Exploring LoRa Signal Propagation in Indoor

and Outdoor Environments: A Comparative

Study
Bhanu Pratap Reddy Bhavanam1 Prashanth Ragam
School of Computer Science and School of Computer Science and
Engineering Engineering
VIT-AP University VIT-AP University
Inavolu, Beside AP secretariat, Inavolu, Beside AP secretariat,
Amaravati, AP, India Amaravati, AP, India
[email protected]
2024 IEEE International Conference on Smart Power Control and Renewable Energy (ICSPCRE) | 979-8-3503-7700-2/24/$31.00 ©2024 IEEE | DOI: 10.1109/ICSPCRE62303.2024.10675046

[email protected]

Abstract—Wireless technology has emerged as a pivotal its growth necessitates efficient and reliable wireless
enabler of modernization across various domains, significantly communication protocols .Figure 1 demonstrates the work
transforming the landscape of communication and flow in IoT.
connectivity. The advent of the Internet of Things (IoT) has
catalyzed this transformation, introducing a myriad of
innovative solutions that leverage different Radio-Frequency
(RF) protocols. Among these, Long Range (LoRa) has emerged
as a leading contender, distinguished by its utilization of the
Industrial, Scientific, and Medical (ISM) band, its low power
consumption, and its extended range capabilities. These
attributes have positioned LoRa at the forefront of IoT
communication, demonstrating its versatility and effectives
across a wide range of applications. This study undertaken a
comparative analysis of LoRa signal propagation in Indoor-to-
“Indoor (I2I) and Outdoor-to-Outdoor (O2O) environments,
specifically focusing on signal attenuation. The research is
conducted at VIT University, Andhra Pradesh, and involves a Fig.1. How the IoT works
case study that examines the performance of LoRa in both
indoor and outdoor settings. For the I2I scenario, the study Among these protocols [4], Radio Frequency (RF)
measures received signal strength (RSSI) and signal-to-noise technologies play a pivotal role. They offer a robust and
ratio (SNR) over a distance of 72.9 meters between transmitter scalable approach to data transmission, enabling
and receiver located on 8 floors building. In contrast, the O2O
communication between devices without the constraints of
scenario assesses LoRa’s performance over a distance of 1050
meters in an open ground environment, with RSSI and SNR of
physical cables. Within the realm of RF protocols, LoRa
-129 and -10 respectively. The findings of this comparative (Long Range) stands out for its unique capabilities. LoRa
study reveal that LoRa exhibits greater signal attenuation in prioritizes long-range, low-power communication, making it
indoor environments compared to outdoor settings, there by ideal for applications where battery life and extended
highlighting the technology’s unique characteristics and coverage are paramount. This comparative study delves into
potential applications in both indoor and outdoor the propagation characteristics of LoRa signals in both
communications scenarios. indoor and outdoor environments [5]. By understanding how
LoRa behaves in these contrasting settings, we can optimize
Key Words-LoRa, Indoor to Indoor, Outdoor to Outdoor
its use for various IoT deployments [6].
TABLE I LIST of Abbreviations LoRa was introduced by Cycleo a French company in the
year 2009 and later in 2012 it was adopted by Semtech and
Acronym Full Form made an open standard for all LPWAN communication. It
LoRa Long Range stands for long range which demolishes several drawbacks
I2I Indoor to Indoor in wireless networks to support IoT in various applications.
O2O Outdoor to Outdoor
CR Coding Rate From all over the world members working together for
SF Spreading Factor LoRa have formed an open nonprofit organization named as
BW Bandwidth LoRa Alliance it is also reputed from the telecommunication
ISM Industrial, Scientific, and Medical sector as International Telecommunication Union(ITU-
CSS Chirp Spread Spectrum
T).LoRaWAN functions by sub-GHZ spectrum works for
ISM (Industrial Scientific and Medical (ISM) without taking
I. INTRODUCTION any licenses it can be used for free by anyone [7].
The amplest feature of LoRa [8] [9] was the chirp
The evolution of communication technologies has seen a
spread spectrum (CSS) as a modulation approach facilitates
dramatic shift from reliance on wired connections to the
for Long range communication with less power
freedom and flexibility of wireless connections [1]. This
consumption. The spreading factor (SF) can be opted from
transition has been driven by the ever-increasing demand for
SF7 to SF12 it differs from a higher number that provides
data connectivity in a world dominated by the Internet of
long distance with less date rate support which spreads its
Things (IoT) [2]. The IoT [3] refers to the vast network of
signal over a wide frequency band. The appropriate stability
interconnected devices collecting and exchanging data and

