Albena Mihovska, Associate Professor at Aarhus University, introduces the MOTOR5G project, which uses Artificial Intelligence to enable smarter and safer wireless ecosystems.
The HORIZON 2020-funded MOTOR5G project is focused on embedding Artificial Intelligence (AI) into 5G communication systems for the smarter use of network-generated data, and the automated enabling of network operators and service providers to adapt to changes in traffic patterns, security risks and user behaviour. It, therefore, paves the way towards safe and reliable next-generation wireless ecosystems.
MOTOR5G considers aspects such as the use of drone-based technology for enhanced multi-antenna and data forwarding techniques; the use of AI for novel adaptive digital beamforming techniques applied on realistic antenna arrays; communications in the millimetre-wave bands; blockchain-based approaches to spectrum management and sharing; and the use of Machine Learning (ML) for enhanced Quality of Experience (QoE). In parallel, research focuses on novel business models to sustain profitable operations beyond 5G ecosystems.
Embedding Artificial Intelligence for cross-layer applications
The MOTOR5G project focuses on the development of neural networks (NN) that can replace a basic physical layer network process called adaptive beamforming. It is a technique used in wireless communications to improve signal quality and reliability by adapting the direction of the radiation beam transmitted or received by an elementary antenna in real time.
In the context of network prediction, adaptive beamforming can be used to improve the accuracy of network performance prediction by taking into account the changing properties of the wireless channel. By adjusting the weights in real-time, adaptive beamforming algorithms are used to account for changes in the wireless channel, such as fading and interference, which can affect the signal quality and network performance. This can help improve the accuracy of network predictions by providing a more realistic model of the wireless channel. In addition, Deep Learning (DL) applications in adaptive beamforming can be used to predict the optimal weights for future situations.
Deep Learning for robust 6G performance
The number of network devices continues to rise as we advance towards 6G communication systems. A new range of frequencies is allocated, while the earlier resources remain underutilised. Cognitive Radio (CR) enables the dynamic spectrum management of frequencies while detecting the unoccupied bands with the aid of spectrum sensing. By adopting DL for spectrum sensing, the performance of the 6G networks can be made more robust. MOTOR5G will develop novel DL algorithms developed and validated on radio frequency (RF) datasets to achieve efficient sharing of radio resources.
Blockchain-based approach to dynamic spectrum access using MOTOR5G
The current network resources are becoming insufficient to satisfy the connectivity requirements due to the drastic increase in mobile devices and mobile users. This has resulted in a demand for higher bandwidth availability and new network resources. The MOTOR5G project investigates Dynamic Spectrum Access (DSA) and network slicing for efficient resource utilisation based on Quality of Service (QoS) requirements of the user. A blockchain-based solution for DSA and network slicing is proposed and investigated, to enhance the security, fairness, and auditability of the resource sharing process.
Machine Learning for enhanced Quality of Experience
Quality of Experience is a measure that evaluates a communication network’s performance as end users perceive it. The assessment of QoE is a highly complex issue, as QoE is a very broad and multidisciplinary concept, affected by a wide range of parameters, such as throughput, latency, packet loss, and jitter. Network planning that relies exclusively on QoS is insufficient to meet the divergent requirements of Future Wireless Networks (FWNs). MOTOR5G has proposed an application-oriented QoE enhancement framework, focused on emerging multimedia services such as Virtual Reality (VR), and the deployment of ML algorithms centred on network performance optimisation. More precisely, ML algorithms will be developed and integrated into Open Radio Access Network (O-RAN) architecture.
Objective video quality assessment (VQA) to support QoE-oriented video transmission
The rapid development of multimedia technologies and the increasing demand for new forms of content, such as Augmented Reality (AR) and VR, make it difficult for mobile networks to meet the specifications required for such applications. 360° videos tend to have significantly higher bitrate demands than regular videos and are subjected to strict latency requirements. As a result, mobile network operators should address the anticipated massive traffic in their future designs. QoE in VR applications require novel algorithms that must be developed to consider both the new network designs and the emerging videos’ requirements and characteristics.
Simultaneous Localization and Mapping (SLAM) for mm-wave
The mm-wave band offers huge available bandwidth, which makes it beneficial to be used in positioning applications for its increased resolution capability to distinguish closely spaced objects. However, it is not suitable to be employed in long-range applications as it faces severe attenuation due to the propagation environment. Therefore, mm-wave signals remain to be a suitable choice for indoor applications. The MOTOR5G project has proposed a methodology based on the use of Deep Learning to improve the estimation of the angle-of-arrival (AoA) from a low-cost mm-wave sensor. The project will demonstrate a data-driven mechanism that takes noisy measurements of the AoA from a mm-wave sensor and maps it to an AoA estimate with better accuracy.
Safe and secure design for unmanned aerial vehicles
Unmanned aerial vehicles (UAVs) play a vital role in non-terrestrial network (NTN) wireless communications because of their different altitudes, ability to hover, and manoeuvrability. Wireless communication poses a challenge to information security, especially with the broadcast nature of UAVs. Conventional networks have utilised infrastructure-based security standards, but power limitations in UAV communication require infrastructure-less security. Physical layer security (PLS) is a promising approach to mitigate security threats from eavesdroppers due to its cost-efficient, time-efficient, and seamless architecture. PLS can be ensured through keyless and key-based approaches by exploiting the channel state information (CSI).
Novel business models using MOTOR5G
Massive technology transformations are very challenging with respect to the way current mobile telecommunications networks have been designed and deployed. Beyond purely technological impacts, these transformations impose requirements for adaptability and timeliness. MOTOR5G examines existing and emerging business models in the telecommunications ecosystem and analyses these findings to derive perspectives for Beyond-5G (B5G) and 6G networks.
The MOTOR5G project is funding the mobility and development of research talent in Europe
The MOTOR5G project is funding 15 early-stage researchers (ESRs), supporting their research and training towards the award of the PhD degree. Through their involvement in the research and network activities provided under the project’s umbrella, they are able to greatly improve their skills, both in the research-oriented and scientific dimension and in terms of transversal skills. With that, the project will contribute to creating highly skilled young researchers for the European labour market.
This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie Grant Agreement No 861219
Please note, this article will also appear in the fifteenth edition of our quarterly publication.
