arXiv subjects

João Morais

Publications and source records attributed to João Morais.

At least 19 recordsLinked to original sources

arXiv2026

Wireless Dataset Similarity: Measuring Distances in Supervised and Unsupervised Machine Learning

This paper introduces a task- and model-aware framework for measuring similarity between wireless datasets, enabling applications such as dataset selection/augmentation, simulation-to-real (sim2real) comparison, task-specific synthetic data generation, and informing decisions on model training/adaptation to new deployments. We evaluate candidate dataset distance metrics by how well they predict cross-dataset transferability: if two datasets have a small distance, a model trained on one should perform well on the other. We apply the framework on an unsupervised task, channel state information (CSI) compression, using autoencoders. Using metrics based on UMAP embeddings, combined with Wasserstein and Euclidean distances, we achieve Pearson correlations exceeding 0.85 between dataset distances and train-on-one/test-on-another task performance. We also apply the framework to a supervised beam prediction in the downlink using convolutional neural networks. For this task, we derive a label-aware distance by integrating supervised UMAP and penalties for dataset imbalance. Across both tasks, the resulting distances outperform traditional baselines and consistently exhibit stronger correlations with model transferability, supporting task-relevant comparisons between wireless datasets.

cs.LG↗

arXiv2025

Comparing Stochastic and Ray-tracing Datasets in Machine Learning for Wireless Applications

Machine learning for wireless systems is commonly studied using standardized stochastic channel models (e.g., TDL/CDL/UMa) because of their legacy in wireless communication standardization and their ability to generate data at scale. However, some of their structural assumptions may diverge from real-world propagation. This paper asks when these models are sufficient and when ray-traced (RT) data - a proxy for the real world - provides tangible benefits. To answer these questions, we conduct an empirical study on two representative tasks: CSI compression and temporal channel prediction. Models are trained and evaluated using in-domain, cross-domain, and small-data fine-tuning protocols. Across settings, we observe that stochastic-only evaluation may over- or under-estimate performance relative to RT. These findings support a task-aware recipe where stochastic models can be leveraged for scalable pre-training and for tasks that do not rely on strong spatiotemporal coupling. When that coupling matters, pre-training and evaluation should be grounded in spatially consistent or geometrically similar RT scenarios. This study provides initial guidance to inform future discussions on benchmarking and standardization.

eess.SP↗

arXiv2024

Localization in Digital Twin MIMO Networks: A Case for Massive Fingerprinting

Localization in outdoor wireless systems typically requires transmitting specific reference signals to estimate distance (trilateration methods) or angle (triangulation methods). These cause overhead on communication, need a LoS link to work well, and require multiple base stations, often imposing synchronization or specific hardware requirements. Fingerprinting has none of these drawbacks, but building its database requires high human effort to collect real-world measurements. For a long time, this issue limited the size of databases and thus their performance. This work proposes significantly reducing human effort in building fingerprinting databases by populating them with \textit{digital twin RF maps}. These RF maps are built from ray-tracing simulations on a digital replica of the environment across several frequency bands and beamforming configurations. Online user fingerprints are then matched against this spatial database. The approach was evaluated with practical simulations using realistic propagation models and user measurements. Our experiments show sub-meter localization errors on a NLoS location 95\% of the time using sensible user measurement report sizes. Results highlight the promising potential of the proposed digital twin approach for ubiquitous wide-area 6G localization.

cs.IT↗

arXiv2023

DeepSense 6G: A Large-Scale Real-World Multi-Modal Sensing and Communication Dataset

This article presents the DeepSense 6G dataset, which is a large-scale dataset based on real-world measurements of co-existing multi-modal sensing and communication data. The DeepSense 6G dataset is built to advance deep learning research in a wide range of applications in the intersection of multi-modal sensing, communication, and positioning. This article provides a detailed overview of the DeepSense dataset structure, adopted testbeds, data collection and processing methodology, deployment scenarios, and example applications, with the objective of facilitating the adoption and reproducibility of multi-modal sensing and communication datasets.

