Step Length Measurement Through Signal Power Analysis and

Accelerometer Device: A Laboratory Comparison

Christopher S. Ramon

1 a

, Santiago R. Coronel

1 b

, Jostin L. Ruiz-Zambrano

1 c

ul I. Villalta-Encalada

1 d

, Julio C. Chuqui-Calle

2 e

, Luis J. Serpa-Andrade

1 f

Freddy L. Bueno-Palomeque

1 g

and Pa

ul A. Chasi-Pes

antez

1 h

GIHEA - Grupo de Investigaci

on en Hardware Embebido Aplicado, Universidad Polit

ecnica Salesiana, Cuenca, Ecuador

Grupo de Investigaci

on en Ciencias de la Actividad F

ısica, Universidad Polit

ecnica Salesiana, Cuenca, Ecuador

Keywords:

Step Length Measurement, Race Walking, Received Signal Strength Indicator, Signal Power, Inertial Sensor,

Accelerometer.

Abstract:

The monitoring and analysis of kinetic and kinematic parameters integrated into sports disciplines such as race

walking can provide valuable information for developing personalized training programs and evaluating tech-

nique execution. In this study, we developed two systems for measuring step length based on microcontrollers

ESP32. The ﬁrst system is based on measuring the received signal strength between two antennas, while the

second relies on inertial sensors (MPU-9250). Both systems were tested in a laboratory setting using a tread-

mill and video recording to assess their accuracy. The results showed that the system based on signal strength

measurement exhibited low precision at distances within a range of a few centimetres. On the other hand, the

inertial sensor-based system demonstrated higher accuracy when compared to video recordings. Although the

measurements differed statistically between these two methods (p-value = 0.001), the proposed inertial system

recorded a step length of 65 (61-69) cm, while the video recordings measured 67 (64-70) cm. The error dis-

tribution analysis showed that 39% of measurements had an error of 3.2 cm, 32% had an error of 7.5 cm, and

29% had an error not exceeding 12 cm. The proposed system shows potential for step length quantiﬁcation

using the MPU-9250 sensor; however, further testing is required to reduce the measurement errors.

1 INTRODUCTION

Race walking is a sport discipline that combines as-

pects of normal walking with specialized biomechan-

ical techniques, characterized by the absence of a

ﬂight phase and the need for continuous contact of

at least one of the lower limbs with the ground (Ca-

poraso et al., 2020). Stride length and various factors,

such as height, muscle strength, ﬂexibility, time ﬂight,

and technique, converge and their analysis enables

the identiﬁcation of movement patterns, biomechan-

ical alterations, and optimization of athletes’ walking

https://orcid.org/0009-0009-9244-9238

https://orcid.org/0009-0005-7083-2392

https://orcid.org/0009-0009-1887-7752

https://orcid.org/0009-0000-5905-3770

https://orcid.org/0009-0003-2150-2173

https://orcid.org/0000-0003-1338-7868

https://orcid.org/0000-0003-1626-0272

https://orcid.org/0000-0002-4364-7483

technique (Pavei et al., 2019; Barreto-Andrade et al.,

2016).

The aforementioned parameters require that the

technique necessitates rigorous monitoring and con-

trol, both in training stages and in ﬁeld competitions,

to optimize performance and prevent injuries. How-

ever, in addition to their complexity, the current tools

available for measuring and analysing these biome-

chanical parameters are limited in accessibility, mak-

ing their use in amateur training groups challenging

(Taborri et al., 2019).

Studies involving kinematic analysis using high-

speed cameras or force platforms face practical chal-

lenges in outdoor environments, mainly due to their

high complexity in acquisition and setup. Although

precise, these technologies are less accessible for real-

world applications outside controlled laboratory envi-

ronments due to their reliance on stable setups and de-

tailed calibrations to ensure data accuracy. In (Bernar-

dina et al., 2024), the feasibility of using action sports

cameras for 3D motion reconstruction of swimmers

Ramon, C. S., Coronel, S. R., Ruiz-Zambrano, J. L., Villalta-Encalada, R. I., Chuqui-Calle, J. C., Serpa-Andrade, L. J., Bueno-Palomeque, F. L. and Chasi-Pesántez, P. A.

