Over the course of the last decades, the demands on modern communication systems have been rising continuously. Emerging data-intense applications in combination with challenging user experience requirements continue to drive innovation. At the same time, the need for resource-aware technologies has never been greater than during the pervasive climate crisis.
Typically, modern digital communication systems are implemented as integrated circuits. To improve energy and resource efficiency of these systems, the available computational resources and chip area have to be spent cautiously to minimize costs and the ecological footprint. In these systems, the hardware complexity directly correlates with the computational complexity of the algorithms constituting the receiver and transmitter of the system. For reliable and powerful communication, sophisticated algorithms are necessary that conventionally operate with high-resolution numbers. Using these high-resolution numbers requires a lot of hardware resources that consume a substantial amount of energy.
The information bottleneck method is a method is an information-theoretic framework that can be used to design reduced complexity receiver operations, that are optimized to maximize the amount of preserved relevant information about the transmitted data. This way, the overall required hardware and energy resources of the receiver can be reduced, while only losing a small amount of performance.
In his master thesis, Jasper Brüggmann applied the information bottleneck method to the turbo equalization principle. Turbo equalization is a well-known procedure for iterative mitigation of inter-symbol interference and error correction. This work combines results obtained in prior research by the Institute of Communications, where the information bottleneck method has been applied extensively to low-density parity check (LDPC) decoders, as well as receivers for inter-symbol interference channels in an award-winning conference paper. Extending these results to the turbo equalization principle showed significant performance improvements in addition to a reduced implementation complexity.
The Karl H. Ditze foundation was established in 1979, annually granting awards for excellent bachelor theses, master theses, dissertations and student initiatives since the year 2000. Besides scientific excellence, practical significance and multidisciplinary work are important evaluation criteria.
Further information: TUHH: Karl H. Ditze-Preise 2024 vergeben
