Advanced Topics in Communication Systems: Codes on Graphs
Lecturer (assistant) | |
---|---|
Number | 0000003226 |
Type | lecture with integrated exercises |
Duration | 4 SWS |
Term | Sommersemester 2025 |
Language of instruction | English |
Position within curricula | See TUMonline |
Dates | See TUMonline |
- 24.04.2025 15:00-16:30 N2408, Seminarraum
- 29.04.2025 11:30-13:00 N2408, Seminarraum
- 06.05.2025 11:30-13:00 N2408, Seminarraum
- 08.05.2025 15:00-16:30 N2408, Seminarraum
- 13.05.2025 11:30-13:00 N2408, Seminarraum
- 15.05.2025 15:00-16:30 N2408, Seminarraum
- 20.05.2025 11:30-13:00 N2408, Seminarraum
- 22.05.2025 15:00-16:30 N2408, Seminarraum
- 27.05.2025 11:30-13:00 N2408, Seminarraum
- 03.06.2025 11:30-13:00 N2408, Seminarraum
- 05.06.2025 15:00-16:30 N2408, Seminarraum
- 12.06.2025 15:00-16:30 N2408, Seminarraum
- 17.06.2025 11:30-13:00 N2408, Seminarraum
- 24.06.2025 11:30-13:00 N2408, Seminarraum
- 26.06.2025 15:00-16:30 N2408, Seminarraum
- 01.07.2025 11:30-13:00 N2408, Seminarraum
- 03.07.2025 15:00-16:30 N2408, Seminarraum
- 08.07.2025 11:30-13:00 N2408, Seminarraum
- 10.07.2025 15:00-16:30 N2408, Seminarraum
- 15.07.2025 11:30-13:00 N2408, Seminarraum
- 17.07.2025 15:00-16:30 N2408, Seminarraum
- 22.07.2025 11:30-13:00 N2408, Seminarraum
- 24.07.2025 15:00-16:30 N2408, Seminarraum
Admission information
Objectives
After attending the course, students will be able to
- Analyse factor graphs
- Design (spatially-coupled) LDPC codes
- Design (spatially-coupled) turbo-like codes
- Understand product-like codes from a modern coding theory lens
- Interpret LDPC decoder for high-throughput applications
- Interpret product-like decoders for high-throughput applications
- Illustrate rateless-codes
- Employ neural decoders for linear block codes
- Describe applications of (graph-based) coding beyond communications and storage
- Analyse factor graphs
- Design (spatially-coupled) LDPC codes
- Design (spatially-coupled) turbo-like codes
- Understand product-like codes from a modern coding theory lens
- Interpret LDPC decoder for high-throughput applications
- Interpret product-like decoders for high-throughput applications
- Illustrate rateless-codes
- Employ neural decoders for linear block codes
- Describe applications of (graph-based) coding beyond communications and storage
Description
The focus of this course is on modern coding theory. In contrast to classical coding theory, which finds its roots in algebra, modern coding theory refers to a broad family of coding techniques unified by the use of sparse-graph constructions, iterative decoding algorithms, and probabilistic methods. For this reason, modern codes are also sometimes referred to as graph-based codes or probabilistic codes. Iterative message-passing methods have revolutionized coding, and more generally, communications and storage.
The course covers recent advances in modern coding theory: (capacity-achieving) spatially-coupled codes, turbo-like codes, product-like codes from a modern coding theory lens, coding for ultra high-throughput applications, neural decoders, and the application of coding to areas beyond communication and storage.
The course covers recent advances in modern coding theory: (capacity-achieving) spatially-coupled codes, turbo-like codes, product-like codes from a modern coding theory lens, coding for ultra high-throughput applications, neural decoders, and the application of coding to areas beyond communication and storage.
Prerequisites
It is strongly recommended to have taken courses in information theory or channel coding. Knowledge of basic probability, linear algebra, and combinatorics (at the level of standard first-year undergraduate courses) will be expected.
Teaching and learning methods
2 slots of 90 minutes a week, of which 1 will be a lecture and 1 will be a tutorial.
It is planned to have the lecture on Tuesdays and the corresponding tutorials on Thursdays.
It is planned to have the lecture on Tuesdays and the corresponding tutorials on Thursdays.
Examination
Final exam
Recommended literature
Lecture notes will be provided.