COREnext Scientific Output:  Key Publications Advancing Research in Connectivity, Computing and Sub-THz Technologies

Throughout the COREnext project, partners produced a substantial body of scientific work that reflects the breadth of research carried out across communications engineering, semiconductor design, embedded systems, security and computing architectures. In total, 37 scientific papers were published, including conference papers, journal articles, a poster abstract, a book chapter and a white paper. Together, they demonstrate both the depth and continuity of the project’s scientific contributions.

Research on polymer microwave fibre (PMF) and sub-THz communication features prominently across multiple publications. Partners such as CHALM, CEA-Leti, Bordeaux-IMS, IFAG and Radiall reported advances in high-data-rate links at D-band, H-band and Y-band, exploring new modulation schemes, efficient coupler designs, waveguide characterisation and transmitter architectures. These studies include demonstrations of gigabit-class performance over polymer waveguides, compact antenna-in-package concepts and ray-tracing-based channel models for industrial and data-centre environments. The findings have been presented at major venues including the International Microwave Symposium, EuMW, ECTC and IEEE JC&S, further reinforcing COREnext’s role in shaping future short-range communication systems.

In parallel, the project contributed significantly to many-core computing, software-defined radio architectures and RISC-V based systems. ETHZ, BI, TUD and IHP authored papers on scalable clusters, vector processing, memory architectures, data movement for accelerator-rich systems and techniques for reducing contention in shared-memory designs. These outputs cover topics such as the TeraPool cluster, scalable memory mechanisms, many-core baseband processors and dynamic allocation schemes. Publications appeared in DATE, VLSI-SoC, GLSVLSI, ASPLOS, HotOS and EuroSys, underlining the relevance of COREnext research in the computing-systems domain.

Trustworthiness and secure system design formed another important thread. Researchers from BI and other partners examined modular trusted execution environments, software-defined CPU modes, remote attestation integration and trustworthy communication mechanisms for mobile systems. These studies, published through workshops such as SysTEX and high-level conferences including USENIX ATC, add a valuable dimension to the project’s efforts to integrate security and reliability into next-generation communication and computing infrastructures.

A dedicated white paper, Trustworthiness – The Key to Europe’s Digital Future, expands on these themes by examining digitalisation through the lens of trust and security. It outlines how these considerations will influence Europe’s position in high-end consumer goods and notes that the value of products is shifting towards embedded connectivity and sensing capabilities that link devices into wider digital ecosystems. The paper stresses that Europe’s continued leadership will depend on delivering technologies that are reliable, secure and aligned with societal expectations.

Further contributions include analytical work on Open RAN adoption, mmWave multiuser MIMO beamforming, LDPC accelerators and adaptive RISC-V systems for non-terrestrial sub-THz communication, demonstrating the interdisciplinary nature of the consortium.

Together, these 37 publications present a cohesive scientific narrative, advancing the state of the art in high-frequency communication links, enabling scalable computing architectures for next-generation radio systems and embedding trustworthiness at the heart of future network technologies. The collective contributions of COREnext partners reinforce Europe’s research position and support innovation pathways relevant to emerging 6G, semiconductor and computing developments.

 

 


COREnext Partners Engage Worldwide: Insights from 62 Events Shaping Europe’s Future Connectivity Landscape

Across the full duration of COREnext, partners took part in 62 events, ensuring a strong and sustained presence within the European and global research, industrial and policy landscape. This broad participation reflects the consortium’s active commitment to dissemination, technical exchange and engagement with communities shaping future connectivity technologies. Activities ranged from high-profile conferences to focused workshops, internal seminars and domain-specific symposia, each providing opportunities to present results, receive feedback and strengthen collaborations.

During EuCNC & 6G Summit 2025, partners showcased six demonstrations that drew interest from academic, industrial and policy audiences, positioning COREnext firmly within Europe’s 6G ecosystem. At European Microwave Week (2024–2025) and IMS 2024, CHALMERS, Bordeaux-IMS and CEA exchanged insights with hardware manufacturers, SMEs and standardisation bodies, highlighting advances in polymer microwave fibre communication, D-band and H-band technologies, and packaging concepts.