Step Length Measurement Through Signal Power Analysis and Accelerometer Device: A Laboratory Comparison.

DOI: 10.5220/0013619100003988

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 13th International Conference on Sport Sciences Research and Technology Support (icSPORTS 2025), pages 119-125

ISBN: 978-989-758-771-9; ISSN: 2184-3201

119

both in air and underwater was explored, highlighting

a more economical and ﬂexible alternative to tradi-

tional motion capture systems.

From the perspective of portable solutions, the re-

search conducted in (Zrenner et al., 2018) has demon-

strated that using inertial sensors to measure running

speed and stride length can be a practical and portable

tool, such as RunScribe and Garmin Running Dynam-

ics Pod that utilize accelerometers for data collec-

tion. RunScribe offers a device equipped with the

InvenSense MPU-9255 motion sensor, which inte-

grates a gyroscope, accelerometer, and magnetometer

in a three-axis system. The inclusion of these tech-

nologies allows for the measurement of biomechan-

ical variables such as ground reaction force (Lewin

et al., 2022). Similarly, the Garmin Running Dynam-

ics Pod uses an accelerometer to capture torso move-

ment and measure key metrics like cadence, stride

length, ground contact time, and vertical ratio (Drob-

nic et al., 2023). The rapid advancement of sen-

sors and the reduction in manufacturing costs have

driven the increasingly widespread use of technology

in sports, encouraging its development and adaptation

to speciﬁc training needs (Campoverde-G

arate et al.,

2022; Flores-Morales et al., 2016; Cho, 2017; Mali

and Dey, 2020).

Building on the difﬁculties exposed, this project

assesses the viability of developing solutions for mea-

suring stride length in race walking by comparing the

use of inertial measurement sensors with the data pro-

vided by their accelerometers, and the measurement

of power using the received signal strength indicator

(RSSI) parameter through WiFi communication be-

tween two ESP32 microcontrollers.

Previous works proposes the use of RSSI for lo-

calization in both indoor and outdoor sports sys-

tems. In (Pricone and Caracas, 2014), they use RSSI

to estimate distance through trilateration, adapting a

path loss model with shadowing to adjust the signal

strength received based on distance, which is essen-

tial for real-time tracking applications. In the study

presented by (Ruiz Zambrano and Villalta Encalada,

2024), power measurement was evaluated as a tool to

relate it to step length. However, the results indicated

that the system was not suitable for short distances.

Based on these results, we hypothesized that stabiliz-

ing the transmission power and improving temporal

synchronization of transmitted values of power data

between transmitter and receiver will provide more

accurate results when measuring short distances.

The use of the MPU-9250, which integrates ac-

celerometers and gyroscopes, enables gesture detec-

tion in various sports disciplines such as running,

weightlifting, or race walking. (C

ardenas-Rodr

ıguez

et al., 2023; Mali and Dey, 2020; Flores-Morales

et al., 2016).

Our study proposes to compare two methods for

measuring step length: the method based on measur-

ing the transmitted power between the emitter and re-

ceiver and the accelerometer-based method. Our ob-

jective was to evaluate both methods with the inten-

tion of developing, in the future, a system for moni-

toring step length in race walking as a tool for training

monitoring in sports.

2 MATERIALS AND METHODS

2.1 Measurement Based on Power

Transmission (RDU)

To conduct this research, a distance measurement

system based on RSSI was ﬁrst developed (RDU -

RSSI Distance Unit), based in the measurement of

the power level of a received signal, useful for dis-

tance estimation (Vikas et al., 2016). For distance

calculation we used two ESP-WROOM-32 develop-

ment board. For communication, the integrated IEEE

802.11 b/g/n WiFi module operates at 2.4 GHz, set-

ting up the two ESP32 microcontrollers as an access

point and a client as shown in Fig. 1 where the system

operation is generally explained.

Figure 1: Diagram for communication between microcon-

trollers in the step length estimation system through trans-

mission power measurement.