979-8-3503-7700-2/24/$31.00 ©2024 IEEE

Authorized licensed use limited to: UNIVERSIDADE DE SAO PAULO. Downloaded on December 11,2024 at 07:54:59 UTC from IEEE Xplore. Restrictions apply.
for the fulfillment of the LoRa device can be accomplished nodes near roads. They utilized a reputation model based on
by picking out the suitable spreading factor for the beta distributions for association duration and a Bayesian
application as required. To deploy loRa devices in different Game strategy for private information collection. The
countries accordingly have to select the specified spreading experiment, named Traffic Information Acquisition with
factors for each region of the country and Some of the LoRa (TILR), involved 32 LoRa nodes, 3 gateways, and a
specimens are for Europe 868 MHZ and 433 MHZ bands video camera, achieving 92% accuracy for right or left
proposed by the European telecommunications standards turns, 94% for U-turns, 91% for direction, 90% for speed
institute (ETSI), for the United sates915 MHZ,433 MHZ estimation, and 90% for road crossings, without multi-hop
bands proposed by federal communication commission communication to conserve battery life. Preti Kumari et al
(FCC). developed an energy-efficient smart metering system using
LoRa communication protocols open the way to edge computing with LoRaWAN, featuring a star-of-stars
establish the connection between LoRa devices and network architecture. They utilized machine learning for
loRaWAN gateways to communicate with each other. Based data reduction and proposed a strategy for selecting
on the necessity of the application as it depends upon power spreading factors in LoRa networks. The system achieved a
utilization and latency, it can be selected among LoRaWAN 95% success rate in transferring Energy Multivariate Time
class A, class B [10], and class C Protocols [11]. Class A Series (EMS) data, using Python and TensorFlow for
protocol is used for the applications when they require less simulation. The hardware included a Raspberry Pi-3, an
power utilization and more latency. Class B is used for RFM95W-868S2 module, and a Dragino LG01-SIOT
applications that require more power utilization and less gateway, with power utilization recorded at 2µJ, 32µJ, and
response time than Class A. class c applications are used 11µJ for idle, transmission, and reception phases,
when require high power utilization and quick response respectively.
time. Adoptive Data Rate [12] [is the dominant feature of
LoRa it keeps a stable communication link for this it allows
the devices to monitor their SNR and RSSI of arriving II. EXPERIMENTAL SITE SPECIFICATIONS: INDOOR AND
messages and then accordingly they can tune their data rate, OUTDOOR LOCATIONS
SF, and transmit power to increase performance, and The attenuation of LoRa signals exhibits considerable
preserve battery life. The fulfillment of ADR relies on variation across distinct environments, including those
several parameters including physical distances among influenced by climatic conditions [16]. This case study aims
devices, network density of devices, and obstacles in the to understand the LoRa signal strength within the chosen
network surroundings which shows the best results of ADR. environment, highlighting LoRa’s distinctive capability for
I. Literature Review Long-Range communication. For the experimental
Numerous applications came in to utilization of LoRa configuration, the study was conducted at VIT-AP
for a wide array of domains. From smart agriculture to University, situated in inavolu, beside AP Secretariat in
urban infrastructure monitoring, LoRa has been deployed in Amaravathi, Andhra Pradesh, India. Specifically, a
various scenarios to facilitate reliable and cost-effective designated building was chosen for the I2I environment,
communication. This literature review aims to explore the while open ground was selected for the O2O environment,
diverse applications of LoRa technology, focusing on its facilitating the range test of LoRaWAN.
utilization in both indoor and outdoor environments. By
Faber et al [13] underscores LoRa's advantages over NB- III. PREPARING LORA FOR EXPERIMENTATION
IoT and LTE-M, particularly in free licensed bands,
recommending parameter optimizations and redundancy A. Configuring LoRa Devices
techniques for reliable communication. Experimental To ensure a systematic range test has to be conducted
findings demonstrate LoRaWAN's efficacy in urban (3KM) before the field test, the peripherals in table are utilized, as
and rural (10KM) settings, with average signal strengths and listed in the table 1 below.
Signal-to-Noise Ratios (SNR) outlined. Integration of Chirp
Spread Spectrum (CSS) modulation and Forward Error TABLE II LIST of PHERIPHERALS
Correction (FEC) methods like Hamming Code enhance
LoRa's performance under varied channel conditions. Equipment name Quantity
Experimental setups incorporate GPS-equipped transmitters Universal Expansion Board 2
Pycom board with Antennas 2
and LoRa gateways for robust signal measurement and
Power Bank 1
analysis, laying a foundation for optimized LoRa-based Plastic storage boxes 2
communication systems. Ekaterina et al [14] evaluated
LoRaWAN's performance at 868 MHz across various
environments at Brno University using LoRa Nodes and Two pycom boards, model Smtech SX1276, were used for
Gateways. They employed the LG3308 GW, routers, FTDs,
the field test. One functioned as the transmitter, while the
and a display unit for field tests. They proposed GW
Trilateration and Weighted Centroid for localization, tested other acted as the receiver. To establish communication,
with ML algorithms, and found mean localization errors of 5 each board was individually connected to a laptop for
meters underground, 6.6 meters in open ground, and 4 configuration. MicroPython 3 served as the programming
meters outdoor. environment, providing access to built-in functionalities and
Kumari et al [15] conducted an experiment to detect essential Python libraries optimized for interacting with
reckless driving and estimate vehicle speed using embedded

