende. Vigo, Spain4 Universidad de A Coruna, Rua Maestranza. A Coruna, Spain

� Author to whom correspondence should be addressed; Email: [email protected]

Version November 16, 2009 submitted to Sensors. Typesetted by LATEX using class file mdpi.cls

Abstract: Wireless sensor networks have increasingly grown in the last years, supporting the

concepts of contextual services and ambient intelligence, and enabling new applications at

different areas. Nowadays, several Spanish Information Technologies research programs are

tackling the use of WSN in specific scenarios, such as e-Health, e-Inclusion or e-Sport. This is

the case of the project “Ambient Intelligence Systems Support for Athletes with Specific Pro-

files”, which intends to assist sportsmen with different performance profiles in their training.

In this paper, we describe the main developments and outcomes from this project. The archi-

tecture of the system comprises a WSN deployed in the training area which provides bidirec-

tional communication with athletes’ mobile equipments, performs location tasks, and gathers

environmental variables (humidity, wind speed, temperature, ligth intensity, etc.). Users are

equipped with a monitoring unit which obtain data from the athletes’ training (pulse, oxime-

try, speed). Both sets of data are then used by a decision engine to direct users’ training. An

implementation of this model is presented in this work for a particular scenario where cross

runners are training on a net of tracks with different difficulty. The decision engine, imple-

mented by means of an interpolation algorithm based on (m, s)-splines, heads the runners

towards the most suitable track in each moment to achieve their training goals.

Keywords: Ambient Intelligence; Contextual Services; Wireless Sensor Networks; Sport

Training; Support Vector Machine

Version November 16, 2009 submitted to Sensors 2 of 21

1. Introduction

Wireless sensor networks (WSNs) link information technology with the physical world. Currently,

WSNs are used in a broad range of applications: industrial process control, ecosystem monitoring or

health monitoring among others. Nevertheless, the application areas are constantly growing. It has been

stated that WSNs will provide intelligence to everything around us [1]. As sensor devices shrink, and

their computing and communication capabilities increase, WSNs will get seamlessly integrated into the

physical world (they are transparent to the users, in the sense that the users only perceive the output of

the interfaces).

WSNs may be sensitive and responsive to the presence of people. Depending on the sensed data

(e.g. biometrics, weather conditions, etc.), network devices can situate the user in a context, enabling ap-

plications that react to their changing environments, such as context-aware personalized services. How-

ever, there are some pending research challenges, regarding hardware requirements, service accessibility,

decision-making intelligence, human interfaces, system architectures, security, reliability, etc.

To some extent, the emergence of these networked systems of embedded systems is simply a natural

evolution of the historical trend in computing and communications technologies, which are already of

use in different fields. As an example, advanced sports training is often assisted by more or less complex

computing devices carried by the sportsmen, which provide useful telemetry about their biometrics and

practice related events. Among other parameters, they can monitor heart rate, track routes, measure

speeds and distances, and so forth [2]. All these data are collected while training and recorded in a wrist

or arm unit, to be analyzed by an external software, after each training session. This approach seems

static in some sense, because the external software does not adapt itself to the sportsmen progress, nor it

provides any advice while training.

Supported by the Spanish Ministry of Education, Culture and Sports [3], project Ambient intelligence

systems to assist sportsmen with specific profiles, pursues to provide personalized assistance to sportsmen

in their training.

As shown in Figure 1, there are three driving aspects in this project: sportsmen profile, environment,

and computing and communications technologies. Technology will provide adaptive coordination be-

tween the user and the environment, adapting the behavior of the system as it responds to changes in the

environment or the user conditions, such as weather or sportsmen location. Since training conditions can

rapidly vary, the system will be expected to take real time decisions to meet the user needs at their full

potential. Thus, a decision engine, as well as efficient communications and localization protocols, are

two key working areas in the project. This article presents its main advances so far.

The rest of the paper is organized as follows: Section 2. discusses related work. Section 3. presents

the system architecture, its components and the communication protocols. Sections 4. and section 5.

describe a system prototype and the methodology used for the implementation of the decision engine,

respectively. A case of study is analyzed in Section 6.. Finally, section 7. concludes the paper.

2. Related work

As described in [4], the use of chronometers, photocells, contact platforms, microphones, photo or

video cameras, magnetic resonance and X-rays machines, movement sensors and, in summary, every


Figure 1. General vision of ambient intelligence aimed at Sports

sensing device for physical or chemical parameters, is nowadays common in athletes’ training. Many

of these devices are not exclusive for elite sportsmen, but available to the general public (e.g. heart

rate monitors). They are used for training monitoring, by sportsmen themselves or by their coaches, to

improve physical performance.

These availability of these devices was the first step towards contextual services, as mentioned in

Section 1.. Next, sport feedback systems introduced real-time user data collection and localization.

Communications capabilities are provided by standard wireless network protocols (IEEE 802.11, Blue-

tooth/Wibree, Zigbee, etc.). Either these networks or specific ones (UWB, RFID, GPS, etc.) may support

location [5].

Future developments aim at expanding the range of monitored data (environmental conditions, more

rich user data), and providing useful actions and information based on them. WSNs [6] represent one of

the enabling technologies for that evolution. Several context-aware applications for sportsmen training

have already been introduced. For example, in [7], the authors propose a feedback system for elite sports

training. They demonstrate its functionalities with use cases in table tennis, biathlon and rowing. Their

work shows how sensors may expand available data to study exercise execution. In the case of table

tennis, the system detects impacts of the ball on the table, computing its position by means of vibration

triangulation, to display throw accuracy. In the biathlon case, a laser positioning system analyzes the

motion of the rifle barrel before and after the shot. Finally, in the rowing case, the training system

calculates the effort that is applied to the oar. This system lacks flexibility, because it depends on the

sport class, and it is not adequate for outdoor training. In the positive side, sportsmen movements are not

restricted, because they do not need to carry any electronic device. Nevertheless, the training scenario is

always supervised and controlled. Thus, the sportsman may perceive the system as intrusive.