The microcontroller conﬁguration process is de-

tailed below:

• Terminal TX Conﬁguration: An ESP32 micro-

controller was conﬁgured as an access point, it

creates a WiFi network with a predeﬁned SSID

(Service Set Identiﬁer) and password to estab-

lish a secure communication link. It continuously

transmits a preconﬁgured message with a times-

tamp corresponding to the moment the message

icSPORTS 2025 - 13th International Conference on Sport Sciences Research and Technology Support

120

is sent, ensuring a constant ﬂow of data between

devices.

• Terminal RX Conﬁguration: A second ESP32 mi-

crocontroller was conﬁgured as a device that con-

nects directly to the ﬁrst microcontroller TX act-

ing as a host using the assigned credentials to re-

ceive and process the messages. Its main func-

tions include decoding the received message, ex-

tracting, and logging the RSSI value associated

with the message.

• Communication: To enable communication, the

client receives the message from the access point.

This process ensures that the transmission time

and transmission losses are recorded using the

RSSI parameter for each instance of communica-

tion. The time interval between transmitted mes-

sages was set to 75 ms and the initial transmission

power was set at -19 dBm as the upper limit to

ensure a stable connection.

The distance estimation through the RDU system is

based on the logarithmic relationship between signal

intensity and distance (Wu et al., 2015; Moya Ve-

lasco, 2019). This model describes signal loss as a

logarithmic function of distance with Equation 1:

(d) = P

) + 10nlog





+ X

, (1)

where n is a path loss factor, typically between 2 and

5, depending on the environment, X

is a random vari-

able with a normal distribution, and d

is the reference

distance, usually 1m. This model is suitable for out-

door applications due to its low cost and acceptable

accuracy (Cao and Li, 2013).

To capture the data, measurement devices were

placed on the subject’s body (see Fig. 2). The RDU

system was positioned at the front part of each shoe.

Building upon the work presented in (Ruiz Zambrano

and Villalta Encalada, 2024), we implemented in soft-

ware a temporal synchronization stage to manage

transmission of the measured power data. Synchro-

nization was achieved by sending two parameters in

an array: a timestamp and the corresponding RSSI

value. Path loss was then determined by calculat-

ing the difference between transmitted and received

power for each recorded value across the predeﬁned

time intervals.

Both of the systems (RDU and SIMPU) used a

push-button placed at the bottom of the shoe, which

helped synchronize data collection since the push-

button is responsible for initiating the systems to col-

lect data in a coordinated manner.

Figure 2: A: Positioning of the SIMPU (red dotted line)

and RDU (white dotted line) systems mounted on one leg

of the subject for treadmill testing. B: Image focused on the

inertial sensor (blue dotted line) and microcontrollers for

data management.

2.2 Measurement Based on

Accelerometer (SIMPU)

At this stage of the project, we designed a system

based on the MPU-9250 inertial sensor (SIMPU),

which measures acceleration in three axes, allowing

for distance estimation using the Attitude and Head-

ing Reference System (AHRS) algorithm. Addition-

ally, we included a Kalman ﬁlter to enhance the accu-

racy of these angles through data fusion and Bayesian

estimation, considering both system and observation

noise (Chunyang et al., 2015; Ter

an Pineda, 2017).

This ﬁlter predicts the system state and then corrects

it based on the observed measurements (Ter

an Pineda,

2017; Krishnaveni et al., 2022). The prediction equa-

tion is:

= (I − K

H)P

−

(2)

Where P

is the estimation error covariance after

correction, I is the identity matrix, K

is the Kalman

gain, H is the observation matrix, and P

−

is the pre-

dicted estimation error covariance.

The AHRS algorithm processes the corrected data

from the MPU-9250 sensor to continuously estimate

the device’s orientation in real-time. This orientation

is represented by Euler angles: roll (θ), pitch (φ), and

yaw (ψ), which describe how the athlete’s foot tilts

and rotates during race walking (Wang et al., 2014;

Madgwick, 2014), as seen in Fig. 3.