Events such as ECTC 2024 and RTAS 2024 enabled partners to engage with semiconductor, embedded-systems and real-time computing communities. Presentations covered topics including interconnect architectures, trustworthy system design and predictable cross-core communication, securing strong visibility for COREnext’s systems-level research.

Beyond these flagship events, partners were active across a wide set of specialised meetings. These included EuroSys & ASPLOS 2025, OSDI & ATC 2025, Embedded World 2025, Leti Innovation Days 2025, Silicon Saxony Day 2025, IP-SOC 2024, Design & Automation Conference 2025, DATE 2024, and numerous focused workshops on PMF technologies, trusted execution environments, and transceiver architectures. Contributions ranged from theoretical and experimental studies of plastic waveguides to demonstrations of high-data-rate links, chip-package co-design, channel modelling and secure system architectures.

Through this extensive programme of participation, COREnext partners ensured that project results were shared widely, discussed within expert communities, and aligned with ongoing industrial and academic developments. This broad engagement strengthened the project’s impact and maintained its visibility across the full spectrum of stakeholders involved in Europe’s future connectivity technologies.


High-Datarate Interconnects over Plastic Optical Fibre - Advancing Next-Generation Connectivity in COREnext

Within the COREnext project, one of the recent demonstrations led by INFINEON has explored how plastic optical fibre can support high-speed data transfer inside future communication systems. The work focuses on a high-rate data link operating in the H-Band, using a Plastic Multimode Fibre (PMF) and an in-package PMF coupler designed specifically for this frequency range. This approach offers a different route toward compact and efficient data interconnects, reducing dependency on traditional materials while supporting the scalability required for emerging workloads.

WATCH THE VIDEO

The system is built on a combination of MATLAB-based baseband signal processing and an analogue beam-steering transceiver architecture. Four independent RF front-ends and antennas allow real-time beam control, enabling the system to dynamically adapt while maintaining consistent transmission performance. The demonstration also incorporates an interactive monitoring interface, providing direct insight into link behaviour and allowing users to operate and observe the setup live — a useful feature for evaluating system stability, responsiveness and performance under different conditions.

A key achievement lies in the successful demonstration of the first in-package PMF coupler operating in the H-Band. Its compact structure, reduced optical loss and cost-efficient design make it a strong candidate for next-generation interconnects. By reducing energy consumption while sustaining very high throughput, this solution could support future environments such as large-scale data centres and highly distributed computing architectures, where bandwidth demands continue to grow.

The objective of the demonstration is to illustrate how fibre-based links can be implemented at high frequencies without compromising efficiency or system footprint. In doing so, it points to potential pathways for improving energy-aware communication at scale, and for building digital infrastructures that remain flexible as demand increases. As development continues, the work stands as a reference point for future system integration within COREnext and beyond — offering a glimpse of what next-generation network interconnects may look like.

 


AI-Enabled Physical Security Using Radio Frequency Fingerprinting 

As part of the COREnext project, researchers demonstrated an AI-enabled physicallayer security solution based on radio frequency (RF) fingerprinting - a technique that enables the authentication of wireless devices by exploiting subtle, hardware-based signal characteristics. The demonstration showcased how machine learning can identify, authenticate, and protect against impersonation attacks in communication network systems. 

Enhancing Physical Layer Security with RF Fingerprinting 

Every radio transmitter, even when manufactured to the same specifications, exhibits unique hardware imperfections – such as variations in amplifiers, oscillators, and other circuits. These imperfections leave a distinct “fingerprint” on the transmitted signal. 

The RF fingerprinting method leverages these differences to create a unique identifier for each device, making it possible to authenticate transmitters not only by their software credentials or classical cryptography- based methods, but by their physical-layer signal signatures too. 

In this demonstration, 5G New Radio (NR) signals at 2.5 GHz are used as authentication signal. By using techniques such as operating the radio amplifiers at high gain, the researchers amplified the nonlinearities in the signal — the very characteristics that make each device’s transmission unique. 

WATCH THE VIDEO HERE

Device Authentication through Machine Learning 

The setup included three software-defined radio devices from the same manufacturer, two of which were identical models — a challenging scenario for classification. A convolutional neural network (CNN) was trained to analyse short (1.4 ms) signal transmissions and classify each device based on its RF fingerprint. 