Authorized licensed use limited to: UNIVERSIDADE DE SAO PAULO. Downloaded on December 11,2024 at 07:54:59 UTC from IEEE Xplore. Restrictions apply.
Pycom boards. The following table II shows the basic
parameters configured with their specific values.

Table III. Reference for Configuration

Parameter Configured value

Transmitted Power 14 dBm
Spreading Factor (SF) 7
Bandwidth (BW) 125 KHz
Height of Anteena 0.19 m
Coding Rate 4/5
Frequency 868 MHZ

The deployment procedure for the LoRa network

necessitates prior configuration to initiate communication
between the transmitter and receiver. This setup is crucial Fig 3. Indoor-to-Indoor (I2I) LoRa Receiver Setup
for collecting RSSI and SNR values, which are essential for
Table IV Sample I2I RSSI and SNR measurements at various Distances
demonstrating signal attenuation. Before proceeding to note
live readings, it is imperative to conduct empirical checks. Distance (m) RSSI (dB) SNR (dB)
1 -42 6
5 -62 5
B. Evaluating LoRa Range in I2I Environments 40.8 -99 7
48.9 -124 -6
39.9 -103 0
Wireless communication within indoor environments is 22.12 -87 6
inherently challenged by various obstacles, including walls, 8 -85 6
floors, and structural elements, all of which contribute to 43.8 -110 5
signal attenuation. To comprehensively evaluate the range 51.9 -124 -6
42.9 -112 3
of LoRa technology in Indoor-to-Indoor (I2I) environments,
an experimental setup was devised. Initially, the transmitter C. Evaluating LoRa Range in O2O Environments
LoRa setup was positioned on the first floor in room number This section presents the experimental procedure of
101 as demonstrated in figure1, subsequently; the receiver LoRa signal propagation in outdoor settings, contrasting its
LoRa mobility setup was systematically relocated to behavior with that observed in I2O scenarios. Unlike the
designated points on the second floor at predetermined confined structure of indoor spaces, outdoor environments
intervals as shown in figure2, At each position, the Received offer greater freedom for signal propagation, although
Signal Strength Indication (RSSI) and Signal-to-Noise Ratio susceptible to external influences such as temperature,
(SNR) were recorded as shown in Table IV. This process humidity. The experimental methodology involves the fixed
was iterated across all eight floors, with the transmitter setup position of the transmitter setup while the receiver LoRa
remaining fixed while the receiver setup was sequentially setup is mobilized to various points within the outdoor space
moved to distinct locations on each floor. The experimental as shown in the below figure2. Real-time readings of
setup configuration is illustrated in the figure1. Notably, the Receiver Signal Strength Indication (RSSI) and Signal-to-
first setup pertains to the stationary transmitter setup, while Noise (SNR) are exactly recorded along with corresponding
the LoRa receiver setup is navigated to different points for distance measurements as shown in table V.
each floor, with corresponding RSSI and SNR values
carefully documented.