In MarathonNet [8] a WSN monitors runners in marathon events. Sensors on runners collect data

about heart rate, time and location. These data are sent to a central database via base stations along


the track, where they are off-line analyzed. Base stations can communicate with the central database by

means of GPRS, WLAN, or a wired network link. The sensor nodes in our work have similar function-

ality, but they also act as information routers.

The system in [9] assists professional skiers. Using accelerometers and force-sensing resistors, skiers

can obtain data about their movements and visualize them, along with video footage, once the exercise

has finished. The research in [10] investigates the combination of two data sources, video and sensed

parameters, to improve the performance of inexpert and elite athletes. This system allows to compare

and evaluate different athletes’ actions and performance, identifying advantages and disadvantages. By

watching recorded video, a coach may give customized feedback to each skier.

Major competition events, like the Olympic games, typically pioneer the adoption if new technologies.

WSN can help referees in their tasks. As an example, in [12], the author introduces the SensorHogu

systems, which embeds piezoelectric force sensors on body protectors, to recognize valid scoring kicks

in Taekwondo.

None of the previous works consider real-time feed-back. In other words, a human -the coach- carries

this task after interpreting sensed data. Some systems aim at simplifying this manual task: The com-

mercial product Team2Pro [11] allows to record and study fitness data in real time for up to 28 players.

Nevertheless, without human intervention the previous system is useless.

A system that aims at providing automatic real-time feedback is described in [13]. The system moni-

tors dynamic data from cyclists and their bicycles. Then, it compares those sensed data with a predefined

set of values to adjust training parameters. For instance, the system can advise the group to change the

formation, split, or to increase or decrease the speed. However, decisions are based on simple rules based

on thresholds on monitored variables. Besides, environmental conditions are not considered (e.g. wind

speed may be critical in cycling).

To sum up, most previous systems do not sense environmental data, but only from sportsmen, and they

only provide limited real-time feedback in some cases. Our proposal tries to overcome such limitations,

providing useful automated feedback to athletes based on an ambient intelligent system, inferring the

behavior of the athlete under changing training conditions.

Our prototype calculates an estimate of the athlete’s heart rate (HR) as the exercise endurance or

the environmental conditions changes. The relationship between HR and sports activity is affected by

many factors, such as temperature, age, sex, mental stress, etc. The effect of ambient temperature in cy-

clist’s HR was studied in [14], which concluded that temperature can be a hardening factor for training.

Analysis of HR dynamics by methods based on chaos theory and nonlinear system theory have received

attention recently, due to observations that suggest that the mechanisms involved in cardiovascular reg-

ulation likely interact with each other nonlinearly [15, 16, 17]. Thus, characterizing HR and athlete

performance is even more difficult.

There exist several numerical and statistical methods to determine the relationships hidden in com-

plex processes where many variables can be involved. These methods have been utilized to predict

physiological parameters, such as HR, systolic blood pressure, or body temperature.

Support vector machines (SVM) have been successfully applied to medical decision support. Refer-

ence [18] describes how to mine medical knowledge from time series of high-dimensional numerical data

describing patients in intensive care. A SVM is used to learn how and when a drug dose must change.


Also, in [19], it is proposed to improve the diagnosis of tuberculosis infections by means of SVM image

classification. Nevertheless, for sportsmen training purposes, a SVM statistical method requires a huge

amount of time series data of many training sessions, in order to feed the knowledge base of each athlete

appropriately. Therefore, the approach is impractical.

Based on HR and 3D acceleration signals, preprocessed as time sequences, the strategy in [20] em-

ploys Feed-Forward Neural Networks (FFNN) to predict the next time step in the HR sequence. The

predicted HR follows the variance of the real HR, although with noticeable differences in some cases.

The work in this paper is based on numerical approximation techniques based on splines, which are

extensively applied in signal processing and surface fitting areas [21, 22, 23]. Among the different

spline techniques, we selected (m, s)-splines [24] for their favorable characteristics to our research: they

allow to face multi-variable problems, the problem domain is not required to be a mesh grid, and the

computational load is low.

3. System architecture

Our goal was designing a general architecture, to support sportsman and environmental monitoring,

in order to provide real-time feedback for training improvement. The infrastructure is mainly oriented

towards open-field sports, such as running, cycling or skying. In these scenarios, sportsman performance

depends not only on physical conditions, but also heavily on terrain conditions (slope, temperature, wind,

fog, etc.). In indoor sports environmental sensing is less critical and sportsman sensing becomes highly

specialized, as described in Section 2.. Therefore, our approach is less applicable.

Two types of data sets are combined to characterize user performance:

• A static set, which provides information that does not change along training, from the environment

(e.g. terrain slopes, intrinsic exercise, pathway configuration, sensing nodes position, etc.) and the

user (e.g. performance profile, age, skill, training goals, etc.).

• A real-time set, with updated environment information (temperature, wind, visibility, etc.) and

sportsman data (heart rate, body movement, elapsed training time, etc.).

A control element processes these data and issues commands to direct user exercise. Section 6.

describes an implementation of these elements for a realistic case study. The rest of this section describes

the architecture and protocols of the system.

