Source coding is a key technology for mobile radio, MP3, JPG, and video transmission. We discuss the theoretical foundations for lossy source coding (rate distortion theory and asymptotic quantisation theory) as well as practical coding schemes and vector quantisation. For practically relevant correlated source signals we extend the theory and we also discuss practical code design that are structurally based on results from information theory. Those schemes include predictive coding, which is used e.g. in speech transmission in mobile radio, as well as subband coding, which is part of MP3-Audio.
The lectures are delivered (in blocked format) on Tuesday, 2pm, in room CG0118. Details of the schedule will be announced in the first lecture.
First lecture: Tuesday, 10 October 2017
Schedule of lectures (14:00--16:15, 3 hours, room CG118):
10 Oct, 17 Oct, 24 Oct, 7 Nov, 14 Nov, 21 Nov, 28 Nov, 5 Dec., 12 Dec, 19 Dec
Nominal teaching hours: 15 × 2 = 10 x 3 = 30 hours.
Printed copies of the lectures slides will be available for the students of the course.
Desirable prior knowledge: discrete-time signals and systems, basics of information theory.
There are no formal compulsory requirements.
