A Bayesian Decision Architecture for Adaptive IoT Monitoring under Energy, Connectivity, and Measurement Uncertainty

• José Jefferson Pires Gonçalves IFPB
• Bruno Neiva Moreno IFPB
• Emanuel Filho UFCG
• Danyllo Albuquerque UFCG

DOI: https://doi.org/10.5753/ise.2026.32579

Resumo

Adaptive IoT monitoring systems require decision mechanisms that can operate under uncertainty, especially when deployed in rural environments with limited energy, intermittent connectivity, and noisy sensor measurements. In water-monitoring applications, fixed-threshold rules may trigger inadequate decisions because they usually ignore the interaction between domain conditions and operational constraints. This paper presents an emerging result from an ongoing research project that investigates Bayesian Networks as an interpretable decision layer for adaptive IoT monitoring. We propose an initial Bayesian decision architecture for rural cistern monitoring, integrating evidence about cistern level, measurement uncertainty, precipitation, communication-link quality, and battery status. The model infers water risk and transmission viability to support an operational decision. We conducted a preliminary evaluation using five simulated scenarios covering normal operation, water-supply risk, uncertain measurement, low transmission viability, and poor operational conditions. The results indicate that the model produces coherent inferences and can adjust the final decision when water-related urgency and system-level feasibility diverge. Although still preliminary, the architecture provides a structured basis for engineering adaptive decision mechanisms in constrained IoT systems. Future work will incorporate real sensor data, refine the conditional probability tables, adopt more granular states, and compare the Bayesian policy with fixed-threshold baselines.

Palavras-chave: Bayesian Networks, Adaptive IoT, Decision Architecture, Rural Water Monitoring, Uncertainty, LoRaWAN

Referências

A. Alghamdi and E. Khairullah. 2022. IoT-based water monitoring systems: A review. IEEE Access (2022).

M. Azghadi et al. 2024. Energy-aware modeling for IoT-based monitoring systems. Sensors (2024).

E. Bertone et al. 2022. Bayesian network approaches in water systems management. Water Research (2022).

C. N. de Castro. 2025. Semiárido, escassez hídrica e articulação de políticas públicas. Technical Report. Instituto de Pesquisa Econômica Aplicada (Ipea).

J. Cavalcanti et al. 2024. Adaptive decision-making in IoT systems under uncertainty. Journal of Network and Computer Applications (2024).

A. Dawoud et al. 2022. Smart water monitoring using IoT technologies. IEEE Access (2022).

S. Jabbar et al. 2024. Statistical reliability in IoT sensor networks. Sensors (2024).

Finn V. Jensen and Thomas D. Nielsen. 2007. Bayesian Networks and Decision Graphs (2 ed.). Springer.

E. de O. Marques. 2026. Tecnologia social: impactos do programa um milhão de cisternas na saúde pública e nos gastos do SUS com a participação da Fundação Banco do Brasil. Technical Report. Instituto de Pesquisa Econômica Aplicada (Ipea).

S. J. Moe, J. F. Carriger, and M. Glendell. 2021. Bayesian networks for water resource decision support. Science of the Total Environment (2021).

R. Oliveira. 2025. Estratégias inteligentes para monitoramento hídrico com IoT. Revista Brasileira de Computação Aplicada (2025).

R. Oliveira, M. Conceição, and F. Neto. 2020. LoRaWAN para aplicações IoT em zonas rurais brasileiras: uma revisão sistemática. In Simpósio Brasileiro de Redes de Computadores.

Judea Pearl. 1988. Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference. Morgan Kaufmann.

T. D. Phan et al. 2016. Applications of Bayesian belief networks in water resource management: A systematic review. Environmental Modelling and Software 85 (2016), 98–111.

T. D. Phan et al. 2019. Applications of Bayesian networks as decision support tools for water resource management under climate change and socio-economic stressors: A critical appraisal. Water (2019).