eess.SP↗

arXiv2022

Device-Agnostic Millimeter Wave Beam Selection using Machine Learning

Most research in the area of machine learning-based user beam selection considers a structure where the model proposes appropriate user beams. However, this design requires a specific model for each user-device beam codebook, where a model learned for a device with a particular codebook can not be reused for another device with a different codebook. Moreover, this design requires training and test samples for each antenna placement configuration/codebook. This paper proposes a device-agnostic beam selection framework that leverages context information to propose appropriate user beams using a generic model and a post processing unit. The generic neural network predicts the potential angles of arrival, and the post processing unit maps these directions to beams based on the specific device's codebook. The proposed beam selection framework works well for user devices with antenna configuration/codebook unseen in the training dataset. Also, the proposed generic network has the option to be trained with a dataset mixed of samples with different antenna configurations/codebooks, which significantly eases the burden of effective model training.

eess.SP↗

arXiv2022

Multi-Modal Beam Prediction Challenge 2022: Towards Generalization

Beam management is a challenging task for millimeter wave (mmWave) and sub-terahertz communication systems, especially in scenarios with highly-mobile users. Leveraging external sensing modalities such as vision, LiDAR, radar, position, or a combination of them, to address this beam management challenge has recently attracted increasing interest from both academia and industry. This is mainly motivated by the dependency of the beam direction decision on the user location and the geometry of the surrounding environment -- information that can be acquired from the sensory data. To realize the promised beam management gains, such as the significant reduction in beam alignment overhead, in practice, however, these solutions need to account for important aspects. For example, these multi-modal sensing aided beam selection approaches should be able to generalize their learning to unseen scenarios and should be able to operate in realistic dense deployments. The "Multi-Modal Beam Prediction Challenge 2022: Towards Generalization" competition is offered to provide a platform for investigating these critical questions. In order to facilitate the generalizability study, the competition offers a large-scale multi-modal dataset with co-existing communication and sensing data collected across multiple real-world locations and different times of the day. In this paper, along with the detailed descriptions of the problem statement and the development dataset, we provide a baseline solution that utilizes the user position data to predict the optimal beam indices. The objective of this challenge is to go beyond a simple feasibility study and enable necessary research in this direction, paving the way towards generalizable multi-modal sensing-aided beam management for real-world future communication systems.

eess.SP↗

arXiv2022

Position Aided Beam Prediction in the Real World: How Useful GPS Locations Actually Are?

Millimeter-wave (mmWave) communication systems rely on narrow beams for achieving sufficient receive signal power. Adjusting these beams is typically associated with large training overhead, which becomes particularly critical for highly-mobile applications. Intuitively, since optimal beam selection can benefit from the knowledge of the positions of communication terminals, there has been increasing interest in leveraging position data to reduce the overhead in mmWave beam prediction. Prior work, however, studied this problem using only synthetic data that generally does not accurately represent real-world measurements. In this paper, we investigate position-aided beam prediction using a real-world large-scale dataset to derive insights into precisely how much overhead can be saved in practice. Furthermore, we analyze which machine learning algorithms perform best, what factors degrade inference performance in real data, and which machine learning metrics are more meaningful in capturing the actual communication system performance.

eess.SP↗

arXiv2022

Location- and Orientation-aware Millimeter Wave Beam Selection for Multi-Panel Antenna Devices

While initial beam alignment (BA) in millimeter-wave networks has been thoroughly investigated, most research assumes a simplified terminal model based on uniform linear/planar arrays with isotropic antennas. Devices with non-isotropic antenna elements need multiple panels to provide good spherical coverage, and exhaustive search over all beams of all the panels leads to unacceptable overhead. This paper proposes a location- and orientation-aware solution that manages the initial BA for multi-panel devices. We present three different neural network structures that provide efficient BA with a wide range of training dataset sizes, complexity, and feedback message sizes. Our proposed methods outperform the generalized inverse fingerprinting and hierarchical panel-beam selection methods for two considered edge and edge-face antenna placement designs.

eess.SP↗

arXiv2022

Parkour Spot ID: Feature Matching in Satellite and Street view images using Deep Learning

How to find places that are not indexed by Google Maps? We propose an intuitive method and framework to locate places based on their distinctive spatial features. The method uses satellite and street view images in machine vision approaches to classify locations. If we can classify locations, we just need to repeat for non-overlapping locations in our area of interest. We assess the proposed system in finding Parkour spots in the campus of Arizona State University. The results are very satisfactory, having found more than 25 new Parkour spots, with a rate of true positives above 60%.