3.1. System components

Figure 2 shows a typical system deployment. It consists of the following elements:

• Infrastructure Nodes (IN), which cover the training area. These nodes sense environmental vari-

ables and relay data to/from the User Equipment or other IN to the Control Node. Communication

relies on a wireless link. In addition, IN nodes also serve as an auxiliary network to perform sports-

men location in the training area. An IN is composed by a processing unit, a wireless interface

including directional communication antennas, and a power supply. Directional antennas increase

the communication range between IN nodes, so that less nodes are necessary, thus reducing net-

work installation costs.


Figure 2. Data flow of the sensed and localization information, and response from CN.

(a) Environmental and sportsmen data acquisition (b) Location procedure

(c) Decision transmission

• User equipment (UE). This is the device carried by the user, which includes a wireless transceiver

to communicate with the IN, and sensing elements to monitor sportsman physical parameters. The

UE also includes one or more user interfaces (acoustic or visual) to deliver training orders. The

UE has the same functional blocks as an IN but, as its weight and volume are constrained, memory

and processing power are lower and the antennas are less directional.

• Control Node (CN). It can be considered as a particular IN, which collects data from all IN nodes

and the UE, and analyzes them to assist user training. CN processing requires access to the static

information set, which is stored in databases or standard files.

3.2. Network deployment and communication protocols

The scenario described in the previous section is a WSN with fixed sensing elements (IN) and mobile

ones (UE). As in a typical WSN, energy constraints are demanding for these elements, since they usually

run unattended. Thus, power saving is a mandatory requisite of protocol design. In order to reduce the

burden of protocol signaling, it is assumed that the IN nodes are arranged in a tree topology. The

normal disposition cross-country tracks, which likely includes crossings and track lines, also suggests

a topology of this type. Therefore, the protocol routes remain unchanged once they are calculated and,


Figure 3. State machine diagrams.

(a) State machine of Infrastructure Node (b) State machine of User Equipment

thus, they can be considered static. For instance, the routing protocol may run at network startup, relying

on some level discovery protocol (e.g. [26]), and reducing the packets necessary for routing tasks. Then,

regular INs can deliver data to the CN just sending packets to their parent nodes, or to other INs or UEs

forwarding packets to their leaf nodes. In addition, detailed coverage planning has to be calculated for

the IN positions to guarantee network connectivity. For network planning purposes, there exists some

specialized tools, but they usually ignore practice-related considerations (for instance, if wind sensing is

interesting, some open ground IN positions may be desirable). An optimization method to determine the

best sites and the connectivity level may be necessary (e.g., [27]).

Note that communication ranges differ regarding IN-IN and IN-UE links, since UE antennas are less

directional than IN ones (see Section 4. for details). Low directivity antennas reduce the communication

range, and precise aiming is not required, and thus it is less obtrusive for the physical activity.. In

addition, if UE output power is lower than IN output power, UE uplink range will be likely lower than

that if the downlink. These effects have to be taken into account in the design of data acquisition and

location protocols, which are described next.

Data exchange. Figures 3(a) and 3(b) show the state diagram of the data acquisition and processing

of the UE and the IN, respectively. Real-time data is harvested by the INs and UEs periodically (every

TIN Data and TUE Data seconds respectively). As previously stated, the set of measured parameters

depends on the sport, and spans from general variables like temperature or pressure, to particular ones,

like visibility conditions for ski practice, wave height in surfing, etc. Once the data are collected, the INs

send them to their parent nodes (see Figure 2(a)), which forward them again until they eventually reach

the CN. To save battery, the UE processes data internally to compute their moving averages and standard

deviations. The resulting packets with those statistics are delivered to the CN via the INs only when the

UE responds to location update queries (see 3(b)).


The CN issues orders for specific UEs by means of command packets (see Figure 2(c)). These com-

mands are delivered when the sportsman must undertake some action, such as selecting another pathway,

increasing the running pace, etc. The CN delivers the order by transmitting the command to a specific

IN, which, in turn, handles its transmission to the UE. The specific IN is selected according to the UE

position and its movement direction. Both informations are provided by the location procedure.

Location procedure. A discrete positioning system conducted by the INs computes sportsmen loca-

tion. Figure 2(b) illustrates this process, and figures 3(a) and 3(b) describe the IN and UE state diagrams,

respectively. It operates as follows: the CN periodically starts a search by sending a location request

message, including the identity of the nodes it looks for, or a special value indicating that all nodes must

respond. Immediately, the INs forward this query to all their leaf nodes using a broadcast packet. If

one or more of the sought nodes receive this packet, they answer to the IN with a unicast packet, which

also includes the last sportsman statistics. The receiving IN forwards the packet to the CN. However,

the sportsman may transitory lie out of IN coverage, because it crosses shadow areas (due to the reduced

UE-IN communication range), or because the broadcast packet gets lost. As a consequence, in the proto-

col implementation, each IN retransmits the broadcast query packet every TQUERY seconds for NQUERY

times. Eventually, the CN is informed of the nodes in range with each particular UE. The sportsman

movement direction may also be computed from the previous location records.

4. Implementation

A prototype based on the architecture described in the previous section has been implemented using

standard WSN hardware (MicaZ and Imote2 motes from Crossbow Technology [28]). This prototype

has been tested in a cross-country running application (described in the next section). Nevertheless, it

can be easily adapted to other sport activities, such as cycling, walking, etc. In the following paragraphs

the implementation of each component is described, as well as the parameter selection for the protocol.

