CORRELATE

Sustainable IoT: a strategic necessity 

As the manufacturing industry rapidly digitizes, Industrial IoT (IIoT) systems based on wireless sensors and actuators are becoming central to monitoring and controlling equipment. These devices often rely on batteries, supercapacitors, or energy harvesting. However, their growing energy footprint poses both environmental and economic concerns.

To ensure both sustainable and cost-effective operations, it’s therefore crucial to develop energy-aware embedded systems for truly wireless IIoT systems. This includes hardware and software capable of optimizing energy use, especially through distributed computing.

The CORRELATE project addresses this challenge by minimizing energy consumption through smart device activation, efficient data collection, and strategic deployment of machine learning models in the network.

Tackling key barriers to energy-aware IIoT

The CORRELATE project targets three major challenges that limit energy-efficient monitoring and control in Industrial IoT systems:

• Inefficient data transfer: Instead of sending full data streams, only relevant information—like a narrow frequency range needed for bearing fault detection for example—should be transmitted, cutting down from hundreds of packets to just a few.
• Inefficient computation: As an example, continuous high-frequency monitoring isn't always needed. Since faults develop gradually, tracking asset degradation over time allows for smarter, lower-power processing strategies.
• Inefficient deployment: Machine learning models should run where they consume the least energy—often directly on the device at the network edge.

By minimizing redundancy in time, data, and deployment, CORRELATE aims to significantly reduce energy use while maintaining high performance and accuracy.

Innovation objectives and use cases 

CORRELATE’s main objective is to tackle the lack of coordination between IIoT devices and backend asset management software—a gap that limits energy efficiency, accuracy, and responsiveness. By turning passive, device-agnostic data collection into an interactive cloud-edge feedback loop, the project will enable real-time, intelligent, energy-aware data classification. The CORRELATE innovations will:

• Guarantee an IIoT system’s lifetime of 20 years by reducing the network traffic and eliminating data redundancy by a factor of 50. A further factor of 10 can be gained by exploiting time redundancy, i.e. prioritizing data analysis and precious bandwidth on those assets that are showing signs of failure.
• Maintain industry-level data classification accuracy of 99% when integrating the data classifier into the control loop and achieving energy savings.
• Achieve the energy and accuracy objectives without interfering with the application quality.

These ambitious innovation goals are feasible as CORRELATE builds on three technologies that its partners have developed and proven in-house. These confirm that major bandwidth savings are possible through feature compression and through the optimization potential of energy-aware tiny machine learning deployment, and that industrial-grade accuracy can be achieved by closing the feedback loop.

The project partners have selected three representative use cases that will drive the innovation and prove its feasibility: One in infrastructure monitoring, one in fault monitoring (e.g. bearings, motors) and a last in display control (e.g. HVAC, passenger information).