Examinando por Autor "Marques, Antonio G."

Mostrando 1 - 7 de 7

A Unified Approach to QoS-Guaranteed Scheduling for Channel-Adaptive Wireless Networks
(IEEE, 2007-12-01) Wang, Xin; Giannakis, Georgios B.; Marques, Antonio G.
Scheduling amounts to allocating optimally channel, rate and power resources to multiple connections with diverse quality-of-service (QoS) requirements. It constitutes a throughput-critical task at the medium access control layer of today's wireless networks that has been tackled by seemingly unrelated information-theoretic and protocol design approaches. Capitalizing on convex optimization and stochastic approximation tools, the present paper develops a unified framework for channel-aware QoS-guaranteed scheduling protocols for use in adaptive wireless networks whereby multiple terminals are linked through orthogonal fading channels to an access point, and transmissions are (opportunistically) adjusted to the intended channel. The unification encompasses downlink and uplink with time-division or frequency-division duplex operation; full and quantized channel state information comprising a few bits communicated over a limited-rate feedback channel; different types of traffic (best effort, non-real-time, real-time); uniform and optimal power loading; off-line optimal scheduling schemes benchmarking fundamentally achievable rate limits; as well as on-line scheduling algorithms capable of dynamically learning the intended channel statistics and converging to the optimal benchmarks from any initial value. The take-home message offers an important cross-layer design guideline: judiciously developed, yet surprisingly simple, channel-adaptive, on-line schedulers can approach information-theoretic rate limits with QoS guarantees.
Energy-Efficient Quantization and Resource Allocation for TDMA with Finite Rate Feedback
(IEEE, 2008-09-01) Wang, Xin; Marques, Antonio G.; Giannakis, Georgios B.
We deal with energy efficient time-division multiple access (TDMA) over fading channels with finite-rate feedback for use in the power-limited regime. Through finite-rate feedback from the access point, users acquire quantized channel state information. The goal is to map channel quantization states to adaptive modulation and coding modes and allocate optimally time slots to users so that the total average transmit-power is minimized. To this end, we develop a joint quantization and resource allocation approach, which decouples the complicated problem at hand into three minimization sub-problems and relies on a coordinate descent approach to iteratively effect energy efficiency. A sub-optimal yet simplified alternative algorithm which decouples the original problem into two sub-problems is also presented. Numerical results are presented to evaluate the energy savings and compare the novel approaches.
Enhanced graph-learning schemes driven by similar distributions of motifs
(2024) Rey, Samuel; Roddenberry, T. Mitchell; Segarra, Santiago; Marques, Antonio G.
We develop a novel convolutional architecture tailored for learning from data defined over directed acyclic graphs (DAGs). DAGs can be used to model causal relationships among variables, but their nilpotent adjacency matrices pose unique challenges towards developing DAG signal processing and machine learning tools. To address this limitation, we harness recent advances offering alternative definitions of causal shifts and convolutions for signals on DAGs. We develop a novel convolutional graph neural network that integrates learnable DAG filters to account for the partial ordering induced by the graph topology, thus providing valuable inductive bias to learn effective representations of DAG-supported data. We discuss the salient advantages and potential limitations of the proposed DAG convolutional network (DCN) and evaluate its performance on two learning tasks using synthetic data: network diffusion estimation and source identification. DCN compares favorably relative to several baselines, showcasing its promising potential.
Interpretable clinical time-series modeling with intelligent feature selection for early prediction of antimicrobial multidrug resistance
(Elsevier, 2022) Martínez-Agüero, Sergio; Soguero-Ruiz, Cristina; Alonso-Moral, Jose M.; Mora-Jiménez, Inmaculada; Álvarez-Rodríguez, Joaquín; Marques, Antonio G.
