The latest publication by PhD student Sebastian Górecki, who is conducting his research under the supervision of Prof. Sławomir Hausman and Prof. Piotr Korbel, focuses on the design of energy-efficient AIoT devices that combine IoT sensors with local data processing and TinyML models.
The starting point is a seemingly simple question: Does transmitting less data always mean lower energy consumption? The results show that this is not necessarily the case. The authors point out that the actual energy balance should be considered over the device’s entire operating cycle - from data acquisition and preprocessing, through local inference using TinyML, to modem activation, network registration, synchronization, transmission and possible retransmissions, and finally the return of the device to sleep mode.
💡 The key idea of the publication is cross-layer energy management, meaning the joint optimization of sensing, data processing, and wireless communication, taking into account factors such as battery level, harvested energy from the environment, and radio link quality.
Experiments using LTE-M showed that, for messages ranging from 24 to 512 B, energy consumption did not increase monotonically with the amount of data transmitted. This means that signaling and protocol overhead also play a significant role.
The publication also highlights the potential of semantic compression. Instead of transmitting the full signal, a device can analyze it locally and transmit only the information relevant to a given task. In an example involving acoustic beehive monitoring, the amount of data was reduced from 196 608 B to just 8-24 B of information describing the signal’s features.
🐝 The example of precision beekeeping demonstrates how this approach can be applied in practice: energy-efficient environmental measurements can be used to determine when to activate more energy-intensive measurements involving sound, images, or vibrations.
📖 We encourage you to read the publication:
Sebastian Górecki, Sławomir Hausman, Piotr Korbel, “Cross-Layer Energy Management in Embedded AIoT Sensor Networks: A Structured Review, Full-Cycle Energy Model, and Break-Even Analysis”, Electronics, 2026, 15(17), 4000. https://doi.org/10.3390/electronics15174000
📸 Photos: Sebastian Górecki
Image
Image
Image
