Marco Sousa

Dr. Marco Sousa

orcid

Research Focus & Metrics

Academic Career Length

17Years (2010-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
25 July 2022 in IEEE Access
Verified
Journal Article
Open access
MSMarco Sousa
Marco Sousa
Pedro VieiraPedro Vieira
Pedro Vieira
MQMaria Paula Queluz
Maria Paula Queluz
ARAntónio Rodrigues
António Rodrigues
The performance of any Mobile Wireless Network (MWN) is dependent on the appropriate level of radio coverage, with Path Loss (PL) models being a valuable resource for its evaluation. Recently, advancements in Machine Learning (ML) and Deep Neural Networks (DNNs) have been applied to radio propagation to produce new data-driven PL models. Notoriously, these advancements have also allowed the inclusion of non-classical inputs, such as satellite images. However, data-driven PL models are often developed under the assumption that training and test data distributions are similar, which is a weak assumption in real-world scenarios. Thus, generalization (i.e., the model’s ability to perform on different data distributions) is a crucial aspect of data-driven PL models in the context of Mobile Network Operators (MNOs). This paper proposes a new data-driven PL model, the Ubiquitous Satellite Aided Radio Propagation (USARP) model, developed to enhance the geographical generalization capabilities of empirical PL models, by using satellite images. The USARP model considers self-supervised learning to extract general data representations of the radio environment from satellite images, improving the PL prediction Root Mean Square Error (RMSE) of the 3rd Generation Partnership Project (3GPP) PL model in the order of 9 dB, and for a data distribution distinct from the training data. Moreover, it was demonstrated the potential of the USARP model in terms of geographical and radio environment generalization. Although the generalization capabilities of ML regression algorithms are limited, the chosen USARP architecture and the use of regularization techniques had a positive impact on its geographical generalization performance.

29 January 2025 in IEEE Access
Verified
Journal Article
Open access
LMLuís Mata
Luís Mata
MSMarco Sousa
Marco Sousa
Pedro VieiraPedro Vieira
Pedro Vieira
MQMaria Paula Queluz
Maria Paula Queluz
ARAntónio Rodrigues
António Rodrigues
The urgent need to meet Environmental, Social, and Governance net-zero commitments and the financial risks posed by rising energy costs, are placing increasing pressure on Mobile Network Operators (MNOs) to optimise energy use. Given the significant energy consumption from Base Stations (BSs) and the fact that Fifth Generation (5G) deployments are not meeting expectations for reduced energy consumption when compared to Fourth Generation (4G), MNOs are shifting from a load-centric optimisation approach to a service-centric paradigm that balances spectral efficiency and energy efficiency. This paper introduces the Energy Sustainability Framework, which provides a composite index to evaluate the spectral and energy efficiencies of BSs, summarised across four classes: A, B, C and D. This framework supports network auditing, benchmarking, monitoring, and energy efficiency optimization while ensuring the desired Quality of Service (QoS). By leveraging Machine Learning (ML) and Explainable Artificial Intelligence (XAI) techniques applied to live data - including Performance Management (PM) and Energy Management (EM) metrics - integrated with domain expertise, the framework goes beyond deterministic computation to identify the root-cause factors influencing the BSs’ energy class, thereby enabling targeted optimisation. The framework can be flexibly applied at the location (site), BS (cell), or equipment type (hardware) levels. Notably, the implemented ML and XAI model achieved an F1-Score of 0.78 for 5G scenarios and 0.83 for 4G scenarios in predicting the Energy Sustainability Class, relying on a reduced set of indicators encompassing the most influential factors. An additional analysis, based on domain expertise, provided evidence of potential causal relationships between the class and observed variables involving channel quality, Multiple Input Multiple Output (MIMO) utilisation, network topology and energy-saving functionalities. For example, the results show that using higher order modulation coding schemes reduces the probability of a BS falling into class D by 50%. These findings provide practical insights for developing more sustainable network operations.

18 June 2025 in IEEE Access
Verified
Journal Article
Open access
MCMadalena Cilínio
Madalena Cilínio
TSThaína Saraiva
Thaína Saraiva
MSMarco Sousa
Marco Sousa
Pedro VieiraPedro Vieira
Pedro Vieira
ARAntónio Rodrigues
António Rodrigues
Anomaly detection is a critical component of Self-Organizing Networks (SON), enhancing network efficiency and resilience. This paper proposes a novel framework for detecting collective anomalies in univariate Key Performance Indicators (KPIs) in mobile networks. By leveraging data mining and Machine Learning (ML) techniques, the framework enables timely anomaly detection without requiring expert-labeled data. The proposed framework starts by clustering time series from 11 distinct KPIs into four groups. Then, representative KPIs from each cluster are selected to evaluate the anomaly detection performance using two algorithms: the Smart Trouble Ticket Management (STTM) and STUMPY. The STTM algorithm is applied to KPIs with low variability, such as Call Setup Success Rate and Service Drop Rate, showing high accuracy in anomaly detection. For KPIs with higher variability, such as User Downlink (DL) Average Throughput and DL Resource Block Utilization Rate, the STUMPY algorithm is employed, yielding similarly accurate results. The framework, composed of the STTM and the STUMPY algorithms, demonstrates effective anomaly detection, achieving high precision (0.94), recall (0.86), and an F1-score of 0.90, with minimal False Positives (FP). These results underline the framework’s reliability across different types of KPIs, providing a robust solution for anomaly detection in mobile network monitoring, outperforming benchmark algorithms in all key metrics.