The MPU-9250 obtains these orientations from

the integration of angular velocities provided by its

gyroscope, as expressed in the following equation:

Rotation Angle =

angular velocitydt (3)

However, due to the accumulation of drift errors in

the integration process, the AHRS algorithm applies

sensor fusion techniques, combining data from the ac-

celerometer and gyroscope. This approach, optimized

Step Length Measurement Through Signal Power Analysis and Accelerometer Device: A Laboratory Comparison

121

Figure 3: Block diagram of distance estimation based on

accelerometer readings.

by Madgwick (Madgwick, 2014), corrects cumulative

deviations, ensuring more accurate and stable orienta-

tion estimates, even during dynamic movements. The

SIMPU system was afﬁxed to the mid-part of the tibia

to minimize possible tilts that could introduce dis-

turbances into the inertial measurement system (see

Fig. 2).

2.3 Laboratory Tests

Given the exploratory nature of this laboratory study,

the assessments were conducted with a single par-

ticipant (27-year old, height=1.81 m, and mass=105

kg). The subject was in good health, had no suf-

fered recent injuries, and was fully informed about

the scope of the project from the beginning and vol-

untarily agreed to participate. The subject was sub-

jected to a race walking session on a treadmill in

the Effort Laboratory of the Universidad Polit

ecnica

Salesiana, Cuenca Campus. For video recording, an

smartphone camera was used (108-megapixels and 30

fps) and positioned parallel to the subject’s plane of

movement on the treadmill. For comparative eval-

uation, Kinovea software (v1.2, open-source, avail-

able at http://www.kinovea.org) was used, which al-

lowed for an analysis of measurements in real time

against video captures. Data from the RDU system,

SIMPU, and video images were recorded, enabling

cross-validation of the obtained results.

The RDU and SIMPU systems operate indepen-

dently. For the RDU setup, the system is placed on

the right ankle, as shown in Fig. 2, while a micro-

controller is attached to the left ankle to record the

received signal strength. For the SIMPU system, a

single unit is positioned on the right ankle. Once the

systems are in place, the subject walks on the tread-

mill, and when the target speed is reached, an exter-

nal trigger is applied via a push button to synchronize

both systems with the video recording.

The laboratory tests were divided into two stages.

In the ﬁrst stage, three sessions were conducted (one

minute each one). The subject wore visible markers

(white tape) on the key areas where devices were in-

stalled, which facilitated the correct identiﬁcation of

each system during analysis in the reference software.

As a metric reference, a 1.5 m tape measure was used

to calibrate the reference in Kinovea software. To

avoid inconsistencies in comparing results, both sys-

tems (RDU and SIMPU) were evaluated simultane-

ously, ensuring that the measurements obtained reﬂect

the same experimental context and allowing a direct

and unbiased comparison.

In the second stage of testing, we discarded the

system that exhibited the highest error and repeated

the tests using only the remaining system. For

the statistical analysis we used the Wilcoxon rank

sum test to compare the medians of the registered

data, after assessing data normality using the Kol-

mogorov–Smirnov test. This non-parametric test does

not assume normality of the recorded step length mea-

surements. The hypothesis test was based on the

null hypothesis (H

), which stated that the measure-

ments obtained with our system were comparable to

those recorded by video. A signiﬁcant result (p-

value≤0.05) would indicate a difference between the

two sets of measurements.

3 RESULTS

During the ﬁrst stage of testing, a total of 133 steps

were recorded, and the results analysis in this section

was based on this data. The measurements recorded

with both systems and the Kinovea software can be

observed in Fig. 4A. Additionally, a description of

these results is presented in Table 1, which includes

the p-value obtained by comparing the results of the

two proposed systems with those of the Kinovea soft-

ware. In the comparative evaluation, we can observe

that the measurements with the SIMPU system show

a high level of similarity with the Kinovea records (p-

value=0.350). On the other hand, the measurements

taken with the REDU system are statistically different

(p-value<0.001).