Electronic health records provide rich, heterogeneous data about the evolution of the patients’ health status. However, such data need to be processed carefully, with the aim of extracting meaningful information for clinical decision support. In this paper, we leverage interpretable (deep) learning and signal processing tools to deal with multivariate time-series data collected from the Intensive Care Unit (ICU) of the University Hospital of Fuenlabrada (Madrid, Spain). The presence of antimicrobial multidrug-resistant (AMR) bacteria is one of the greatest threats to the health system in general and to the ICUs in particular due to the critical health status of the patients therein. Thus, early identification of bacteria at the ICU and early prediction of their antibiotic resistance are key for the patients’ prognosis. While intelligent data-based processing and learning schemes can contribute to this early prediction, their acceptance and deployment in the ICUs require the automatic schemes to be not only accurate but also understandable by clinicians. Accordingly, we have designed trustworthy intelligent models for the early prediction of AMR based on the combination of meaningful feature selection with interpretable recurrent neural networks. These models were created using irregularly sampled clinical measurements, both considering the health status of the patient and the global ICU environment. We explored several strategies to cope with strongly imbalance data, since only a few ICU patients are infected by AMR bacteria. It is worth noting that our approach exhibits a good balance between performance and interpretability, especially when considering the difficulty of the classification task at hand. A multitude of factors are involved in the emergence of AMR (several of them not fully understood), and the records only contain a subset of them. In addition, the limited number of patients, the imbalance between classes, and the irregularity of the data render the problem harder to solve. Our models are also enriched with SHAP post-hoc interpretability and validated by clinicians who considered model understandability and trustworthiness of paramount concern for pragmatic purposes. Moreover, we use linguistic fuzzy systems to provide clinicians with explanations in natural language. Such explanations are automatically generated from a pool of interpretable rules that describe the interaction among the most relevant features identified by SHAP. Notice that clinicians were especially satisfied with new insights provided by our models. Such insights helped them to trust the automatic schemes and use them to make (better) decisions to mitigate AMR spreading in the ICU. All in all, this work paves the way towards more comprehensible time-series analysis in the context of early AMR prediction in ICUs and reduces the time of detection of infectious diseases, opening the door to better hospital care.
Minimizing Transmit-Power for Coherent Communications in Wireless Sensor Networks with Finite-Rate Feedback
(IEEE, 2008-09-01) Marques, Antonio G.; Wang, Xin; Giannakis, Georgios B.
We minimize average transmit power with finite-rate feedback for coherent communications in a wireless sensor network (WSN), where sensors communicate with a fusion center using adaptive modulation and coding over a wireless fading channel. By viewing the coherent WSN setup as a distributed space¿time multiple-input single-output (MISO) system, we present optimal distributed beamforming and resource allocation strategies when the full (F-) channel state information at the transmitters (CSIT) is available through a feedback channel. We also develop optimal adaptive transmission policies and design optimal quantizers for the finite-rate feedback case where the sensors only have quantized (Q-) CSIT, or, each sensor has F-CSIT of its own link with the FC but only Q-CSIT of other sensors. Numerical results confirm that our novel finite-rate feedback-based strategies achieve near-optimal power savings based on even a small number of feedback bits.
Optimizing Power Efficiency of OFDM Using Quantized Channel State Information
(IEEE, 2006-08-01) Marques, Antonio G.; Digham, Fadel F.; Giannakis, Georgios B.
Emerging applications involving low-cost wireless sensor networks motivate well optimization of orthogonal frequency-division multiplexing (OFDM) in the power-limited regime. To this end, the present paper develops loading algorithms to minimize transmit-power under rate and error probability constraints, using three types of channel state information at the transmitter (CSIT): deterministic (per channel realization) for slow fading links, statistical (channel mean) for fast fading links, and quantized (Q), whereby a limited number of bits are fed back from the transmitter to the receiver. Along with optimal bit and power loading schemes, quantizer designs and reduced complexity alternatives with low feedback overhead are developed to obtain a suite of Q-CSIT-based OFDM transceivers with desirable complexity versus power-consumption tradeoffs. Numerical examples corroborate the analytical claims and reveal that significant power savings result even with a few bits of Q-CSIT.
Power-Efficient Wireless OFDMA Using Limited-Rate Feedback
(IEEE, 2008-02-01) Marques, Antonio G.; Giannakis, Georgios B.; Digham, Fadel F.; Ramos, Javier
Emerging applications involving low-cost wireless sensor networks motivate well optimization of multi-user orthogonal frequency-division multiple access (OFDMA) in the power-limited regime. In this context, the present paper relies on limited-rate feedback (LRF) sent from the access point to terminals to minimize the total average transmit-power under individual average rate and error probability constraints. Along with the characterization of optimal bit, power and subcarrier allocation policies based on LRF, suboptimal yet simple schemes are developed for channel quantization. The novel algorithms proceed in two phases: (i) an off-line phase to construct the channel quantizer as well as the rate and power codebooks with moderate complexity; and (ii) an on-line phase to obtain, based on quantized channel state information, the optimum, rate, power and user-subcarrier allocation with linear complexity. Numerical examples corroborate the analytical claims and reveal that significant power savings result even with suboptimal schemes based on practically affordable LRF.