1 February 2024 in IEEE Access
Verified
Journal Article
Open access
LMLuís Mata
Luís Mata
MSMarco Sousa
Marco Sousa
Pedro VieiraPedro Vieira
Pedro Vieira
MQMaria Paula Queluz
Maria Paula Queluz
ARAntónio Rodrigues
António Rodrigues
On the road to the sixth generation of cellular networks (6G), the need to ensure a sustainable usage of natural resources, amid increased competition and cost pressures, has driven the adoption of Self-Healing Mobile Networks to enhance operational efficiency of current and future wireless networks. This paradigm shift relies on Artificial Intelligence (AI) to increase automation of network functions, notably by applying predictive fault detection and automatic root-cause analysis. In this context, this paper proposes a Deep Learning (DL) model for self-healing operations based on a Spatial Graph Convolutional Neural Network (SGCN), which is applied to evaluate the performance degradation of Base Stations (BSs) and uncover the underlying root-causes. The advantages of the proposed DL model are threefold. Firstly, it is especially suited for wireless network applications, leveraging the SGCN to account for spatial dependencies among BSs and their physical characteristics. Secondly, the proposed model offers the flexibility to process diverse types of predictive features, including Performance Management (PM), Fault Management (FM), or other data types. Thirdly, it incorporates an explainability module that pinpoints the input features, such as PM counters, with the most significant influence on BS performance, thereby shedding light on its root-cause factors. The proposed model was evaluated on a live 4G network dataset and the results confirmed its effectiveness in identifying BS performance degradation. An F1-score of 89.6% was achieved in the classification of performance failures, which includes a 27% reduction in false negatives compared to prior research outcomes. In a live network environment, this reduction translates into substantial improvements in Quality of Experience (QoE) for the end users and cost savings for the Mobile Network Operators (MNOs).

27 March 2015 in Microwave and Optical Technology Letters
Verified
Journal Article
Open access
PDF file
MGMaria Giraldi
Maria Giraldi
CFCindy S. Fernandes
Cindy S. Fernandes
MFMarta Ferreira
Marta Ferreira
MSMarco Sousa
Marco Sousa
Pedro JorgePedro Jorge
Pedro Jorge
Orlando FrazãoOrlando Frazão
Orlando Frazão
In this work, it is proposed a technique to implement an intensity sensor based on the generation of a double-reflecting (ghost) signal in optical time domain reflectometry (OTDR). The intensity sensor is supported by a singlemode-multimode-singlemode (SMS) fiber structure combined with a fiber loop mirror (FLM). The results of the displacement sensitivity show linear behavior for both the first-reflecting and double-reflecting signals with linear slopes of approximately −4.5 dB/mm and −6 dB/mm, respectively. The displacement resolution achieved is ∼0.28 mm. It is also found that the system is able to read periodic displacement variations in the millisecond time scale applied to the sensing head. © 2015 Wiley Periodicals, Inc. Microwave Opt Technol Lett 57:1312–1315, 2015

21 August 2021 in Applied Sciences
Verified
Journal Article
Open access
PDF file
ALAdriano Lopes
Adriano Lopes
JOJoão Oliveira
João Oliveira
PSPedro Sebastiao
Pedro Sebastiao
MSMarco Sousa
Marco Sousa
Pedro VieiraPedro Vieira
Pedro Vieira
Mobile networks management is increasingly critical due to heavy communications usage by customers and complex due to the multiple technologies and systems deployed. Thus, Mobile Network Operators (MNOs) are constantly looking for better software solutions and tools to help them increase network performance and manage their networks more efficiently. In this paper, we present a modular web-based software solution to tackle problems related to mobile network planning, operation and optimization. The solution is focused on a set of functional requirements carefully chosen to support the network life cycle management, from planning to Operation and Maintenance (OAM) and optimisation stages. Based on a 3-tier modular architecture and implemented using only open-source software, the solution handles multiple data sources (e.g., Drive Test (DT) and Performance Management (PM)) and multiple Radio Access Network (RAN) technologies. MNOs can explore all available data through a flexible and user-friendly web interface, that also includes map-based visualization of the network. Moreover, the solution incorporates a set of recently developed and validated RAN algorithms, supporting tasks of network diagnosis, optimization, and planning. Also, with the purpose of optimizing the network, MNOs can investigate network simulations, using the RAN algorithms, of how the network will behave under certain conditions, and visualize the outcome of those simulations.