The measurements of each system were compared

with the Kinovea record, and the absolute error is cal-

culated for each measurement. Fig. 5 shows the er-

ror distribution for the SIMPU and RDU systems. It

can be observed that the SIMPU system exhibits a

more concentrated error distribution with less disper-

sion around zero, indicating greater precision and less

variability in its measurements. In contrast, the RDU

system shows a wider distribution, with scattered er-

rors and a higher frequency of elevated values, sug-

gesting lower accuracy and greater variability in data

icSPORTS 2025 - 13th International Conference on Sport Sciences Research and Technology Support

122

Figure 4: A: Box plot comparing the distribution of step length measurements between SIMPU, RDU, and Kinovea software

in tests stage 1. B: Comparison of step length measurements between SIMPU and Kinovea in the test stage 2.

Table 1: Description of the data obtained by the SIMPU

and RDU systems in meters (m). The median value and in-

terquartile range are presented. The RMSE and the p-value

shown was calculated by comparing the results of each sys-

tem with those obtained from Kinovea.

System Step Length RMSE p-value

First test

Kinovea 0.56 (0.52 - 0.62) - -

SIMPU 0.55 (0.51 - 0.64) 0.057 0.350

RDU 0.80 (0.66 - 0.94) 0.299 <0.001

Second test

Kinovea 0.67 (0.64 - 0.70) - -

SIMPU 0.65 (0.61 - 0.69) 0.045 0.001

Figure 5: Calculation of the absolute error of the measure-

ments with the proposed systems compared to the Kinovea

measurements.

measurement.

The data obtained, which contrast the readings of

the systems with the reference measurements from

Kinovea software, indicate that the SIMPU system

exhibited statistically similar, as evidenced in Fig. 4A.

The values for the SIMPU system remain within a

narrower and more predictable range compared to

those of the RDU system, which shows considerable

variability and a lack of a clear pattern of behavior.

The more centred median and narrower range in the

SIMPU data compared to the RDU data (see Table 1)

reinforce the idea that SIMPU is more stable and re-

liable for these measurements. The RMSE value be-

tween SIMPU and Kinovea was 0.057 m, indicating a

small average error between the measurements.

For the second stage of testing (304 steps

recorded), measurements were conducted exclusively

with the SIMPU system in the same laboratory and

under identical test conditions. The step length results

were incorporated into Fig. 4B. for comparison, while

the median and interquartile range results were added

into the lower section of Table 1. The results show

an maximum error of 20%, similar to that obtained in

the ﬁrst stage of testing. However, when statistically

comparing the results, we observe a p-value=0.001,

indicating that the results are statistically different,

while the small RMSE value (0.045 m) reﬂects a low

average deviation.

4 CONCLUSIONS

The work carried out in this research contrasts two

systems developed for the quantiﬁcation of step

length in race walking. The mentioned sport has been

the inspiration for this study; however, the scope of

the tests conducted for this document has remained

within the laboratory. This project has been directed

toward the development of a system for monitoring

step length in race walking, focusing on training and

monitoring based on easily accessible and highly re-

liable technologies (Mali and Dey, 2020). The re-

sults presented in this work were obtained from a sin-

gle participant; therefore, they are not generalizable.

Step Length Measurement Through Signal Power Analysis and Accelerometer Device: A Laboratory Comparison

123

Nevertheless, they provide preliminary evidence re-

garding the behaviour of the systems developed in our

laboratory for the future evaluation of stride length.

The instability of the RDU complicates the possi-

bility of establishing effective corrections to minimize

errors, whereas the consistency of the SIMPU sug-

gests greater viability for its implementation in step

length measurement systems in sports and biome-

chanical applications. Consequently, the analysis

concludes that the use of SIMPU as a measurement

system is the less imprecise option in step length mea-

surement, highlighting its potential as a low-cost tool

with promising prospects for reducing measurement

error, which are fundamental for the development and

validation of human movement models.

Based on the results obtained from the absolute

error comparison, with -0.01% and 43.77% being the

average error for SIMPU and RDU respectively, it is

concluded that the power-based measurement method

presents problems of readjustment. This is due to the

Wi-Fi communication protocol attempting to main-

tain a stable connection by dynamically adjusting the

transmission power. However, this behavior affects

the transmission value, making it difﬁcult to obtain

consistent measurements.