cs.CV↗

arXiv2021

Improving the compromise between accuracy, interpretability and personalization of rule-based machine learning in medical problems

One of the key challenges when developing a predictive model is the capability to describe the domain knowledge and the cause-effect relationships in a simple way. Decision rules are a useful and important methodology in this context, justifying their application in several areas, particularly in clinical practice. Several machine-learning classifiers have exploited the advantageous properties of decision rules to build intelligent prediction models, namely decision trees and ensembles of trees (ETs). However, such methodologies usually suffer from a trade-off between interpretability and predictive performance. Some procedures consider a simplification of ETs, using heuristic approaches to select an optimal reduced set of decision rules. In this paper, we introduce a novel step to those methodologies. We create a new component to predict if a given rule will be correct or not for a particular patient, which introduces personalization into the procedure. Furthermore, the validation results using three public clinical datasets suggest that it also allows to increase the predictive performance of the selected set of rules, improving the mentioned trade-off.

cs.LG↗

arXiv2021

A New Approach for Interpretability and Reliability in Clinical Risk Prediction: Acute Coronary Syndrome Scenario

We intend to create a new risk assessment methodology that combines the best characteristics of both risk score and machine learning models. More specifically, we aim to develop a method that, besides having a good performance, offers a personalized model and outcome for each patient, presents high interpretability, and incorporates an estimation of the prediction reliability which is not usually available. By combining these features in the same approach we expect that it can boost the confidence of physicians to use such a tool in their daily activity. In order to achieve the mentioned goals, a three-step methodology was developed: several rules were created by dichotomizing risk factors; such rules were trained with a machine learning classifier to predict the acceptance degree of each rule (the probability that the rule is correct) for each patient; that information was combined and used to compute the risk of mortality and the reliability of such prediction. The methodology was applied to a dataset of patients admitted with any type of acute coronary syndromes (ACS), to assess the 30-days all-cause mortality risk. The performance was compared with state-of-the-art approaches: logistic regression (LR), artificial neural network (ANN), and clinical risk score model (Global Registry of Acute Coronary Events - GRACE). The proposed approach achieved testing results identical to the standard LR, but offers superior interpretability and personalization; it also significantly outperforms the GRACE risk model and the standard ANN model. The calibration curve also suggests a very good generalization ability of the obtained model as it approaches the ideal curve. Finally, the reliability estimation of individual predictions presented a great correlation with the misclassifications rate. Those properties may have a beneficial application in other clinical scenarios as well. [abridged]

cs.LG↗

arXiv2021

Reduced-quaternionic Mathieu functions, time-dependent Moisil-Teodorescu operators, and the imaginary-time wave equation

We construct a one-parameter family of generalized Mathieu functions, which are reduced quaternion-valued functions of a pair of real variables lying in an ellipse, and which we call $λ$-reduced quaternionic Mathieu functions. We prove that the $λ$-RQM functions, which are in the kernel of the Moisil-Teodorescu operator $D+λ$ ($D$ is the Dirac operator and $λ\in\mathbb{R}\setminus\{0\}$), form a complete orthogonal system in the Hilbert space of square-integrable $λ$-metamonogenic functions with respect to the $L^2$-norm over confocal ellipses. Further, we introduce the zero-boundary $λ$-RQM-functions, which are $λ$-RQM functions whose scalar part vanishes on the boundary of the ellipse. The limiting values of the $λ$-RQM functions as the eccentricity of the ellipse tends to zero are expressed in terms of Bessel functions of the first kind and form a complete orthogonal system for $λ$-metamonogenic functions with respect to the $L^2$-norm on the unit disk. A connection between the $λ$-RQM functions and the time-dependent solutions of the imaginary-time wave equation in the elliptical coordinate system is shown.

math.CV↗

arXiv2019

Relations among spheroidal and spherical harmonics

A contragenic function in a domain $Ω\subseteq\mathbf{R}^3$ is a reduced-quaternion-valued (i.e. the last coordinate function is zero) harmonic function, which is orthogonal in $L^2(Ω)$ to all monogenic functions and their conjugates. The notion of contragenicity depends on the domain and thus is not a local property, in contrast to harmonicity and monogenicity. For spheroidal domains of arbitrary eccentricity, we relate standard orthogonal bases of harmonic and contragenic functions for one domain to another via computational formulas. This permits us to show that there exist nontrivial contragenic functions common to the spheroids of all eccentricities.