Despite the physical similarity between devices, the trained model achieved near 100% accuracy in distinguishing between all three. Visualisations of the model feature space showed well-separated clusters for each device, confirming that the neural network successfully captured subtle signal-level distinctions invisible to traditional security mechanisms. 

Testing the System’s Robustness - Impersonation Attack Scenario 

To assess the system’s resilience, the researchers simulated an impersonation attack where a malicious actor attempts to mimic a legitimate device signal using identical transmission settings, with a device of the same model as the target. This test represented an open-set scenario, meaning the attacker’s device had not been part of the model’s training data. 

When the impersonating device transmitted its signal, the machine learning model successfully detected it as an unknown or anomalous device, labelling it as unauthorised.  

Towards AI-Driven Physical Layer Security 

This demonstration highlights the potential of combining AI techniques with physical-layer security to protect wireless systems from spoofing and impersonation attacks. In addition to traditional cryptographic methods, RF fingerprinting provides a hardware-rooted form of authentication that is extremely difficult to forge or tamper with. 

As the COREnext project continues to advance secure communication technologies, AI-enabled RF fingerprinting stands out as a powerful approach to ensure device authenticity, resilience against attacks, and trustworthy operation in next-generation networks. 

 


Trust Evaluation and IoT Management 

As part of the COREnext project, Wings ICT Solutions has developed and demonstrated an innovative Trust Evaluation and IoT Management mechanism - a digital component designed to optimise the deployment of workloads across Internet of Things (IoT) devices based on their trustworthiness. This work forms part of the project’s broader objective to create secure, adaptive, and high-performance digital infrastructures for next-generation communication systems. 

The growing number of interconnected devices in modern networks creates both opportunities and challenges. While IoT devices enable smart and decentralised systems, they also vary significantly in reliability, performance, and security. To address this, the COREnext team designed a Trust Management Orchestrator capable of dynamically evaluating and assigning tasks to the most trustworthy IoT devices — ensuring secure and efficient operation across the network. 

How It Works - Trust-Based Orchestration 

The demonstrated solution organises and manages IoT devices across three computing layers — extreme edge, edge, and cloud, and follows a multi-step process: 

  1. Clustering: Devices are grouped using the K-means clustering algorithm, classifying them into groups based on their operational characteristics. 
  1. Trust Evaluation: A trust evaluation function calculates a trust index (ranging from 0 to 1) for each device class. Metrics include: 
  • Availability and reliability 
  • Security, data privacy, and integrity 
  • Energy consumption and battery level 
  • Network performance (latency, bandwidth) 
  • Multi-connectivity capabilities 
  1. Orchestration: The Trust Management Orchestrator uses these trust indices to determine the optimal workload placement, ensuring tasks are executed on the most trustworthy and capable devices. 

The system continuously gathers real-time performance data — including CPU utilisation, resource usage, and network metrics — and feeds it into an optimisation algorithm that adapts workload assignments dynamically. 

WATCH THE VIDEO HERE

 

Measurable Gains in Trustworthiness 

To validate the approach, the team conducted extensive simulations comparing the trust-based orchestration mechanism against a standard load-balancing algorithm. 

The results were clear: 

  • The proposed system achieved up to 53% higher trustworthiness in workload allocation compared to traditional load balancing. 
  • Larger device pools increased potential gains, as a greater number of trustworthy allocation options became available. 
  • Conversely, as the number of tasks increased, gains naturally decreased due to the limited solution space. 

These findings demonstrate the significant advantages of trust-aware orchestration in distributed IoT environments, improving both security and reliability without compromising efficiency. 

Towards Secure, Adaptive IoT Management 

The Trust Evaluation and IoT Management component represents a key step toward enabling self-organising, secure, and efficient IoT ecosystems. By integrating dynamic trust metrics into resource management, the COREnext project paves the way for intelligent, autonomous digital infrastructures that can adapt to evolving conditions and maintain high standards of security and performance. 

