IRU-RAN

Project Summary

The project focuses on the development of intelligent resource utilisation methods for efficient next-generation radio access in converged heterogeneous wireless networks.

The considered network scenarios combine terrestrial cellular and ultra-dense networks with communication nodes deployed on unmanned aerial vehicles (UAVs).

Radio Environment Maps (REMs) are used as a central source of information for characterising spectrum occupancy and the spatial and temporal behaviour of the radio environment.

The project integrates statistical modelling, signal processing, artificial intelligence (AI), machine learning (ML), propagation modelling, and optimisation to develop new methods for two- and three-dimensional REM reconstruction, channel assessment, user behaviour prediction, user association, radio-resource allocation, and interference management.

The developed methods will be integrated into a system-level simulation framework for evaluating their performance under realistic heterogeneous-network conditions.

The expected result is a more adaptive and self-configuring radio access network capable of improving spectrum and energy efficiency, throughput, Quality of Experience (QoE), and seamless connectivity.

Concept & Methodology

The central concept of the project is to use Radio Environment Maps as an information layer supporting intelligent and proactive network operation.

REMs will represent the spatial, temporal, and frequency-domain variations of the radio environment and will provide information for spectrum analysis, channel assessment, user association, and radio-resource management.

The project assumes that combining statistical REM reconstruction methods, AI-based algorithms, predictive RF data analysis, propagation models, and optimisation techniques can significantly improve the efficiency and adaptability of heterogeneous radio access networks.

The research will begin with a systematic review of the state of the art related to REM construction, spectrum analysis, channel modelling, and resource allocation.

Publicly available datasets collected in real environments, together with data generated through propagation models and simulations, will be used for the development and evaluation of the proposed methods.

The data will undergo preprocessing, transformation, spatial interpolation, and semantic analysis to obtain suitable representations of the radio environment.

Statistical methods and criteria will be applied to characterise signal fluctuations and spectrum occupancy in two- and three-dimensional space.

AI and machine-learning methods, including neural networks, autoencoders, heterogeneous classifiers, and anomaly-detection techniques, will be investigated for REM reconstruction, RF data analysis, and user behaviour prediction.

Deterministic, statistical, and hybrid propagation models will be used to describe the communication channels in terrestrial, ultra-dense, and UAV-assisted network scenarios.

The reconstructed REMs, propagation models, and predicted user behaviour will support the development of low-complexity user-association and resource-allocation algorithms.

Particular attention will be given to proactive user association between terrestrial access points and UAV-based nodes, user-centric resource virtualisation, and joint resource allocation and interference management.

The complete methodology will be evaluated through system-level simulations using both conventional network indicators and user-oriented Quality of Experience metrics.

Project Objectives

1. Develop accurate radio-environment and propagation models for characterising signal variations, spectrum occupancy, and channel conditions in two- and three-dimensional heterogeneous-network environments.
2. Develop low-complexity methods for REM reconstruction that improve reconstruction accuracy when only sparse, incomplete, or compressed RF measurements are available.
3. Enable the practical construction and application of REMs for spectrum monitoring, network planning, and intelligent radio-resource management.
4. Develop AI- and ML-based spectrum utilisation methods that allow terrestrial access points and UAV-based communication nodes to adaptively access and use the available RF spectrum.
5. Develop predictive user behaviour models using big-data analysis, machine learning, heterogeneous classifiers, and anomaly-detection techniques.
6. Design REM-based user-association and self-configuration schemes for heterogeneous networks containing terrestrial access points, ultra-dense deployments, and communication UAVs.
7. Develop efficient joint resource-allocation and interference-management algorithms with low computational complexity and support for proactive network adaptation.
8. Develop user-centric Quality of Experience assessment methods based on advanced data analysis and integrate them into the proposed network-management framework.
9. Develop system-level simulation models for evaluating the effectiveness, adaptability, and performance of the proposed REM reconstruction, user-association, resource-allocation, and network self-configuration methods.