4.1. IN implementation

Each IN requires sensing, processing, and communicating capabilities, as stated in section 3.. We

selected the MICAz mote [29], based on the CC2420 chip [31], as the IN core. It works in the 2.4 GHz

band, and it is compliant with the low power Zigbee/IEEE 802.15.4 [30] physical interface. Its Medium

Access Control (MAC) protocol is B-MAC [32]. On one hand, B-MAC provides suitable unicast (with

Automatic Repeat reQuest mechanism) and broadcast transmissions, both required for the location and

data transmission protocols. On the other, B-MAC minimizes energy consumption, which is mandatory

in a real WSM deployment. MICAz software relies on the open source event-oriented TinyOS operating

system [33], which is a reliable platform for ad-hoc protocol programming.

Environmental sensing is carried out by a MTS400 sensor board [34], designed for Crossbow motes,

which measures light, temperature, humidity, barometric pressure and seismic activity, although in our

testbeds (section 6.) we only activated temperature sensing.

In addition, each MICAz is equipped with two panel antennas (a 2.4 GHz Stella Doradus 24-8080

planar antenna [35]) to improve communication range. The reason is twofold: (i) to cope with radio

propagation issues (caused by natural obstacles like knolls, woodlands, etc) that may worsen commu-


Figure 4. Deployed hardware

(a) Infrastructure Node (b) Upper User Equipment

(c) Lower User Equipment

(d) Crossroad

nication among INs; and (ii) to increase communication range between the stations, allowing a lower

density of INs, and therefore, reducing installation and operation costs. In actual deployments, it was

verified that communication range may reach up to 180 meters, although the INs were installed with a

maximal distance of 150 meters in between, as a security margin and to increase the precision of the

location algorithm. Notice that, with the default antennae that are embedded in the MICAz motes, the

typical range is less than 100 meters.

Figure 4(a) shows a picture of two IN in a running track.


4.2. CN implementation

The CN can be considered a special IN, because it has the same sensing and data transmission ca-

pabilities. Moreover, the CN also implements the intelligence of the system: it issues commands to

the network and to the users in order to fulfill training goals. Its decisions are transparent to the user,

which only receives the training advice through his/her UE. In this implementation, the decision engine

selects the best path to achieve a target HR. The classification procedure to compute the HR is based on

a (m, s)-spline approximation. Section 5. describes this method in deep.

The location of the CN in our network deployment is shown in Figure 6(b).

4.3. User equipment

As in the IN case, the UE combines sensing and communications functionalities. Nevertheless, the

UE is designed to sense human biometrics, instead of environmental parameters. Human biometrics

variables usually change faster than environmental ones. Therefore, sample frequencies must be higher.

In addition, to avoid continuous packet transmission, these samples are processed in the UE, and only

their statistics are transmitted. The UE must also deliver training advice to the athlete. Thus, the UE is

also the interface between the system and the user. In the implementation, the advices are received and

transmitted to the user as vocal speech messages. To reduce the required bandwidth and energy, an advice

command is just a packet with an identification, and the UE contains in its memory the corresponding

vocal message.

To perform these operations, the UE is composed by several modules, as shown in Figures 4(b) and

4(c). The main module is the Crossbow IMOTE2 IPR2400 [36], a wireless sensor network platform that

can be expanded with extension boards to customize the system to a specific application. IMOTE2 also

includes an 802.15.4 radio (CC2420) with a built-in 2.4 GHz antenna for IN-UE data communications.

In the IMOTE2 there is an embedded IMB400 multimedia board [37] with a codec to play the speech

messages. These messages are recorded into the 32 MB FLASH memory of the IMOTE2, which allows

for up to 50 messages. The IMB400 also has a color image and video camera chip with audio capture,

which could be used in future system applications.

For human biometrics sensing, an integrated pulse oximetry device (iPOD model 3211, from Nonin

Medical company [38]) was used. This sensing device takes measures of heart rhythm and oxygen level

with a low power consumption. The iPOD was chosen due to its lightness, size, and easy integration in

the UE. In fact, only the connection of the iPOD RS-232 interface with the IMOTE2 required a simple

voltage adaptation. An specific driver in TinyOS for connectivity and iPOD control was also developed.

4.4. Protocol parameters selection

In order to tune protocol operation, different parameters and timers (introduced in section 3.) must be

established. Some of them are related to the sportsman activity (e.g. speed), whereas others are related to

environmental change rate. The selection of sportsman parameters was based on an informed guess that

guarantees system operation, limiting message exchange to minimize the impact on energy consumption.

The selection of environmental parameters is discussed next. Table 1 summarizes the parameters for the

specific example of section 6..


• TQUERY is the elapsed time between IN transmissions of location update requests. It is related

to the shadow time, TSHADOW , which an athlete may experience between consecutive INs. The

criterion was TQUERY = TSHADOW , since this guarantees that, in the worst case, the number of

lost packets is just one. Otherwise, for lower values, more packets may be transmitted (and lost)

while the user is in the shadow area. For higher values, the delay of the communication between

the IN and the UE will increase.

To compute TSHADOW let us denote v as the average expected user speed, and DSHADOW as the

length of the shadow area. Then,

TQUERY = TSHADOW =DSHADOW

v(1)

Besides, DSHADOW can be computed as a function of IN antenna directivity gain (GIN ), the

directivity gain of the UE (GUE), IN output power (PIN ), UE output power (PUE), and the distance

between INs (D).

DSHADOW = D(1 − 2

√GUE PUE

GIN PIN) (2)

Hence,

TQUERY =D(1 − 2

√GUE PUE

GIN PIN)

v(3)

Since in our testbed GIN = 11,15 dB, GUE = 1,41 dB, PIN = 0 dBm, PUE = 0 dBm, D = 150

meters, and v = 4,16 m/s (15 km/h), then TQUERY = 13 s.