 


Validating Secure Sub-THz Point-to-Point Links 

As part of the COREnext project, researchers at IMEC have demonstrated a point-to-point sub-terahertz (sub-THz) communication link operating at 140 GHz, validating both its performance and security. The demonstration illustrates how high-frequency wireless communication can deliver ultra-high data rates while maintaining robustness against eavesdropping — a key step toward enabling next-generation 6G systems. 

Sub-terahertz frequencies, such as 140 GHz, are nearly 50 times higher than those used in typical 4G or 5G networks. These bands offer significantly larger bandwidths, enabling data throughputs in the multi-gigabit-per-second range. Such capacity is crucial for emerging applications, including virtual and augmented reality (VR/AR), autonomous systems, and data-intensive industrial connectivity. 

However, these advantages come with challenges. Operating at such high frequencies introduces hardware nonlinearities, including phase noise, IQ imbalance, and power amplifier limitations. To address these issues, IMEC first developed a comprehensive end-to-end MATLAB simulator to model and validate the effects of real hardware impairments in the sub-THz range. Once simulation results were verified, the team proceeded to experimental validation in the lab. 

WATCH THE VIDEO HERE

Experimental Setup and Performance Validation 

The laboratory setup featured IMEC’s custom 140 GHz transceiver chip, connected to the baseband simulator developed in the project. The initial validation involved a simple point-to-point configuration, with a transmitter and receiver aligned directly. 

Results from this experiment showed: 

  • 2 GHz bandwidth achieving a throughput of ~3.5 GB/s, 
  • When expanded to 4 GHz bandwidth, throughput increased to ~7 GB/s, 
  • The link maintained high reliability and a very low error rate. 

These results confirm the feasibility of high-speed, low-latency sub-THz wireless communication and its potential for real-world high-capacity use cases. 

Securing the Link with Analog Beam Steering 

Beyond throughput, the demonstration also addressed communication security - specifically, how to prevent eavesdropping at such high frequencies. 

To test this, the transmitter was mounted on a rotating platform to emulate a real-world scenario with a potential eavesdropper positioned off-axis. By performing analog beam steering, the team could control the direction of the transmitted signal. 

Key findings include: 

  • When the eavesdropper’s receiver was positioned 10° off the main transmission path, it could still capture some of the signal. 
  • Beyond 20°, however, the eavesdropper’s link became unreliable, with negligible data reception. 
  • When beam steering was applied toward the legitimate receiver, the link quality improved even under minor misalignments. 

These results demonstrate that beam steering not only enhances performance but also provides a practical layer of physical security, ensuring that signals remain confined to intended users. 

Towards Secure, Ultra-High-Speed Wireless Systems 

Through this demonstration, IMEC successfully validated both the feasibility and security of sub-THz communication at 140 GHz. The ability to achieve multi-gigabit throughput while maintaining link integrity and resisting interception highlights the potential of sub-THz technologies as a foundation for future 6G networks. 

By combining hardware innovation, accurate modelling, and intelligent beam control, the COREnext project continues to push the boundaries of what’s possible in secure, high-frequency wireless communication. 

 


Securing FPGA Offloading in the Cloud 

As part of the COREnext project, researchers at Nokia have demonstrated a secure FPGA offloading framework for cloud computing environments. The demonstration showcases how data processing tasks can be safely offloaded from CPUs to FPGAs (Field Programmable Gate Arrays) in containerised, cloud-native architectures, such as those using Kubernetes and Docker, without compromising data confidentiality. 

FPGA Acceleration in Cloud Environments 

In modern cloud infrastructures, microservices are deployed on CPUs and can offload computationally demanding tasks, such as digital signal processing (DSP) or machine learning algorithms, to FPGA accelerators. This approach significantly boosts performance and energy efficiency. 

However, this setup introduces security challenges. When multiple microservices share an FPGA, each allocated to a different logical array, a malicious service could potentially gain unauthorised access to another’s data. To prevent such breaches, the team at Bell Labs developed a multi-layer encryption and key management mechanism ensuring secure communication between CPUs and FPGAs. 

