The rapid evolution of cellular networks, driven by the proliferation of mobile devices and the exponential growth of the Internet of Things (IoT), has significantly advanced wireless communication technologies. Fifth generation of wireless communications technology (5G) enhanced data rates, latency, and network capacity, resulting in the emergence of new applications. However, the sixth generation (6G) is foreseen to support a new set of use cases with diverse requirements. This paper explores the critical role of artificial intelligence (AI) in shaping the trajectory from 5G to 6G. We discuss AI applications in 5G for network planning, resource allocation, traffic management, and security, as well as propose infrastructure upgrades, like edge servers and enhanced network topologies, to support AI in 6G. Additionally, we outline a visionary perspective on AI’s potential contributions to 6G, highlighting its role in enabling innovative services and applications. By providing this forward-looking perspective, this paper aims to stimulate discussion and guide the development of intelligent and autonomous 6G networks.

Frequency-modulated continuous-wave (FMCW) radars are becoming increasingly important in industrial applications due to their low cost, simple signal processing, and high resolution. The research is driven by the growing trend of radar applications in industrial settings, including tasks like autonomous robot navigation, radar-based monitoring of humans, and enhancing safety across various scenarios, such as smart factories, traffic control, or environmental monitoring. The network of distributed sensor nodes is a feature to be implemented by the sixth generation of wireless mobile networks. As a result, 6G will become an enabler technology that will develop even further use cases and advance the expansion of these technologies. Therefore, the mutual interference of these sensor systems is becoming highly relevant since they can be avoided by scheduling the individual resources of the sensor nodes by a network structure. This paper investigates the interference behavior of FMCW radar sensors in a dense indoor scenario. In these environments, in particular, the reliability of the sensor system plays an important role, and disruption can have critical consequences like not recognizing a target, triggering a false alarm, or the complete failure of the sensing. To correctly classify the expected interference in the overall context, the paper investigates the physical layer effects on the performance, such as multipath propagation, channel spread, delay, Doppler shift, and attenuation. By evaluating these effects using state-of-the-art signal processing, it was revealed that a high channel spread in particular and the dynamics of the radar channel can significantly reduce the detection quality, as the interference power can exceed the received power by up to 20 dB, depending on the scenario. The simulation results given in this paper show that a potential communication between the radars and compliance with a synchronization delay, the interference problems can be reliably prevented with an appropriately designed anti-aliasing filter. Our simulations have quantified the detection performance under interference through the entire signal chain by evaluating the physical layer, the radar signal processing with non-idealities, and the hardware. The paper has investigated interference probability in indoor environments by examining radars with uncoordinated transmissions. Integrating findings from the physical propagation characteristics, the deviations in chirp durations, and the frequency of interference cases offers insights into the potential risk of interference.

FMCW radar systems are increasingly used in diverse applications, and emerging technologies like JCAS offer new opportunities. However, machine learning for radar faces challenges due to limited application-specific datasets, often requiring advanced simulations to supplement real-world data. This paper presents a setup for generating synthetic radar data for indoor environments, evaluated using CNNs. The setup involves comprehensive modeling, including far-field antenna simulations, variations in human radar cross-section, and detailed representations of indoor environments with their corresponding propagation channel properties. These synthetic data are used to train CNNs, and their performance is assessed on real measurement data. The results demonstrate that CNNs trained on synthetic data can perform well when tested on real measurement data. Specifically, the models trained with synthetic data showed performance comparable to models trained with real measurement data, which required a minimum of 300 samples to reach similar levels of accuracy. This result demonstrates that synthetic data can effectively train neural networks, providing an alternative to real measurement data, particularly when collecting sufficient real-world samples is difficult or costly. This approach significantly reduces the time required for generating datasets, and the ability to quickly label data in simulations simplifies and accelerates post-processing. Additionally, the generated datasets can be made more heterogeneous by introducing varying signal conditions, enhancing the diversity and robustness of the training data.

The rapidly increasing share of fluctuating electricity from photovoltaics calls for accurate approaches to estimate cloud motion, the primary source for the varying power supply. While local sensor networks are prominent in targeting forecast horizons too short for image‐based methods, they have minimal spatial coverage. This work presents the first step towards expanding those approaches to spatially scalable sensor networks: With the motivation of using automotive light sensors as a sensor network, two excerpts from a microscopic traffic simulation serve as simulative sensor networks. A fractal‐based cloud shadow pattern passes the sensor network areas with defined velocities and directions, which shall be estimated using the cumulative mean absolute error method. The evaluation results indicate that the more extensive observation areas compensate for the dynamics in the sensor network when compared to a reference work with a static sensor grid. Furthermore, this work shows how the estimates deteriorate with lower vehicle penetration rates (PR) and longer building shadows due to a lower solar elevation angle. At a penetration rate of 40%, the root mean square errors for both sensor networks are still below 5 ms⁻¹. In conclusion, the spatiotemporal characteristics of a vehicle network offer a potential for estimating cloud movements.