• NQUERY is the number of times that an IN tries to locate UEs nearby. It is configured to be the

expected number of intervals of TQUERY s that a user stays in the IN range:

NQUERY =D

v

1

TQUERY(4)

With the reference configuration, NQUERY = 3.

• TCN is the elapsed time between two location procedures initiated by the CN. The minimal interval

corresponds to the expected time in IN range:

TCN = NQUERY TQUERY =D

v(5)

Higher values reduce energy consumption, at the expense of a lower location accuracy.

With the reference configuration, the minimal value should be TCN = 36 s.


Figure 5. HR evolution in three training sessions.

(a) Training session 1 (b) Training session 2

(c) Training session 3 (d) Training sessions average

5. Decision engine: (m, s) degree splines classification

The CN implements the decision engine, which infers future athlete performance and selects the most

suitable training parameters. In our development, the goal is maintaining the user HR within a selected

target range. The HR range depends on the activity level [39, 40]:

• Moderate activity. The sportsman heart rate is between 50% and 60% of his maximum value.

• Weight control activity. The sportsman heart rate is between 60% and 70% of his maximum value.

• Cardio training activity. The sportsman heart rate is between 70% and 80% of his maximum value.

• Anaerobic activity. The sportsman heart rate is between 80% and 90% of his maximum value.

• Maximum consumed oxygen volume activity. The sportsman heart rate is between 90% and 100%

of his maximum value.

Where the maximum HR is:

Maximum HR = 220 − sportsman’s age (6)


The decision consists in the next track the athlete must follow. Training advices are calculated by the

decision engine from the following variables:

• Current track environmental temperature, which is considered to be divided in two ranges: “hot”

for temperatures over 25 Celsius degrees or “cold” for lower temperatures.

• Future track temperature (Hot or Cold).

• Current track hardness: Difficult (D) or Easy (E) tracks.

• Future track hardness (Difficult or Easy).

• Average heart rate during the last track.

• Variance of the heart rate during the last track.

All these variables have influence on HR performance [14, 20]. The INs monitor the temperature

of the tracks, while the UE measures the user HR. The information on track hardness is static and it is

selected a priori taking into account the length and the slopes of each circuit. Note that the CN is aware

of the position of the different UEs that are present in the training field, and knows the direction of the

movement. When a user approaches a bifurcation, the CN selects the next track in two steps:

First, for all possible next tracks, the CN runs a classification algorithm which selects whether:

1. The user will increase his average HR over 5% of his maximum HR

2. The user will decrease his average HR below 5% of his maximum HR

3. The user Will keep his average HR within ± 5% of his maximum HR

Then, based on these results, and according to the user profile, the CN selects the appropriate tracks

with the following rules:

• If the target HR is below the current HR and a HR drop is predicted, the track belongs to the set of

suitable tracks.

• If the target HR is above the current HR and an increase of HR is predicted, the track belongs to

the set of suitable tracks.

• If the current HR is within ± 5% of the user maximum HR, and the expected HR will lie in this

interval, the track belongs to the set of suitable tracks.

Then, the user is notified the selected track. If there are several suitable tracks, the choice is random.

If none is selected, the system informs the user that all of them are valid.

The key of this process is the classification method, which must cope with the uncertainty of HR

prediction. As an example of its difficulty, Figure 5 shows results for three training sessions, which

were performed as part of the experimental validation in section 6.. It is interesting to observe in Figure

5(d) how the same part in different sessions produces so different average HR values. This is because

sportsmen psychology and physical condition have an important role in performance. They can induce


the sportsman to increase physical activity, and, accordingly, HR in the easy section of the course; or the

sportsman to rest in the same section, leading to activity and HR decrease. For example, this is possibly

the reason why session 3 ends several minutes before sessions 1 and 2 do. Note as well that HR increases

are not directly related to hardness or temperature increases: e.g. the hardness increase from track 7 to

track 8 (see figure 5(d)) produces a sudden HR drop.

To implement the classification algorithm different approaches have been considered, such as SVM,

but finally an interpolation method that computes the expected HR has been selected. Based on the

computed HR, the classification region is selected (increase, decrease or same level). Interpolation is

performed by means of (m, s)-degree splines. As it has been mentioned in Section 2., interpolation

techniques with splines functions have been extensively studied and successfully applied in computer-

aided geometric image processing, and curve and surface generations. These techniques can be easily

extended to approach solutions in multi-dimensional problems, as the one in this work.

5.1. (m, s) degree splines

Interpolation with (m, s)-degree splines has the following advantages that make them suitable to

implement the decision engine:

• Data points must not belong to a mesh grid (as in our case), unlike in other multi-variable inter-

polation methods (e.g. tensor product with finite elements, or with a basis of any vector space of

functions). Nevertheless, data points must differ for the problem to have a unique solution.

• Reduced computing cost to obtain the interpolation function. As described next, finding the solu-

tion requires to solve a system of linear equations with a matrix of symmetric coefficients, which

conditions standard solution methods (e.g Gaussian elimination). Once the interpolation function

is obtained, computing the next HR under new conditions in immediate.

The theory of (m, s)-degree splines derives from the study of the semi-Hilbert space of functions,

Xm,s, as is described in [24].

Let d ∈ N∗ be the number of dimensions considered (e.g. six in our problem). Given a set of elements

Ad∈Rd, and its corresponding values β ∈ R at those locations (e.g. HR in our problem), the goal is to

determine a mapping σdε as smooth as possible that approximates the observed data.