Additionally, the Wi-Fi protocol parameter of

ESP32 is a partially closed system that does not al-

low the modiﬁcation in a customized manner to opti-

mize power usage. Consequently, in situations where

the distance between devices changes, signiﬁcant er-

rors in signal transmission are generated, as observed

in the obtained results. This demonstrates the limita-

tions of the RDU for accurate measurements in sce-

narios where connection conditions vary.

Regarding the SIMPU system, the data obtained

demonstrates superior performance when using this

device for measurements. As a more controlled sys-

tem, it is easier to make adjustments in the capture of

acceleration and position data, thanks to the integra-

tion of its accelerometer and gyroscope.

In the second stage of testing between SIMPU and

Kinovea, we can statistically observe that the mea-

surements are different (p-value=0.001), indicating a

difference in their distribution due to their signiﬁcant

disparity. Based on the recorded measurements, there

is a 39% probability that the measurement has an ab-

solute error of less than 5%, a 32% probability that it

falls within the range 5% < absolute error < 10%, and

a 29% probability that it exceeds 10% (up to approx-

imately 20%, which is the maximum recorded error).

Considering a median value of 0.65 m, an absolute er-

ror of 5%, 10%, and 20% corresponds to a step length

reading error of 3.2 cm, 6.5 cm, and 12 cm, respec-

tively.

While it does not achieve the precision necessary

to conform to the metrics established in the mea-

surement of step length in race walking, its results

are signiﬁcantly better than those obtained with the

RDU system. The maximum error recorded is ap-

proximately 25%, positioning it as the better option

among the two systems evaluated. As future work, it

is proposed to develop an embedded system with the

SIMPU system and conduct ﬁeld tests with athletes.

ACKNOWLEDGEMENTS

This work was supported by project “RWA+ Race

Walking Analytics” (Resolution No. 005-001-

2024-01-18) from Universidad Polit

ecnica Salesiana,

Cuenca, Ecuador.

REFERENCES

Barreto-Andrade, J., Villarroya-Aparicio, A., Contreras-

Calle, T., Brito-V

asquez, V., and Loaiza-D

avila, E.

(2016). Biomec

anica de la marcha atl

etica. an

alisis de

las presiones plantares durante su desarrollo, revisi

actualizada. Lecturas: educaci

on f

ısica y deportes,

21(217):1–9.

Bernardina, G. R. D., Andrade, A. G. P. d., Monnet, T.,

Cerveri, P., and Silvatti, A. P. (2024). Simultaneous

in-air and underwater 3d kinematic analysis of swim-

mers: Feasibility and reliability of action sport cam-

eras. Journal of Biomechanics, 168(112078).

Campoverde-G

arate, J. P., Chuqui-Calle, J. C., Serpa-

Andrade, L. J., and Bueno-Palomeque, F. L. (2022).

Detection of ﬂight phase in race walking based on

pressure sensors. In 2022 IEEE 40th Central Amer-

ica and Panama Convention (CONCAPAN), pages 1–

5. IEEE.

Cao, Z. and Li, D. (2013). Exploration of parameter factor

in the rssi model. Applied Mechanics and Materials,

268-270:1822–1831.

Caporaso, T., Grazioso, S., Gironimo, G. D., and Lan-

zotti, A. (2020). Biomechanical indices represented

on radar chart for assessment of performance and in-

fringements in elite race-walkers. Sports Engineering,

23:4.

ardenas-Rodr

ıguez, M. F., Paute-Tigre, C. G., and Bueno-

Palomeque, F. L. (2023). In-ﬁeld recording of six

biaxial angles and plantar pressures in weightlifting

through a wearable system. Journal of Medical Sig-

nals & Sensors, 13(4):290–299.

Cho, H. (2017). Design and implementation of a

lightweight smart insole for gait analysis. In 2017

IEEE Trustcom/BigDataSE/ICESS, pages 792–797.

IEEE.