WATCH THE DEMONSTRATION HERE

The Security Architecture 

The solution integrates two layers of encryption: 

  1. AES Encryption for Data Protection
    Before data is sent from the CPU to the FPGA for processing, it is encrypted using the Advanced Encryption Standard (AES). This ensures that even if a malicious service accesses the data stream, it cannot decrypt or interpret the information. 
  2. ECC Encryption for Secure Key Exchange
    To address the challenge of securely sharing the AES key itself, the system uses Elliptic Curve Cryptography (ECC) for asymmetric encryption. 
  • Each FPGA logical array is wrapped with a security layer that includes a True Random Number Generator (TRNG). 
  • The TRNG generates a unique pair of public and private keys. 
  • The public key is sent to the CPU, which encrypts its AES key using this public key. 
  • The encrypted AES key is transmitted back to the FPGA, where the private key decrypts it. 

Once the AES key is securely received, the FPGA decrypts incoming data, processes it, and re-encrypts the results before returning them to the CPU, ensuring end-to-end protection of sensitive data. 

Secure Communication in Action 

During the live demonstration, the Bell Labs team connected remotely to a cloud server in France to showcase the encryption process in real time. The system dynamically generated a new ECC key pair for each session, ensuring that every transaction used a fresh public key while maintaining the same AES key on the CPU side. 

Each run of the demo illustrated the following workflow: 

  • The CPU requests a public key from the FPGA. 
  • The CPU encrypts its AES key with that public key. 
  • The encrypted AES key is sent to the FPGA and securely decrypted. 
  • Data is transferred, decrypted, processed, and re-encrypted automatically. 

Every time a new session began, the FPGA regenerated a new public key, confirming the dynamic and secure key exchange mechanism in operation. 

Towards Secure, High-Performance Cloud Acceleration 

This demonstration validates that secure FPGA offloading is achievable within cloud-native infrastructures. The combined use of AES for data encryption and ECC for secure key exchange provides a robust defense against potential attacks and data leaks. 

By embedding security directly into the hardware and orchestration layer, the COREnext project moves a step closer to enabling trustworthy, high-performance computing in distributed and virtualised environments. 

 KNOW MORE ABOUT CORENEXT SOLUTIONS


Demonstrating the M3 Platform - Tackling Hardware Vulnerabilities with Trusted Communication 

At a recent demonstration, Michael Roitzsch from the Barkhausen Institut presented the M3 hardware and operating system architecture, designed to tackle the increasing risk of hardware-level vulnerabilities in modern processors and accelerators. As these components grow in complexity, they are more prone to security flaws, as shown by high-profile cases such as vulnerabilities in Intel processors. The M3 platform aims to provide a more resilient structure capable of withstanding such attacks. 

WATCH THE DEMONSTRATION HERE

 

To illustrate the concept, the team used a simple tic-tac-toe game as a test scenario. The game involved processors, memory, a game engine verifying legal moves, a bot making automated moves, and a human participant. A deliberate vulnerability was introduced into one processor, allowing the bot to bypass the engine’s checks and overwrite moves directly in memory. This demonstrated how real-world hardware bugs can undermine intended safeguards, compromising system integrity. 

The solution proposed by the M3 platform is the use of Trusted Communication Units (TCUs), lightweight devices that sit in front of processors and memory to perform independent access control checks. Unlike complex processors, TCUs are simple enough to be implemented without flaws, ensuring that components only interact with authorised elements. In the demo, when the bot was moved to a processor protected by a TCU, unauthorised actions were blocked while legal moves remained possible.  

This highlights the M3 approach to creating more secure and reliable computing infrastructures, particularly in environments where multiple stakeholders share hardware resources. 

Know more about COREnext Solutions

 


COREnext Partners at European Microwave Week 2025

The European Microwave Week (EuMW) 2025, held in Utrecht, brought together researchers, industry leaders, and innovators in the field of microwave, RF, and wireless technologies. COREnext partners were prominently represented across several sessions, workshops, and demonstrations, showcasing their contributions to advancing high-frequency communication technologies.