The problem can be formulated as the minimization of the penalized sum of squares:

Jdε (v) =

∣∣ρdv − β|2 + ε∣∣ v|2m,s (7)

where ε > 0, v ∈ Xm,s, and ρd is an operator defined as

ρdv = (v(a))a∈Ad (8)

Value ε∣∣v|2m,s represents a smoothness penalty on function v. The semi-norm |v|m,s is specifically

defined in [24, 25].

Then, the approximation spline of degree (m, s) relative to Ad, β, and ε, is a function σdε such that

∀v ∈ Xm,s, Jdε

(σd

ε

) ≤ Jdε (v).


Considering Pm as the ring of real polynomials of d variables, x1, x2, . . . , xd, and degree ≤ m, namely

{p ∈ Pm, p(x) =

∑α1+···+αd≤m

xα11 · · ·xαd

d ; x ∈ Rd

}(9)

In [24] it is demonstrated that σdε is unique and it can be obtained by solving the following system of

equations:

{εC∗λb +

∑a∈Ad λaK2m+2s−n(b − a) +

∑Mj=1 Cjpj(b) = βb, ε > 0∑

a∈Ad λapj(a) = 0, ∀j = 1, . . . , M(10)

where {pj, 1 ≤ j ≤ M} is a space basis of Pm−1, M = dimPm−1, and λa, a ∈ Ad, λb, b ∈ Ad, and

Cj , 1 ≤ j ≤ M are coefficients to be determined. The function K2m+2s−n(x) is

K2m+2s−n(x) =

{|x|2m+2s−n , si 2m + 2s �= 2l, l ∈ N

∗

|x|2m+2s−n log|x| , si 2m + 2s = 2l, l ∈ N∗ (11)

The constant C∗ has the following expression:

C∗ =

{(2π)2m

C1, si 2m + 2s �= 2l, l ∈ N

∗(2π)2m

C2, si 2m + 2s = 2l, l ∈ N

∗ (12)

with

C1 =π(2m+2s−n/2)Γ(n

2−m−s)

Γ(m+s), C2 = 2π(2m+2s−n/2)(−1)(m+s−n/2+1)

Γ(m+s)(m+s−n2 )!

(13)

Therefore, approximation (m, s) − splines has a coefficient matrix whose diagonal is εC∗. In this

work the value of ε is zero, and then σd is an interpolation function.

6. Validation experiment: cross-country training circuit

The prototype has been validated by means of a cross-country training experiment. The selected area

is located near Cartagena (Spain), and it is shown in figure 6. It consists of two interconnected loops (red

and blue) with different hardness and environmental conditions due to:

• Closeness to the sea in some areas of the circuit. For example, training by the coast is characterized

by constant winds.

• Different terrain slopes, since part of the circuit is on a hill whose height is 97 meters over sea

level, with some slopes of 14%.

• Shadow, depending on training hours and the trees along the circuit (as can be observed in the red

circuit in Figure 6(a). It has influence on temperature.

• Different lengths: 1.1 and 0.9 kilometers for the red and blue tracks, respectively.


Figure 6. Aerial sights of the training circuit.

(a) Cross training circuit

(b) Deployed infrastructure

The red track is considered difficult due to of its length and more pronounced slope. The blue track is

tagged as easy. We deployed ten INs in the difficult sector of the training area and eight in the easy one,

as shown in Figure 6(b). The CN is placed in the junction of both loops. The network was configured

with the protocol parameters that are summarized in Table 1.

The tests were performed for two different sportsmen. For each of them a specific classification

function was computed from his training data (knowledge base), which were obtained during twelve

training sessions. Some test results for three training sessions are shown in Figure 5. The same sessions

were repeated by both sportsmen, and they followed the same rules:

1. There was an initial warm-up (labeled as WU in figure 5). During it the sportsman trains in the

easy sector of the circuit. The data from this sector is discarded.

2. Each training session consists of ten loops in the circuit after the warm up sector. At each loop the

athlete ran either in the hard or in the easy track. The order was the same for all the sessions (Table

2), and it determined the course profile. Besides, the ambient temperature was also monitored.

The sessions were performed at the same hour in different days.

3. The HR sensor in the UE took one sample per second, and it discarded the measurements with

errors. Then, the system calculated the average HR in each loop from the correct samples.


Table 1. Protocol parameter selection

Parameter Value

TQUERY 13 s

TCN 60 s

TINDATA120 s

TUEDATA1 s

NQUERY 3

Table 2. Course profile of a training session.

Loop 1 2 3 4 5 6 7 8 9 10 11 12 13

Hardness WU E E E D E E D D E D D D

With the collected data, the classification mappings of section 5. were computed. The conformity

of the classification method was measured by means of its degree of discrepancy between the decision

engine outcome and the following real sportsman HR. To perform these conformity tests, several samples

were taken from the training set. From the knowledge base of the first sportsman, which included a

total of 119 HR values with their associated variable values, 14 records were randomly chosen. For

the second sportsman, with 110 HR and associated variable values in his knowledge base, 12 records

were randomly chosen. These records represent 11.8% and 10.9% of their respective knowledge bases.

Then, the classification method was applied (to compute the expected HR rate) and the success rate was

measured (notice that for every sample the real HR value was available). Table 3 summarizes the results

of four conformity tests for the two sportsmen.

Table 3. Conformity test results.

