Project Details
Projekt Print View

High resolution parameter estimation for millimeter wave propagation in dynamic scenarios

Subject Area Electronic Semiconductors, Components and Circuits, Integrated Systems, Sensor Technology, Theoretical Electrical Engineering
Term from 2019 to 2024
Project identifier Deutsche Forschungsgemeinschaft (DFG) - Project number 418074791
 
Final Report Year 2024

Final Report Abstract

A significant contribution of this project is the extension of multidimensional parameter estimation framework to include frequency responses of estimated specular components, particularly for high bandwidths in the millimeter-wave (mmWave) and sub-THz range. This allows for a more concise modeling of the propagation characteristics. The findings emphasize the importance of simplifying the data model to reduce estimation efforts, particularly by maintaining parameter independence. For instance, the ”narrowband assumption”, which assumes a low relative bandwidth and ignores frequency response, is usually employed to simplify the model. However, this work goes beyond that by incorporating angle dependent frequency responses of antenna arrays, which are particularly beneficial in high bandwidth scenarios. All these novelties are summarized in the development of a new gradient-based maximum likelihood estimator. The project also improves parameter estimation in settings with high relative bandwidth by introducing a parametric data model and a gradient-based maximum likelihood estimator for this purpose. The developed model accounts for delay-dispersive responses of dynamic objects within the channel. As such, it addresses the challenges posed by the large relative bandwidth systems, such as frequency dependent Doppler effects and possibly dispersive/extended reflections within the channel. As such, the project developed an efficient estimation routine of the dispersion characteristics of the objects, modeling it as a smooth function in frequency domain. The approach is validated through simulations and real-world measurements, demonstrating the model’s ability to accurately represent and estimate the characteristics of multiple objects. The project also addresses the challenge of estimating Dense Multipath Components (DMC) in radio channel measurements in scenarios with multiple modes, which are expected to be more likely present in mmWave channels. Existing methods typically handle single-mode scenarios, but the research presents a novel approach using a deep learning architecture to estimate the number of modes and their characteristics. The proposed method combines a neural network-based initialization with a maximum likelihood estimation algorithm. The neural network is trained on synthetic data and can reliably predict the number of modes present in real measurement data. The project shows that the developed method can robustly estimate up to three overlapping modes, which is a significant advancement over existing methods. Overall, the project contributes to the fields of radio channel sounding, channel parameter estimation estimation, and antenna modeling. The advancements in high-resolution parameter estimation, modeling of dynamic objects, multi-modal DMC estimation, and related software tools provide valuable resources for researchers and practitioners in telecommunications and signal processing.

Publications

 
 

Additional Information

Textvergrößerung und Kontrastanpassung