math.CV↗

arXiv2019

The interacting multiverse and its effect on the cosmic microwave background

We study a toy model of a multiverse consisting of canonically quantized universes that interact with each other on a quantum level based on a field-theoretical formulation of the Wheeler-DeWitt equation. This interaction leads to the appearance of a pre-inflationary phase in the evolution of the individual universes. We analyze scalar perturbations within the model and calculate the influence of the pre-inflationary phase onto the power spectrum of these perturbations. The result is that there is a suppression of power on large scales, which can describe well the Planck 2018 data for the cosmic microwave background anisotropies and could thus indicate a possible solution to the observed quadrupole discrepancy.

gr-qc↗

arXiv2018

What if gravity becomes really repulsive in the future?

The current acceleration of the Universe is one of the most puzzling issues in theoretical physics nowadays. We are far from giving an answer on this letter to its true nature. Yet, with the observations we have at hand, we analyse the different patterns that the gravitational potential can show in the future. Surprisingly, gravity not only can get weaker in the near future, it can even become repulsive; or equivalently, the gravitational potential may become negative. We show this remark by using one of the simplest phenomenological model we can imagine for dark energy. We have as well reviewed the statefinder approach of these models. For completeness, we have also showed the behaviour of the density contrast of dark matter and dark energy for these simple (yet illustrative models). Our results are displayed at present and how they evolve in the future.

gr-qc↗

arXiv2018

Pre-inflation from the multiverse: Can it solve the quadrupole problem in the cosmic microwave background?

We analyze a quantized toy model of a universe undergoing eternal inflation using a quantum-field-theoretical formulation of the Wheeler-DeWitt equation. This so-called third quantization method leads to the picture that the eternally inflating universe is converted to a multiverse in which sub-universes are created and exhibit a distinctive phase in their evolution before reaching an asymptotic de Sitter phase. From the perspective of one of these sub-universes, we can thus analyze the pre-inflationary phase that arises naturally. Assuming that our observable universe is represented by one of those sub-universes, we calculate how this pre-inflationary phase influences the power spectrum of the cosmic microwave background (CMB) anisotropies and analyze whether it can explain the observed discrepancy of the power spectrum on large scales, i.e. the quadrupole issue in the CMB. While the answer to this question is negative in the specific model analyzed here, we point out a possible resolution of this issue.

gr-qc↗

arXiv2017

What if? - Exploring the Multiverse through Euclidean wormholes

We present Euclidean wormhole solutions describing possible bridges within the multiverse. The study is carried out in the framework of the third quantization. The matter content is modelled through a scalar field which supports the existence of a whole collection of universes. The instanton solutions describe Euclidean solutions that connect baby universes with asymptotically de Sitter universes. We compute the tunnelling probability of these processes. Considering the current bounds on the energy scale of inflation and assuming that all the baby universes are nucleated with the same probability, we draw some conclusions about what are the universes more likely to tunnel and therefore undergo a standard inflationary era.

gr-qc↗

arXiv2017

Cosmological perturbations in an effective and genuinely phantom dark energy Universe

We carry out an analysis of the cosmological perturbations in general relativity for three different models which are good candidates to describe the current acceleration of the Universe. These three set-ups are described classically by perfect fluids with a phantom nature and represent deviations from the most widely accepted $Λ$CDM model. In addition, each of the models under study induce different future singularities or abrupt events known as (i) Big Rip, (ii) Little Rip and (iii) Little Sibling of the Big Rip. Only the first one is regarded as a true singularity since it occurs at a finite cosmic time. For this reason, we refer to the others as abrupt events. With the aim to find possible footprints of this scenario in the Universe matter distribution, we not only obtain the evolution of the cosmological scalar perturbations but also calculate the matter power spectrum for each model. We have carried the perturbations in the absence of any anisotropic stress and within a phenomenological approach for the speed of sound. We constrain observationally these models using several measurements of the growth rate function, more precisely $fσ_8$, and compare our results with the observational ones.

astro-ph.CO↗