Energy-Efficient Adaptive Polling in LoRaWAN-IIoT for Bakery Production: Model, AoI Analysis, and Experimental Validation

The paper proposes an energy-efficient architecture for an industrial IIoT monitoring system based on LoRaWAN for bakery production and develops a mathematical model linking radio channel and network traffic parameters with node energy consumption, delivery reliability, and data freshness. The model includes the calculation of Time-on-Air as a function of SF/BW/CR and payload length, the node energy balance across "measurement–processing–transmission–reception–sleep" phases, a probabilistic assessment of Packet Delivery Ratio (PDR) considering coverage and collisions, and the Age of Information (AoI) metric for quantitative control of telemetry relevance. An adaptive polling mechanism is proposed, combining event-driven transmissions and periodic heartbeat, with individual period ranges for different types of measurements (accounting indicators and technological parameters). Experimental validation on a facility testbed confirms the proximity of calculated and measured energy costs: the total model error per cycle does not exceed 3.4%, and the largest contribution to energy consumption is provided by radio transmission (about 83% per cycle). It is shown that the adaptive mode improves the "energy–freshness" trade-off: compared to the baseline mode of 300 s, a reduction in energy consumption from 10.20 to 6.50 mAh/day is achieved, while simultaneously decreasing AoI p95 from 285 to 180 s. The results obtained can be used for engineering tuning of LoRaWAN-IIoT systems in workshop conditions, taking into account the requirements for autonomy, reliability, and data relevance.