The Universitätsgesellschaft Ilmenau - Freunde, Förderer, Alumni e. V. awards Dr.-Ing. Liana Khamidullina the Dagmar Schipanski Prize for the best dissertation of the year. The scientist receives the award for her work on the development and analysis of new tensor-based signal processing concepts.
Modern technology is inconceivable without signal processing: microphones convert sound waves into electrical signals, radar systems detect movements, EEG devices record brain activity and mobile phone networks transmit voice and image information. While traditional signal processing often works with one-dimensional or two-dimensional data - such as curves - the complexity of the data is also increasing with technological progress: signals are now often measured simultaneously across many channels, in different frequency ranges or at several locations. This results in multidimensional data structures, so-called tensors - such as data cubes with the dimensions space × time × frequency.
Tensor-based signal processing makes targeted use of these structures. Instead of simplifying or reducing the data, it is mathematically recorded and analyzed in its full multidimensionality. Patterns, correlations and signal components can be recognized more precisely and complex systems can be controlled more efficiently.
In her dissertation "Tensor Decompositions and Algorithms for Efficient Multidimensional Signal Processing", which was written in 2023 and was awarded summa cum laude, Dr. Liana Khamidullina developed new methods to harness this potential. The methods she has developed enable a more efficient evaluation of multidimensional data - a challenge that is relevant in numerous interdisciplinary fields of application, such as wireless communication, biomedical signal processing or the analysis of social networks.
Of particular interest is their derivation of the so-called Multilinear Generalized Singular Value Decomposition (ML-GSVD) - a novel tensor decomposition that extends the Generalized Singular Value Decomposition (GSVD), which was previously limited to two matrices, to more than two matrices and thus enables more efficient resource allocation in mobile radio systems with multiple antennas - a decisive factor for future communication systems. In three concrete application scenarios, Dr. Khamidullina shows how ML-GSVD can be used in wireless communication systems. In addition, Dr. Khamidullina proposes an algorithm for high-resolution tensor-based localization of targets in the near field of multi-antenna radar systems. She is also developing tensor-based methods for data fusion and feature extraction from simultaneously measured EEG and MEG recordings, which were developed in cooperation with the Department of Biomedical Engineering at Ilmenau University of Technology and Jena University Hospital.
Dr. Khamidullina's dissertation attracted a great deal of attention in the scientific community. Significant parts of her work have been published in renowned journals, including the IEEE Transactions of Signal Processing, the most important journal of the IEEE Signal Processing Society, as well as in nine international peer-reviewed conference papers. In addition, her publications have been widely cited - including by researchers in Japan, the UK and Brazil.
In November 2024, Dr. Khamidullina was awarded the coveted Dr. Wilhelmy VDE Prize 2024 by the VDE (Association for Electrical, Electronic & Information Technologies) for her dissertation as one of the two best female doctoral students in electrical engineering in Germany.
Prof. Martin Haardt, Head of the Communications Research Laboratory Group at TU Ilmenau, supervised the dissertation and attests to its scientific excellence. According to him, the "groundbreaking results" of the dissertation provide extremely significant contributions in the field of multi-antenna signal processing and the design of efficient multidimensional algorithms for processing multi-channel data with a very broad interdisciplinary range of applications:
Dr. Khamidullina has written an outstanding and very mature dissertation at the highest theoretical level on a highly topical research problem that has numerous interdisciplinary practical applications. The dissertation makes an excellent scientific contribution to the advancement of knowledge and thus significant pioneering work in the field of multidimensional signal processing.