Fig.4. Outdoor-to-Outdoor (O2) LoRa Transmitter Setup

Fig.2. Indoor-to-Indoor (I2I) LoRa Transmitter Setup

Authorized licensed use limited to: UNIVERSIDADE DE SAO PAULO. Downloaded on December 11,2024 at 07:54:59 UTC from IEEE Xplore. Restrictions apply.
 Received Signal Strength Indicator (RSSI) averages at -
115.97dBm, showing weak signal strength.

Fig 5. Outdoor-to-Outdoor (O2) LoRa Receiver Setup

Table V Sample O2O RSSI and SNR measurements at various Distances

Fig.6. RSSI analysis of I2I LoRa communication
Distance (m) RSSI (dB) SNR (dB)

0 6 -51
50 -1 -116
100 -8 -125
150 -7 -126
200 -7 -127
300 -8 -127
350 -10 -130
400 -10 -130
450 -10 -129

IV. VISUALIZING RESULTS: GRAPHICAL ANALYSIS OF

EXPERIMENTAL DATA
A. Analysis of I2I Experimental Data: Graphical
Insights

The subsequent analysis focuses on the examination of Fig.7. SNR analysis of I2I LoRa communication
Received Signal Strength Indication (RSSI) and Signal-to-
Noise Ratio (SNR) within Indoor-to-Indoor (I2I) Table VI Statistical overview of I2I LoRa signal data
environments as shown in figure 3 and figure 4.Graphical
representations depict the trends observed in RSSI and SNR Parameter MIN MAX AVG STDEV
values concerning distance variations between the
Distance 1.0 72.9 42.3799 18.7171
transmitter and receiver setups. In the first graph, the x-axis SNR -11 7 -3.0652 5.2425
denotes the distance between the transmitter and receiver, RSSI -130 -42 -115.978 17.6792
while the y-axis illustrates the corresponding RSSI values.
Similarly, the second graph illustrates the relationship
between distance and SNR values, with the x-axis
representing distance and the y-axis displaying SNR values.
A discernible trend emerges in both graphs, indicating
degradation in RSSI and SNR values as the distance
between the transmitter and receiver increases. This
phenomenon is attributable to the inherent limitations posed
by indoor environments, wherein signal integrity is
compromised due to structural interfaces between the
transmitter and receiver setups. From table VI the average
distance between devices is approximately 42.37 meters,
with a standard deviation of 18.71 meters, indicating a wide
variability in signal strength across different measurements.
The Signal-to-Noise Ratio (SNR) averages around -3.06 dB, Fig 8: Performance Metrics Comparison across Different Regression
suggesting a relatively noisy environment, while the Models for I2I

Authorized licensed use limited to: UNIVERSIDADE DE SAO PAULO. Downloaded on December 11,2024 at 07:54:59 UTC from IEEE Xplore. Restrictions apply.
 The histogram figure 8 presents a comparative analysis
of four regression models applied to I2I environment
highlighting their performance through key statistical
metrics. Each model aims to predict the relationship
between independent variables and a dependent
variable of interest. Multiple Linear Regression
demonstrates an MAE of 3.7821, MSE of 20.0884,
RMSE of 4.4820, and an R² of 0.5641. Polynomial
Regression exhibits improved performance in terms of
MAE of 3.3392, MSE of 26.1911 RMSE of 5.1177 and
R² of 0.4317.Ridge Regression shows MAE of 3.7873,
MSE of 20.1536, RMSE of 4.4892 and R² of 0.5627.
Lasso Regression projects nearest to ridge regression as
MAE of 3.7880, MSE of 20.1675, and RMSE of 4.4908
and R² of 0.5624.
Fig.10. SNR analysis of I2I LoRa communication
B. Analysis of O2O Experimental Data: Graphical
Insights Table VII Statistical overview of O2O LoRa signal data