Scientific Paper Presentation

A highlight of the scientific programme was the presentation of the paper “Ultra-Broadband Frequency Multiplier (x8) Chain in 90-nm SiGe BiCMOS Technology at H-band”. The work was elaborated by Frida Strömbeck and Herbert Zirath from Chalmers University of Technology, together with Klaus Aufinger from Infineon Technologies AG. This contribution illustrated advances in frequency multiplier chains and their role in enabling higher performance within integrated circuit design.

Full-Day Workshop on Polymer Microwave Fiber Communication

On Sunday, the full-day workshop “Polymer Microwave Fiber (PMF) Communication for sub-THz, Low-Cost High Data Rate Short-Range Systems” took place in Mission 2. Chaired by Frida Strömbeck and Herbert Zirath (Chalmers University), the session gathered leading experts to discuss progress in PMF design, interconnects, and complete system development.

The workshop explored PMF as a promising low-cost solution for short-range links under ten metres, aiming at data rates above 100 Gbps. Such systems are key for intra-box and module-to-module communication in future technologies. Speakers included Patrick Reynaert (KU Leuven), Herbert Zirath (Chalmers University of Technology), Jose-Luis Gonzalez (CEA-Leti), Laurent Petit (Radiall), Sergio Gambarucci (Infineon Technologies), Alejandro García Tejero (HUBER+SUHNER), Sining An (Ericsson), and Samir Lagoug (IMS Group).

Demonstration at the Radiall Booth

In addition to the workshop, COREnext partner Laurent Petit presented the D-band PolymerMediaFiber (PMF) modular high data rate connectivity demonstration kit at the Radiall booth. The kit featured two integration setups – flange connectors and direct seamless PCB connectors – and showcased several types of polymer fibre designed for high-capacity data transmission. The demonstration highlighted the flexibility of PMF technology and its potential for modular system integration in future communication networks.

About EuMW 2025

The European Microwave Week is one of the most significant international events in the field, combining scientific conferences, workshops, and industry exhibitions. The 2025 edition in Utrecht provided an important platform for the exchange of knowledge, ideas, and technology between academia and industry, further underlining the relevance of collaborative projects such as COREnext.

 


Validating Advanced AI and RF Concepts in T.6.1 – An Interview with Mamoun Guenach and Florent Torres 

In this interview, Mamoun Guenach from IMEC and Florent Torres from Ericsson outline the progress and objectives ofT6.1, which focuses on validating AI-based signal processing and theoretical models developed in earlier stages of the project.  

These validations aim to ensure robust and trustworthy wireless communications for a wider range of frequency bands such as sub-6 GHz, sub-10 GHz, and sub-terahertz. 

A key aspect of T6.1 is a mixed approach combining software components from the link-level simulators (PHY models) with real hardware components, so-called “hardware-in-the-loop” validation. In fact, there are three proof-of-concept demonstrations: 

  1. Sub-6 GHz Proof of Concept: Led by Ericsson, this focuses on RF device identification using radio frequency fingerprints. Using 5G signals and neural networks, the team achieved over 80% identification accuracy for devices from the same manufacturer, with some early results exceeding 90%. The team is also addressing cybersecurity challenges, such as impersonation attacks using replayed RF signals. In closed-set scenarios, high accuracy has been achieved in authenticating even identical devices. In open-set conditions, where attackers are not part of the training data, preliminary results are promising, though more work is ongoing. 
  1. Sub-10 GHz Proof of Concept: Conducted by NXP and targeting fingerprinting accuracy in the presence of manufacturing variations, particularly in Doherty power amplifiers used in 5G base stations. Early simulation datasets have been shared with Ericsson for validation, highlighting strong partner collaboration. 
  1. Sub-terahertz Proof of Concept: Managed by IMEC, this work assesses how residual RF non-idealities after mitigation impact the D-band radio link reliability and security. This is particularly crucial to control the beam-steering to enhance D-band link security.  A link-level simulator has been developed and integrated with custom IC designs (developed outside the  COREnext project) to demonstrate advanced features such as beam steering. 

Overall, T6.1 demonstrates strong collaboration and promising technical results, forming a solid foundation for future developments within the project. 

WATCH THE INTRVIEW HERE:


Subscribe to our newsletter

Horizon Europe – Grant Agreement number 101092598
Funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union. The European Union cannot be held responsible for them