Biography

Georg Böcherer obtained his MSc degree in Electrical Engineering and Information Technology from the ETH Zürich. He spent one semester at the EPF Lausanne and wrote his master thesis at the Federal University of Pernambuco in Recife, Brazil. From 2007 to 2012 he worked towards his PhD degree at the Institute of Theoretical Information Technology at RWTH Aachen University. Since April 2012, he is with the Institute for Communications Engineering at the Technical University of Munich. His PhD thesis on probabilistic shaping received the E-Plus Dissertation Award. His work on coding for the ICT Cubes received the best paper award at ISWCS 2011. His Probabilistic Amplitude Shaping (PAS) scheme for capacity achieving and rate adaptive communication won a Bell Labs Prize in 2015. He was a co-organizer of the Munich Workshop on Coding and Modulation (MCM 2015).

Teaching

Lecturer

• Summer term 2015: Multi-User Information Theory
• Summer term (since 2014): Coded Modulation
• Winter term (since 13/14): Channel Coding
• CDHK: Kodierung variabler Länge: Datenkompresion und Verteilungsanpassung

Teaching Assistant

• Multi-User Information Theory
• Information Theory
• Ferienakademie: Network Information Theory and Coding
• Guest lecture Dr. Ungerböck: Coding Theory: Classical and Modern Channel Coding

Research Interests

Coded Modulation with Probabilistic Shaping

In Shannon theory, the key step in calculating capacity of a communication channel is to determine the capacity-achieving input distribution. In practical systems however, almost exclusively uniform input distributions are employed. This results in the so called shaping gap, i.e., a reduction of the maximum theoretically achievable rate for reliable communication. The goal is to develop practical tools that provably allow to overcome the shaping gap. To this end, the generation of capacity-achieving distributions, forward error correction and modulation are jointly considered.

• Recent result: High Throughput Probabilistic Shaping with Product Distribution Matching
• Rate Adaptation and Reach Increase by Probabilistically Shaped 64-QAM: An Experimental Demonstration
• Bandwidth Efficient and Rate-Matched Low-Density Parity-Check Coded Modulation
• Protograph-Based LDPC Code Design for Shaped Bit-Metric Decoding

Simulation of Random Processes

The generation of random numbers with a distribution that matches (resembles) a given target distribution is a core problem relevant to many communication scenarios such as data compression, constrained coding, coded modulation, and information theoretic security. The appropriate measure of resemblance from an information theoretic point of view is the informational divergence between the generated distribution and the target distribution. I am interested in theoretic limits, algorithms, and applications.

• Recent result: Constant Composition Distribution Matching
• Recent result: Informational Divergence and Entropy Rate on Rooted Trees with Probabilities

Projects

• Probabilistic Amplitude Shaping (PAS) Webdemo
• Constant Composition Distribution Matching
• ICT Cubes

Links

• Complete list of publications
• Find me on researchgate.net