Base Records Percentage of valid decisions

dimension analyzed Test 1 Test 2 Test 3 Test 4

sportsman-1 119 11.80% 76.80% 76.40% 85.70% 71.40%

sportsman-2 110 10.90% 75.00% 75.00% 75.00% 83.30%

The (m, s)-spline decision engine outperforms a SVM for the same knowledge base (the outcomes of

a SVM practically resembled a toss coin test), that reflects the uncontrolled effects of sportsman subjec-

tive sensations about the intensity of the physical activity (Figure 5) and the nonlinear behavior of HR

fluctuations (mentioned in Section 2.). We have carried out tests that also considered environment tem-

perature and track hardness for two of three previous tracks (instead of only one), but SVM performance

did not improve.


Moreover, in the deployment, the communication and location delays and the location accuracy were

measured. The round trip time of a message from the CN to a particular UE (or viceversa) was less than

0.55 ms per hop for a packet length of 10 bytes including headers. The average location delay was 5.73

s, (the procedure may suffer temporary fading of the communications with the UE due to shadow areas).

The average number of nodes that detected a UE was 1.3.

7. Conclusions and future work

This work demonstrates the feasibility of ambient intelligence technologies applied to outdoor sports

practice. The environment, the athlete condition, and the track configuration are taken into account

to provide real-time feedback directly to the user. A main outcome is the design of the classification

method, which features a success rate that is close to 80%. Besides, both the infrastructure and the

user equipment, which were developed with off-the-shelf sensor technology, behaved well for ambient

intelligence applications. The results of this work may be extrapolated to other contexts than sports.

As future work, we aim at developing less intrusive UE devices, and design testbeds for different

sports.

Acknowledgements

This work has been supported by grants DEP2006-56158-C03-01/02/03, TEC2007-67966-01/02/TCM

CON-PARTE-1/2 (Ministerio de Educacion y Ciencia, Spain), TSI-020301-2008-16 ELISA and TSI-

020301-2008-2 PIRAmIDE (Ministerio de Industria, Turismo y Comercio, Spain), and it has been also

developed within the framework of “Programa de Ayudas a Grupos de Excelencia de la Region de Mur-

cia”, funded by Fundacion Seneca, Agencia de Ciencia y Tecnologia de la Region de Murcia (Plan

Regional de Ciencia y Tecnologia 2007/2010).

References

1. Aarts, Emile; Harwig, Rick; Schuurmans, Martin. ”Ambient Intelligence” in The Invisible Future:

The Seamless Integration Of Technology Into Everyday Life. 2001. McGraw-Hill Companies.

2. Training plan bank with Polar devices,

http://www.polarusa.com/us-en/training with polar/training plan bank

3. Ministry of Education, Culture and Sports webpage, http://www.educacion.es/portada.html, [in

spanish].

4. Perez-Soriano, P.; Llana-Belloch, S. Instrumentation in Sports Biomechanics. Journal of Human

Sport and Exercise 2007; vol. 2, pp. 26-41.

5. Mao G.; Fidan B.; Anderson, B. D.O. Wireless sensor network localization techniques, Computer

Networks; Volume 51, Issue 10, 11 July 2007, pp. 2529-2553,

6. Armstrong, S. Wireless connectivity for health and sports monitoring: a review. British Journal of

Sports Medicine, 2007; 41:285-289.

7. Baca, A.; Kornfeind, P. Rapid Feedback Systems for Elite Sports Training, Pervasive Computing,

2006; pp. 70-76.

8. Pfisterer, D.; Lipphardt, M.; Buschmann, C.; Hellbrueck, H.; Fischer, S.; Hendrik Sauselin, J.


MarathonNet: adding value to large scale sport events - A connectivity analysis. Conference on

Integrated Internet Ad-hoc and Sensor Networks, 2006.

9. Michahelles, F.; Schiele, B. Sensing and Monitoring Professional Skiers. Pervasive Computing,

2005; pp. 40-46.

10. Hailes, S.; Kalra, D.; Hopper, A.; Lasenby, J.; Kerwin, D.; Wilson, A. SESAME (SEnsing for Sport

And Managed Exercise). http://www.sesame.ucl.ac.uk/index.html. 2006-2010.

11. Team sports products from Polar Company webpage,

http://www.polarusa.com/us-en/b2b products/team sports

12. Chi, Ed H. Introducing Wearable Force Sensors in Martial Arts. Pervasive Computing, 2005; pp.

47-53.

13. Thomas Jaitner, T.; Trapp, M.; Niebuhr, D.; Koch, J. Indoor-Simulation of Team Training in Cy-

cling. Engineering of sport 6; Vol. 1; Developments for sports. International Sports Engineering

Association. Springer Science, 2006.

14. Backx, K; van Someren, K A.; Nevill, A M.; Palmer, G S. Mathematical Prediction of One Hour

Cycle Time Trial Performance Under Different Ambient Temperatures. Medicine & Science in

Sports & Exercise, May 2003; Volume 35, Issue 5; p S30.

15. Kaplan, D.T.; Furman, M.I.; Pincus, S.M.; Ryan, S.M.; Lipsitz, L.A.; Goldberger, A.L. Aging and

the complexity of cardiovascular dynamics. Biophysical Journal; Volume 59, Issue 4, April 1991;

pp. 945-949.

16. Absilz,P.-A.; Sepulchrez, R.; Bilgey, A.; Gerard, P. Nonlinear analysis of cardiac rhythm fluctua-

tions using DFA method. Physica. A, Statistical mechanics and its applications, 1999; vol. 272,

no1-2; pp. 235-244.

17. Huikuri, H.V.; Makikallio, T.H.; Perkiomaki, J. Measurement of heart rate variability by methods

based on nonlinear dynamics. Journal of Electrocardiology; Volume 36, Supplement 1, December

2003; pp 95-99.

