UniServer · Project

Energy-Efficient Micro-Servers That Process Data Where It's Generated, Not in Distant Data Centres

digitalTestedTRL 5

uniserver2020.eu

Right now, most of the data from sensors, cameras, and smart devices gets shipped across the internet to huge data centres far away — which costs energy, money, and time. UniServer built small, super-efficient servers that can sit close to where data is actually created and handle it right there. Think of it like having a mini kitchen in every office instead of sending all lunch orders to one giant restaurant across town. They also figured out how to squeeze more performance from chips by removing the safety margins manufacturers usually bake in, kind of like realizing your car can safely go a bit faster than the speedometer's conservative limit.

By the numbers

EUR 4,815,810

EU funding for development

consortium partners

countries in consortium

64-bit

ARM-based Server-on-Chip architecture

60%

industry partner ratio

total project deliverables

SMEs in consortium

The business problem

What needed solving

Data centres consume enormous amounts of energy, and as IoT devices multiply, sending all data to centralised clouds creates latency, bandwidth costs, and single points of failure. Current server hardware wastes energy by using overly conservative voltage and frequency settings as a safety buffer. Businesses need compact, efficient servers that can work reliably at the network edge, closer to where data is actually generated.

The solution

What was built

UniServer delivered a fully working micro-server prototype based on a 64-bit ARM Server-on-Chip, along with software tools that expose and exploit processor voltage/frequency margins for better energy efficiency. The 25 deliverables include a Predictive Failure Module (evaluated as a demo), runtime resource management policies, and system software for managing distributed edge deployments.

Audience

Who needs this

Data centre operators looking to reduce energy costs and deploy edge infrastructureIoT platform companies needing low-latency local data processingTelecom operators building multi-access edge computing (MEC) networksSmart city solution providers deploying distributed sensor processingCloud service providers expanding into edge-as-a-service offerings

Business applications

Who can put this to work

Cloud and Edge Computing

enterprise

Target: Data centre operators and edge infrastructure providers

If you are a data centre operator dealing with rising energy costs and the need to serve IoT workloads closer to end users — this project developed a 64-bit ARM-based Server-on-Chip micro-server platform that cuts energy consumption by removing pessimistic voltage and frequency margins in processors and memory. The consortium of 10 partners including ARM and IBM validated this on real cloud and edge workloads across 6 countries.

Internet of Things

mid-size

Target: IoT platform companies and smart city solution providers

If you are an IoT solutions provider struggling with latency and bandwidth costs from sending all sensor data to centralised clouds — this project built a distributed micro-server ecosystem that processes data locally, near the source. The system was evaluated with smart emerging applications and includes a Predictive Failure Module to keep edge nodes running reliably without constant manual oversight.

Telecommunications

enterprise

Target: Telecom operators deploying multi-access edge computing (MEC)

If you are a telecom operator rolling out edge computing at cell tower sites and need compact, energy-efficient server hardware — UniServer produced a working prototype of a micro-server that improves both energy efficiency and dependability. With 6 industry partners contributing real-world requirements and 25 deliverables covering hardware and system software, the technology is designed for deployment in space-constrained environments outside traditional data centres.

Frequently asked

Quick answers

What would it cost to adopt this micro-server technology?

The project itself received EUR 4,815,810 in EU funding across 10 partners over 3.5 years. Pricing for the resulting technology components is not published in the project data. Since key partners include ARM and IBM, commercial licensing or product integration would likely go through their existing sales channels.

Can this scale to production-level deployment?

UniServer was evaluated using real applications deployed in both classical cloud data centres and in new edge environments closer to data sources. The technology was ported onto 64-bit ARM-based Server-on-Chip hardware from commercial suppliers (ARM, APM), which is designed for volume production. Scaling to commercial deployment would depend on the chip suppliers' roadmaps.

Who owns the intellectual property and how can I license it?

IP is distributed among the 10 consortium partners across 6 countries, including ARM, APM, and IBM as major industry contributors. Licensing terms are not publicly specified in the project data. Interested parties should contact the coordinator at Queen's University Belfast or the industry partners directly.

How does this compare to existing edge server solutions?

UniServer's key differentiator is its approach to removing pessimistic voltage and frequency margins in processors and memory — something current commercial servers do not exploit. This means better energy efficiency from the same hardware. The project also developed software-only mechanisms to manage this, requiring no custom silicon beyond the ARM Server-on-Chip.

Is there regulatory or compliance risk in using voltage-margin-reduced servers?

The project embedded lightweight software mechanisms for monitoring and managing margin reduction, including a Predictive Failure Module to catch problems before they cause errors. Based on available project data, the approach was validated on real workloads but commercial certification status is not documented.

What is the timeline to get this into my infrastructure?

The project closed in July 2019, so all 25 deliverables and the working prototype are complete. Integration timeline depends on which components you need — the software tools could potentially be adopted faster than new hardware. Contact the consortium partners for current availability.

Consortium

Who built it

This is a commercially serious consortium with 60% industry participation — 6 industry partners out of 10 total, including heavyweight names like ARM (the dominant mobile/embedded chip architecture company), APM (server-on-chip manufacturer), and IBM (enterprise software). The 4 SMEs add agility and niche expertise. Spread across 6 countries (Cyprus, Germany, Greece, Spain, Ireland, UK), the project was coordinated by Queen's University Belfast. The strong industry presence — especially ARM and IBM — means the technology was built with real commercial constraints in mind, not just academic curiosity. For a potential adopter, this means the results are more likely to be production-relevant and that commercial pathways through these major vendors may already exist.

THE QUEEN'S UNIVERSITY OF BELFASTCoordinator · UK
ETHNIKO KAI KAPODISTRIAKO PANEPISTIMIO ATHINONparticipant · EL
IBM IRELAND LIMITEDparticipant · IE
ARM LIMITEDparticipant · UK
SPARSITY SLparticipant · ES
WORLDSENSING SLparticipant · ES
PANEPISTIMIO THESSALIASparticipant · EL
UNIVERSITY OF CYPRUSparticipant · CY

How to reach the team

Queen's University Belfast, UK — reach out through the university's research commercialisation office

Order a report on this consortium See THE QUEEN'S UNIVERSITY OF BELFAST profile →

Next steps

Talk to the team behind this work.

Want an introduction to the UniServer team or a tailored brief on how this edge computing technology fits your infrastructure? Contact SciTransfer — we connect businesses with EU research teams.

Order a report on this topic More in Digital & ICT