Author

Júlio Cezar Estrella

h-index 1 View on Semantic Scholar

We have 2 of 7 papers

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

Conference Aug 2026

From Deployment to Operation: A Reproducible IoT Testbed for Smart-Building HVAC with OTA and Observability

Smart-building research increasingly requires IoT platforms that support reproducible deployment and long-term operation under real conditions. This paper presents a field-operable and reproducible IoT testbed for smart-building HVAC, grounded in an architecture aligned with ISO/IEC 30141 and designed for zero-touch onboarding, fleet-wide Over-The-Air (OTA) lifecycle management, and end-to-end observability. The architecture integrates ESP32-based devices running FreeRTOS, multi-protocol communication using MQTT and CoAP, and a lightweight containerized backend with time-series persistence and monitoring.The testbed was deployed in a university laboratory, where a rule-based HVAC control strategy reduced monthly energy consumption from 101.68 to 45.04 kWh (55.7%) while maintaining thermal comfort. Controlled experiments further evaluate protocol trade-offs, quantifying MQTT QoS impacts on actuation latency and comparing MQTT and CoAP for continuous telemetry under constrained conditions. In addition, the architecture enables progressive edge intelligence through data-driven model deployment without structural changes.Rather than proposing new protocols or control algorithms, this work contributes a standards-aligned, observable, and evolvable IoT testbed, together with empirical evidence and practical design insights for field-oriented smart-building research.

Lucas de Souza Marques, Pedro Porto Teixeira, Renan Correia Monteiro Soares et al. · International Conferences on... · 0 citations

Conference Jul 2026

Improving Federated Learning under Non-IID Data through Clustered Ensembles and Adaptive Client Selection

Federated Learning (FL) enables collaborative model training without sharing raw data, but its effectiveness degrades under Non-IID client data and inefficient participation. In such settings, FedAvg may exhibit slow convergence, high variance across rounds, and elevated communication overhead. We propose a clustered ensemble framework to improve training stability and communication efficiency under label-skewed Non-IID distributions. Clients are clustered by data similarity, and one model is trained per cluster. At inference time, we employ a confidence-based ensemble that selects (or combines) cluster models, while adaptive client selection regulates participation within clusters to reduce redundant communication. Experiments on MNIST, Fashion-MNIST, CIFAR-10, and SVHN show more stable learning dynamics and lower communication cost while maintaining competitive accuracy compared to FedAvg. We also analyze how participation policies affect convergence behavior and client fairness.

Artur Sousa Freitas, A. T. Akabane, J. Estrella · International Conference on... · 0 citations