The subsequent section delves into the scrutiny of Parameter MIN MAX AVG ST DEV
Outdoor-to-Outdoor (O2O) experimental data, presenting Distance 0 1050 594.0625 297.9159
SNR -12 6 -8.1875 3.5960
graphical insights gather from the analysis. Graphical RSSI -132 -51 -125.3125 13.9363
representations provide a fine understanding of the observed
trends in signal parameters among varying distances
between the transmitter and receiver setups. The first graph
depicts the relationship between distance and Received
Signal Strength Indication (RSSI), while the second graph
elucidates the interplay between distance and Signal-to-
Noise Ratio (SNR).A salient observation emerges from the
graphical representations, revealing a discernible trend
characterized by a progressive decline in both RSSI and
SNR values as the spatial separation between the transmitter
and receiver increases. This phenomenon underscores the
influence of external environmental factors, including
terrain variations and atmospheric conditions, on signal
propagation within O2O environments.
Table VII summarized key signal statistics for different
parameters. The measured distances ranged from a
minimum of 0 to a maximum of 1050, with an average
distance of 594.0625 and a standard deviation of 297.9159.
SNR values varied between -12 and 6, with an average of -
8.1875 and a standard deviation of 3.5960. Finally, RSSI Fig 11: Performance Metrics Comparison across Different
measurements spanned from -132 to -51, with an average of Regression Models for O2O
-125.3125 and a standard deviation of 13.9363.
The histogram of Figure 11 compares the performance
of different regression models using several metrics.
Multiple Linear Regression has an MAE of 3.0289, an
MSE of 12.3246, an RMSE of 3.5106, and an R² of -
6.1894. Polynomial Regression shows better
performance with an MAE of 1.8287, an MSE of
5.0992, an RMSE of 2.2581, and an R² of -1.9745.
Ridge Regression records an MAE of 3.0006, an MSE
of 12.1707, an RMSE of 3.4788, and an R² of -6.0596.
Lasso Regression has an MAE of 3.0094, an MSE of
12.1707, an RMSE of 3.4886, and an R² of -6.0995
Fig.9. RSSI analysis of O2O LoRa communication