18. Veropoulos, K.; Cristianini, N.; Campbell, C. The Application of Support Vector Machines to

Medical Decision Support: A Case Study. Advanced Course in Artificial Intelligence (ACAI”99),

Jul. 1999.

19. Morik, K.; Brockhausen, P.; Joachims, T. Combining statistical learning with a knowledge-based

approach - A case study in intensive care monitoring. Proc. 16th International Conference on

Machine Learning (ICML-99), 1999.

20. Yuchi, M.; Jo, J. Heart Rate Prediction Based on Physical Activity Using Feedforwad Neural Net-

work. Hybrid Information Technology, International Conference on, pp. 344-350, 2008 Interna-

tional Conference on Convergence and Hybrid Information Technology, 2008.

21. Lancaster, P.; Salkauskas, K. Curve and Surface Fitting. An Introduction. Academic Press, Lon-

don; 1986.

22. Torrens, J. J. Discrete smoothing Dm-splines: applications to surface fitting. In Proceedings of

the international Conference on Mathematical Methods For Curves and Surfaces II Lillehammer,

1997 (Lillehammer, Norway). Eds. Vanderbilt University, Nashville, TN; pp. 477-484.

23. Unser, M. Splines: A perfect fit for signal and image processing. Signal Processing Magazine,

1999; Vol. 16, no. 6; pp. 22-38.


24. Lopez de Silanes, M.C.; Arcangeli, R. Estimations de l erreur d approximation par splines d inter-

polation et d a justement d ordre (m,s). Numerische Mathematik, no56, Springer Verlag, 1989; pp.

449467.

25. Sharma G.; Shaw M.Q. Thin-plate splines for printer data interpolation. In Proceeding of European

Signal Processing Conference (EUSIPCO), 2006.

26. The Collection Tree Protocol(CTP), TinyOS-2.X webpage,

http://www.tinyos.net/tinyos-2.x/doc/html/tep123.html

27. J. Vales-Alonso; S. Costas-Rodrıguez;M. V. Bueno-Delgado; E. Egea-Lopez; F. Gil-Castineira;

P. S. Rodrıguez-Hernandez; J. Garcıa-Haro; F. J. Gonzalez-Castano. An Analytical Approach to

the Optimal Deployment of Wireless Sensor Networks. Computational Intelligence for Remote

Sensing, Springer Berlin / Heidelberg, Volume 133/2008, Pages 145-161.

28. Crossbow Technology webpage, http://www.xbow.com/

29. Micaz, Crossbow webpage, http://www.xbow.com/Products/productdetails.aspx?sid=164.

30. Zigbee Alliance webpage, http://www.zigbee.org/LearnMore/WhitePapers/tabid/257/Default.aspx

31. CC2420 Chip Documentation, http://docs.tinyos.net/index.php/CC2420

32. BMAC, Versatile Low Power Media Acces for WSN, http://www.polastre.com/papers/sensys04-

bmac.pdf

33. TinyOS webpage, http://www.tinyos.net/.

34. MTS400, Crossbow webpage, http://www.xbow.com/Products/productdetails.aspx?sid=177

35. MTS400, Stella Doradus webpage, http://www.stelladoradus.com/2.4.ghz.planar.antennas.php

36. Imote2, Crossbow webpage, http://www.xbow.com/Products/productdetails.aspx?sid=253.

37. IMB-400 Multimedia Board, Crossbow webpage,

http://www.xbow.com/Products/productdetails.aspx?sid=280

38. Pulse Oximeter and Sensor webpage, http://www.nonin.com/products/ipod-digital-oximeter-and-

sensor-all-in-one/

39. Musnick, D.; Pierce, M. Conditioning for outdoor fitness: functional exercise and nutrition for

every body; The Mountaineers Books, 2004.

40. Mora, J. Triathlon workout planner; Eds.; Human Kinetics, 2006; pp. 45-49.

41. Anlauff, M.; Sunbul, A. Deploying localization services in wireless sensor networks. In Proceed-

ings. 24th International Conference on Distributed Computing Systems Workshops, March 2004;

pp. 782-787.

42. Hu, L.; Evans, D. Localization for Mobile Sensor Networks. International Conference on Mobile

Computing and Networking, 2004.

43. HZein-Sabatto, S.; Elangovan, V.; Wei Chen; Mgaya, R. Localization strategies for large-scale

airbone deployed wireless sensors. Computational intelligence in multi-criteria decision-making,

mcdm ’09. ieee symposium, 2009; pp. 9-15.

44. Ganesh, S. Efficient localization scheme for wireless sensor networks. Wireless Communications

and Sensor Networks, WCSN 2008. Fourth International Conference, 2008; pp. 73-76.

45. Galstyan, A.; Krishnamachari, B.; Lerman, K.; Pattem, S. Distributed online localization in sensor

networks using a moving target. Information Processing in Sensor Networks. IPSN 2004. Third

International Symposium, 2004; pp. 61-70.


46. D’Souza, M.; Wark, T.; Ros, M. Wireless localisation network for patient tracking. Intelligent

Sensors, Sensor Networks and Information Processing. ISSNIP 2008. International Conference,

2008; pp. 79-84.

47. Oh, S.; Sastry, S. Tracking on a Graph. Information Processing in Sensor Networks, IPSN 2005.

Fourth International Symposium , 2005; pp. 195-202.

c© November 16, 2009 by the authors; submitted to Sensors for open access under the terms and condi-

tions of the Creative Commons Attribution license http://creativecommons.org/licenses/by/3.0/.

Documents

Ambient Intelligence Systems for Personalized Sport …ait.upct.es/deportes/publicaciones/Ambient Intelligence Systems for... · Ambient Intelligence Systems for Personalized Sport