Where should a computer do its work? AI brings data processing closer to where data is generated + test

A young engineer holding a clipboard stands in a server room between racks full of network cables.

A drone recognising objects in flight or a sensor that detects a sudden drop in a building’s air quality cannot wait long for an answer from a distant data centre. The European AC3 project has developed a platform in which AI itself decides where each task should run: in the cloud or on a nearby server. Read how it works and what it achieved in tests – and at the end take a short test to see how much you remember.

Contents

⇑The problem: data travels too far

More and more devices around us collect data: cameras, sensors in buildings, drones, laboratory equipment. The simplest solution is to send everything to a large data centre with enormous computing power. The trouble is that the journey there and back takes time. When a drone needs to identify an object in real time, or a sensor detects a sudden change in air quality, waiting for data to travel to a distant data centre can slow the response.

Processing information closer to where it is generated can make computing faster and reduce unnecessary data transfers. The difficulty, however, is that someone has to decide every time where a particular task should run. This is a complex challenge, because conditions keep changing: the workload rises and falls, the quality of the connection varies, and devices differ in power.

The EU-funded AC3 project took on this problem. Its team developed a platform that uses AI and machine learning to manage computing resources across cloud data centres and nearby edge servers, adapting as workloads and conditions change. Project coordinator Christos Verikoukis of Athena Research Centre in Greece explains: “By processing data close to where it is generated, the AC3 platform reduces communication delays and network congestion, enabling the system to react more quickly to changing conditions.”

⇑Cloud and edge: two levels of processing

To understand what AC3 does, it helps to distinguish between two places where programs can run. In practice, the cloud means huge data centres with thousands of servers that we connect to over the internet. They have enormous power, but they are often hundreds of kilometres away from the device sending the data.

Edge computing means that part of the computation is done at the edge of the network, that is, on servers located close to the source of the data, for example in the same building or at a nearby station. Such servers have less power than a data centre, but they respond faster. Computer scientists call the time between sending data and receiving a response latency. In many applications, from controlling a drone to alarms in buildings, this delay decides whether the system works well.

⇑Microservices: a program built from blocks

Modern applications are rarely one big program. More often they consist of multiple smaller software components, known as microservices. Each of them does one thing, such as receiving data from a sensor, analysing an image or sending a notification, and each can run in a different location.

The AC3 platform automates decisions about where these components should be deployed. In doing so, it takes into account computing capacity, network conditions and energy use. Verikoukis explains: “The main innovation of AC3 is the use of AI and machine learning to automatically decide where and how these components should be deployed and managed.” Machine learning is a field of artificial intelligence in which a program is not given ready-made rules but learns them from data, for example from the history of server workloads.

⇑Three tests for the platform

The team tested the platform in three scenarios that show how varied its applications can be:

  • smart building management – sensors track air quality, for example, and the system has to respond quickly to changes;
  • drone-based video surveillance – a drone identifies objects in its camera image in real time;
  • astronomical data analysis – large datasets have to be processed as quickly as possible.

In a smart building it works like this: sensors detect a sudden drop in air quality, and the AC3 platform moves data processing to a nearby server. As a result, the system can trigger an alert or adjust ventilation faster. At the same time, less data is sent to a distant data centre, so the network carries a lighter load.

⇑A drone that hands its work to a server

The most vivid demonstration involved a drone. Detecting objects in an image is a job for an AI model, and such a model needs a fair amount of computing power. The small computer on board a drone cannot always cope, for example when it is carrying out other tasks at the same time.

If a drone cannot run an AI model for real-time object detection efficiently, AC3 can move the workload to a nearby edge server. When the drone has enough capacity again, the task can move back on board. Adlen Ksentini, AC3’s technical manager at EURECOM in France, sums it up: “This intelligent and seamless migration ensures that the object detection algorithm continues to operate optimally, regardless of changes in the drone’s available computing resources.” Migration here means moving a running task from one computer to another without interrupting it.

The team used similar flexibility for the astronomical data. Processing tasks were distributed across multiple computing resources, which means they were carried out in parallel on several machines instead of on just one.

Fun fact: by distributing work across multiple computing resources, the AC3 platform reduced the time needed to process 5 GB of astronomical data by at least 50% compared with stand-alone systems. In the project’s tests, the application also maintained 100% reliability.

⇑Less electricity and code for everyone

Data centres consume a lot of energy, so it is not only speed that matters, but also savings. The AC3 platform can use predictions of computing demand and energy consumption to consolidate workloads onto fewer servers. Unused nodes, that is, individual machines in the network, can then be switched off while applications continue to meet their performance requirements.

Several software components from the project have been released as open source, meaning their code is available for anyone to inspect and develop further. Some tools have also been integrated into the established Kubernetes and OpenShift ecosystems. These are platforms that many companies use to manage applications made up of containers, which are packaged pieces of software ready to run on any server.

From the point of view of a company or institution, the AC3 approach brings several concrete benefits:

  • faster system response, because data is processed close to where it is generated;
  • less network congestion, because less data is sent to a distant data centre;
  • energy savings, thanks to consolidating work onto fewer servers and switching off the rest;
  • open components that developers and researchers can use as a starting point for their own work.

⇑What comes next: from the lab to real services

Verikoukis and Ksentini acknowledge that the next challenge is no longer purely technical. Cloud services and edge-based systems often remain separate, and stronger integration and viable business models are needed to bring them together. In other words, someone has to make money from such a solution, and someone else has to want to pay for it.

The project has shown that intelligent coordination can make computing faster and more efficient. The next step is to help organisations put these capabilities to work in real-world services, from building management to the analysis of scientific data.

⇑Who is this topic for?

Edge computing brings several fields together at once: computer networks, distributed systems, artificial intelligence and energy efficiency. It is a good topic for students of computer science, telecommunications, automation and robotics, electronics or data science. Specialists in the cloud, containers and machine learning are in demand at technology companies, telecom operators and research institutes. Open-source code from projects such as AC3 is also a good opportunity to practise with real tools.

⇑In short

When data has to be processed quickly, what counts is not only how powerful a computer is, but also how far away it is. The platform from the AC3 project uses AI to decide automatically whether a task should run in the cloud or on a nearby edge server. In tests with buildings, a drone and astronomical data, it sped up computing, maintained reliability and helped save energy, and some of its software is available to everyone.

⇑Five questions: how much do you know about edge computing? (test)

5 questions · one minute

1. Why is it worth processing data close to where it is generated?

2. What does the platform from the AC3 project do?

3. What happens when a drone cannot cope with object detection?

4. What result was achieved with the astronomical data?

5. How does the AC3 platform help save energy?


published: 2026-11-17
Comments
StudentNews Group Websites for students and future students — in Poland since 2007, across Europe since 2014. Over 50 websites and millions of users a year (Google Analytics). Studies in Europe Universities in Europe Bachelor’s Master’s Doctorate MBA Study guide Work abroad Travel For partners Advertising and cooperation Reach (online report) Contact advertising@studentnews.eu © 2003–2026 Grupa StudentNews.pl sp. z o.o., Modlnica, Poland
Privacy Policy