Authorized licensed use limited to: UNIVERSIDADE DE SAO PAULO. Downloaded on December 11,2024 at 07:54:59 UTC from IEEE Xplore. Restrictions apply.
V. INTERPRETING RESULTS: DISCUSSION ON EXPERIMENTAL specific environmental conditions can further bolster efforts
OUTCOMES to minimize signal attenuation. Collaborative
The experimental outcomes for the I2I and interdisciplinary research endeavors focusing on optimizing
O2Oscenario, comprising both RSSI (Received Signal LoRa technology's performance in challenging
Strength Indication) and SNR (Signal-to-Noise Ratio) environments are imperative for addressing this critical
values alongside corresponding distance measurements. The aspect of wireless communication.
analysis of these results offers valuable insights into the
behavior of LoRa signals within indoor, outdoor REFERENCES
environments, and their interaction with external settings.
Both RSSI and SNR values exhibit a progressive decline as
[1] Yang SC. Toward a wireless world. IEEE Technology and
the distance between the transmitter and receiver setups Society Magazine. 2007 Jun 11;26(2):32-42.
increases. For indoor the trend is indicative of signal [2] ] Lima WG, Lopes AV, Cardoso CM, Araújo JP, Neto MC,
attenuation, attributed to the presence of structural barriers Tostes ME, Nascimento AA, Rodriguez M, Barros FJ. LoRa
and obstructions within indoor spaces. Among the Technology Propagation Models for IoT Network Planning in
the Amazon Regions. Sensors. 2024
regression models tested, Multiple Linear Regression and
[3] Ma Kavre M, Gadekar A, Gadhade Y. Internet of Things (IoT):
Ridge Regression demonstrated better performance metrics, a survey. In2019 IEEE pune section international conference
with Polynomial Regression and Lasso Regression (PuneCon) 2019 Dec 18 (pp. 1-6). IEEE. r 1;24(5):1621.
following closely. Despite the fluctuations in signal strength [4] Al-Fuqaha A, Guizani M, Mohammadi M, Aledhari M, Ayyash
and noise, the regression models provided reasonably M. Internet of things: A survey on enabling technologies,
protocols, and applications. IEEE communications surveys &
accurate predictions, as evidenced by their error metrics and tutorials. 2015 Jun 15;17(4):2347-76.
R² values. [5] Ayele ED, Hakkenberg C, Meijers JP, Zhang K, Meratnia N,
Similarly for O2O through the analysis of these results, we Havinga PJ. Performance analysis of LoRa radio for an indoor
gain a deeper understanding of LoRa signal propagation IoT applications. In2017 International Conference on Internet of
characteristics within open outdoor environments and their Things for the Global Community (IoTGC) 2017 Jul 10 (pp. 1-
8). IEEE.
susceptibility to external influences. However, as the
[6] Ragam P, Nimaje DS. Performance evaluation of LoRa
distance between the transmitter and receiver expands, the LPWAN technology for IoT-based blast-induced ground
observed RSSI and SNR values progressively decline, vibration system. Journal of Measurements in Engineering. 2019
reflecting the cumulative effects of signal dispersion and Sep 30;7(3):119-33.
environmental interference encountered within outdoor [7] Raza U, Kulkarni P, Sooriyabandara M. Low power wide area
environments. Optimization measures such as antenna networks: An overview. ieee communications surveys &
tutorials. 2017 Jan 16;19(2):855-73.
configurations, transmission power adjustments, and
[8] Kjendal D. LoRa-Alliance regional parameters overview.
frequency channel allocations play a crucial role in Journal of ICT Standardization. 2021;9(1):35-46.
overcoming environmental challenges and maximizing the [9] San Cheong P, Bergs J, Hawinkel C, Famaey J. Comparison of
efficacy of LoRa technology in outdoor settings. Polynomial LoRaWAN classes and their power consumption. In2017 IEEE
Regression emerged as the most effective model, achieving symposium on communications and vehicular technology
(SCVT) 2017 Nov 14 (pp. 1-6). IEEE.
the lowest error metrics compared to the other models.
[10] Soto-Vergel A, Arismendy L, Bornacelli-Durán R, Cardenas C,
However, the negative R² values across all models suggest Montero-Arévalo B, Rivera E, Calle M, Candelo-Becerra JE.
that the regression models struggled to accurately capture LoRa performance in industrial environments: analysis of
the relationship between distance, RSSI, and SNR in the different ADR algorithms. IEEE Transactions on Industrial
O2O context. Informatics. 2023 Jan 31.
[11] Liando JC, Gamage A, Tengourtius AW, Li M. Known and
VI. CONCLUDING REMARKS: SUMMARIZING KEY FINDINGS unknown facts of LoRa: Experiences from a large-scale
measurement study. ACM Transactions on Sensor Networks
AND IMPLICATIONS (TOSN). 2019 Feb 21;15(2):1-35.
In conclusion, this comparative study highlights the [12] Elbsir HE, Kassab M, Bhiri S, Bedoui MH. Evaluation of
significant impact of LoRa technology on contemporary LoRaWAN class B performances and its optimization for better
support of actuators. Computer Communications. 2023 Jan
communication systems. Through a thorough examination 15;198:128-39.
of signal propagation in I2I and O2O environments, it [13] Faber, M. J., van der Zwaag, K. M., dos Santos, W. G. V.,
becomes evident that LoRa demonstrates remarkable Rocha, H. R. D. O., Segatto, M. E., & Silva, J. A. (2020). A
versatility and effectiveness across various contexts. LoRa's theoretical and experimental evaluation on the performance of
extended range capabilities and low power consumption LoRa technology. IEEE Sensors Journal, 20(16), 9480-9489
make it an essential component for numerous IoT [14] Svertoka E, Rusu-Casandra A, Burget R, Marghescu I, Hosek J,
Ometov A. LoRaWAN: Lost for localization?. IEEE Sensors
applications. These findings underscore LoRa's potential to Journal. 2022 Oct 11;22(23):23307-19.
facilitate seamless connectivity and communication in both [15] Kumari P, Gupta HP, Dutta T. A Bayesian game based approach
indoor and outdoor environments, contributing to ongoing for associating the nodes to the gateway in LoRa network. IEEE
advancements in wireless communication systems. Transactions on Intelligent Transportation Systems. 2021 Jan
8;23(5):4583-92.
VII. DIRECTIONS: PRIORITIZING AND SCALING FUTURE [16] Elijah O, Rahim SK, Sittakul V, Al-Samman AM, Cheffena M,
RESEARCH Din JB, Tharek AR. Effect of weather condition on LoRa IoT
communication technology in a tropical region: Malaysia. IEEE
Machine learning techniques improves accuracy and access. 2021 May 14;9:72835-43.
developing adaptive communication protocols tailored to

Authorized licensed use limited to: UNIVERSIDADE DE SAO PAULO. Downloaded on December 11,2024 at 07:54:59 UTC from IEEE Xplore. Restrictions apply.