Project COGNIT: Optimising energy use at the edge

Project COGNIT aims to develop an advanced AI-enabled edge application solution for the energy and other sector domains.

The main goal of COGNIT is to research, develop, validate and make publicly available an AI-enabled adaptive serverless framework for the cognitive cloud-edge continuum. This environment supports a new innovative ‘function as a service’ (FaaS) paradigm for edge application management, based on code offloading.

It enables the on-demand deployment of large-scale, highly distributed and self-adaptive serverless environments using existing data processing resources from cloud/edge infrastructure providers, including local data centres, cloud providers and 5G/telecom operators.

Additionally, it optimises where data is processed according to application demands and behaviour changes while considering energy efficiency heuristics.

In the project, the applicability and viability of the proposed solution will be demonstrated across multiple application domains, including the energy sector domain.

Supporting the energy transition

Supporting the energy transition in Europe requires wide and open access to energy data.

Due to diminishing fossil fuel resources, i.e. oil and gas, and the current geopolitical context, it is critical to develop solutions for the energy independence of households and small energy clusters.

For this, methods for predicting, monitoring and managing energy production and consumption in each environment must be developed, with a clear need for implementing smart edge applications that utilise advanced AI/ML algorithms.

Current deployments in the energy sector are limited by edge applications being deployed in resource-constrained environments, i.e. energy meters, and the very high-security requirements for distribution system operators (DSO), which are usually a result of centralised architectures based on private distributed cloud infrastructures.

Advanced AI/ML applications would benefit from being able to leverage, in a secure way, additional resources across the cloud-edge continuum that are much closer to the data sources and the edge/IoT devices on the ground.

Smart meters as an edge management device

The energy domain use case of the COGNIT project explores the scenario of using smart electricity meters to optimise local green energy usage in a household context, in which energy consumers are also energy producers or prosumers.

The current energy system is carbon-intensive and centralised and electricity must be transmitted over long distances through the transmission and distribution network resulting in high losses. Bottlenecks and disruptions in the network have the potential to affect huge areas and populations.

The energy industry of the future will be based on distributed systems, relying on renewable energy sources (RESs) and energy storage solutions.

This highly distributed model of the energy network features many small local producers of energy, aiming to reduce costs, risks and intensity of greenhouse gas emissions, as well as to eliminate the transmission energy losses.

To make this a reality, there is a need to manage energy production and consumption locally. Electricity meters, already at the interface between the building and the power grid, are ideally positioned to manage such distributed smart energy systems.

Electricity meters can run many user applications to manage major appliances and energy assets installed behind the meter, such as battery storage, PV installations, heat pumps, electric vehicle chargers and electric floor heating, adjusting and optimising operation in real-time based on the user’s preferences.

By empowering electricity meters with apps and connected to services equipped with advanced decision-making algorithms – and eventually pre-trained AI models – they are turned into highly personalised ‘energy assistants’.

You might be interested in:
EU-funded energy projects
The EU Energy Projects Podcast

To optimise local energy usage, decision algorithms should consider weather forecasting data and other relevant data such as energy costs and consumption or production.

Even though electricity meters today are more powerful than ever, resource limitations remain one of the main challenges for software developers targeting them. Thanks to the lightweight operating system Phoenix-RTOS as the basis for developing apps for such resource-constrained platforms, these resources can be managed optimally and additionally benefit from software modularity.

Offloading decision-making and optimisation tasks to the edge allows these resource-constrained devices to take advantage of the low latency benefits of highly distributed edge nodes and be powered up with additional external high-performance computational resources.

Ultimately, this approach leads to cost and carbon savings because of the more effective usage of energy and lowering of the overall demand for coal-based energy.

This energy domain demonstration is being coordinated by Phoenix Systems, a company specialising in smart grid technologies, and cloud platform provider Atende Industries.

Visit our website and let’s connect on LinkedIn and  X.