Main Technical Results Aligned with Project Objectives

1. Development of Compute‑and‑Forward and Distributed Lossy Computation Techniques (Objective 1)
Research, design, and implementation of compute‑and‑forward (CaF) and distributed lossy computation (DLC) for uplink fronthaul between O-RUs and the CU, enabling efficient distributed processing in cell-free massive MIMO (CF-mMIMO).
Validated through an SDR-based demonstrator, these techniques improve uplink performance while reducing computational complexity.

2. Advanced Hybrid Beamforming Solutions for mmWave Fronthaul (Objectives 1 & 2)
Investigation and development of hybrid beamforming mmWave fronthaul solutions, including strategies, algorithms, and codebooks for efficient radio resource management.
Variants of Penalty Convex-Concave Procedure (PCCP) were designed to optimize architectures, mitigate spectral interference, maximize user rates, and minimize energy consumption under SINR constraints.
Implemented schemes cover both point-to-point (PtP) and point-to-multipoint (PtMP) topologies with local processing and adaptive cooperation between APs, DUs, and O-RUs.

3. Low-Complexity Digital Predistortion Techniques Using Machine Learning (Objectives 1 & 2)
Proposed digital predistortion (DPD) methods using OLS-UES and PCA-UES, reducing hardware complexity and training time while improving energy efficiency for CF-mMIMO mmWave hybrid beamformers.
Edge DPD approaches were explored to dynamically reconfigure DPD/DFE functions based on network KPIs and conditions, leveraging micro-orchestration radio mechanisms.

4. Proactive Auto‑Scaling of vRAN Radio Functions (Objectives 2 & 3)
Developed containerized xApp solutions for proactive scaling of radio functions in virtualized RAN using probabilistic resource forecasting (SVPF), outperforming LSTM-based deterministic models.
Integration with Near-RT RIC demonstrated improved load balancing, RRM, and coordination between DUs, facilitating convergence between fixed and mobile networks.
Future multivariate models (GPVA, TFT) were explored for further resource prediction accuracy.

5. CF‑mMIMO O‑RAN Demonstrator and PoC Validation (Objective 4)
Implemented and presented a CF-mMIMO O-RAN demonstrator at MWC2025 using Keysight T&M instrumentation and software tools.
Validated multiple precoding options, realistic channel conditions, RF imperfections, and performance metrics (e.g., SINR, BLER, EVM).
Integration of rApps and xApps enabled cluster formation, resource allocation optimization, and evaluation of new inter-DU interfaces for low-latency cooperation.

6. High-Impact Scientific and Industrial Dissemination (Objective 5)
The project achieved extensive dissemination across scientific, industrial, and educational communities.
Results were shared in international journals, conferences, workshops, and industrial forums, including contributions to standards discussions.
Outreach initiatives to students and the general public, along with an active online presence, strengthened visibility and promoted international collaboration and knowledge transfer.

7. Strengthening National Industrial R&D and Skills (Objective 5)
Enhanced the national R&D ecosystem by promoting high-skilled job creation at CTTC and Keysight, attracting national and international talent, and fostering gender equality in R&D activities.
Developed algorithms, containerized applications, and FPGA-ready hardware represent commercially valuable assets for advanced vRAN and pre-6G networks.
Supported knowledge transfer from public research groups to industry and enabled Keysight to explore white-paper dissemination and future commercial exploitation pathways.