Chunyang, L., Fan, C., Xin, S., Hongtao, C., Junling,

X., and Yujun, X. (2015). Gesture detection and

icSPORTS 2025 - 13th International Conference on Sport Sciences Research and Technology Support

124

data fusion based on mpu9250 sensor. pages 1612–

1615. 12th IEEE International Conference on Elec-

tronic Measurement and Instruments (ICEMI), Qing-

dao, PEOPLES R CHINA, JUL 16-18, 2015.

Drobnic, M., Verdel, N., Holmberg, H.-C., and Supej, M.

(2023). The validity of a three-dimensional motion

capture system and the garmin running dynamics pod

in connection with an assessment of ground contact

time while running in place. SENSORS, 23(16).

Flores-Morales, V. H., Contreras-Bermeo, B. G., Bueno-

Palomeque, F. L., and Serpa-Andrade, L. J. (2016).

Analysis of a mobile system to register the kinematic

parameters in ankle, knee, and hip based in inertial

sensors. In International Congress on Sport Sciences

Research and Technology Support, volume 2, pages

29–33. SCITEPRESS.

Krishnaveni, B. V., Reddy, K. S., and Reddy, P. R. (2022).

Indoor tracking by adding imu and uwb using un-

scented kalman ﬁlter. Wireless Personal Communi-

cations, 123(4):3575–3596.

Lewin, M., Price, C., and Nester, C. (2022). Validation

of the runscribe inertial measurement unit for walking

gait measurement. PLOS ONE, 17(8).

Madgwick, S. O. H. (2014). Ahrs algorithms and calibration

solutions to facilitate new applications using low-cost

mems.

Mali, N. P. and Dey, S. (2020). Modern technology and

sports performance: An overview. International Jour-

nal of Physiology, 5(1):212–216.

Moya Velasco, W. P. (2019). Implementaci

on de un al-

goritmo de localizaci

on en interiores para resolver el

problema de ubicaci

on de objetos usando tecnolog

ıa

bluetooth low energy. B.S. thesis, Quito, 2019.

Pavei, G., Cazzola, D., Torre, A. L., and Minetti, A. E.

(2019). Race walking ground reaction forces at in-

creasing speeds: A comparison withwalking and run-

ning. Symmetry, 11.

Pricone, M. and Caracas, A. (2014). A heterogeneous

rssi-based localization system for indoor and outdoor

sports activities. In Proceedings of the IEEE Con-

ference on Local Computer Networks, Milano, Italy.

IEEE.

Ruiz Zambrano, J. L. and Villalta Encalada, R. I. (2024).

Dise

no de prototipo de medici

on de zancada en mar-

cha ol

ımpica. B.S. thesis.

Taborri, J., Palermo, E., and Rossi, S. (2019). Automatic

detection of faults in race walking: A comparative

analysis of machine-learning algorithms fed with in-

ertial sensor data. Sensors 2019, Vol. 19, Page 1461,

19:1461.

Ter

an Pineda, D. F. (2017). Dise

no e implementaci

de un sistema para visualizar la marcha humana

biomec

anica en la afectaci

on de rodilla ante una go-

nartrosis.

Vikas, C. M., Rajendran, S., Pattar, A., Jamadagni, H. S.,

and Budihal, R. (2016). Wiﬁ rssi and inertial sen-

sor based indoor localisation system: a simpliﬁed hy-

brid approach. 2016 INTERNATIONAL CONFER-

ENCE ON SIGNAL AND INFORMATION PROCESS-

ING (ICONSIP).

Wang, Y., Li, N., Chen, X., and Liu, M. (2014). Design

and implementation of an ahrs based on mems sensors

and complementary ﬁltering. Advances in Mechanical

Engineering, 2014:Article ID 214726.

Wu, F., Xing, J., and Dong, B. (2015). An indoor local-

ization method based on rssi of adjustable power wiﬁ

router. Fifth International Conference on Instrumen-

tation & Measurement, pages 1481–1487.

Zrenner, M., Gradl, S., Jensen, U., Ullrich, M., and Es-

koﬁer, B. M. (2018). Comparison of different algo-

rithms for calculating velocity and stride length in

running using inertial measurement units. Sensors,

18(12):4194.

Step Length Measurement Through Signal Power Analysis and Accelerometer Device: A Laboratory Comparison

125