Nonlinear Signal Processing


General Description

These two courses combine lectures and problem classes in nonlinear signal processing theory and applications. The theoretical part lays foundations in the area of memoryless nonlinearities (limiters/quantizers/classifiers), nonlinear dynamical systems (nonlinear filters/oscillators/chaos theory), nonlinear statistics (higher-order statistics and information theory), as well as parallel distributed processing (neural networks). The application part is devoted to the modeling of natural signals with nonlinear systems, algorithms for time series analysis and signal synthesis, and nonlinear devices in digital communications systems.

As a prerequisite, students should have prior knowledge in digital signal processing theory and, preferentially, in adaptive systems.

Lecture course


Lecture room: HS i12 "INFONOVA Hörsaal" (ICK1130H)

Lecture dates:

Day Date Start End Remark
Thu 05.03.2015 09:15 10:45  
Thu 12.03.2015 09:15 10:45  
Thu 19.03.2015 09:15 10:45  
Thu 26.03.2015 09:15 10:45  
Thu 23.04.2015 09:15 10:45  
Thu 30.04.2015 09:15 10:45  
Thu 07.05.2015 09:15 10:45  
Thu 21.05.2015 09:15 10:45  
Thu 28.05.2015 09:15 10:45  
Thu 11.06.2015 09:15 10:45  cancelled
Thu 18.06.2015 09:15 10:45  
Thu 25.06.2015 09:15 10:45  
Thu 02.07.2015 09:15 10:45  

 For downloading the lecture slides you need a password. If you have been registered for the course, you should have been received it.

Lecture  Slides
Course Organization






Memoryless Systems:
System Representation



Memoryless Systems:
Signal Characterization



Memoryless Systems:
Signal Processing

Fading Memory Systems:
System Representation


Fading Memory Systems:
Signal Characterization


Fading Memory Systems:
Signal Processing


Non-Fading Memory Systems:
System Representation


Non-Fading Memory Systems:
System Representation


Non-Fading Memory Systems:
Signal Processing


Please consult the TUGonline systems for exam dates.

Problem classes

The theoretical course is accompanied by practical problem classes. For the problem classes several meetings are planned, spanning four major blocks:
  1. Median filtering - Introduction into Nonlinear Signal Processing
  2. Memoryless nonlinearity and Higher Order Statistics (HOS)
  3. Fading memory systems and the corresponding HOS
  4. Infinite memory systems (oscillators, chaos ) and information theory

Grading is based on four homework assignments, one for each major block. You are strongly encouraged to work in pairs! The  due date for all four homeworks is TBA. Please submit your homeworks at our mailbox in Inffeldgasse 16c / ground floor. You can find a map HERE. Please also submit your code files to

Please use the newsgroup provided by the ZID at for discussion of the assignments.

What do you have to submit? A written report (one per group, a group consists of at most 2 students) containing your results, figures and the corresponding discussions for all problems marked with a house icon. For analytic problems you can also hand in hand-written sheets, as long as they are readable. Make sure your report contains enough information such that your results are traceable. Mail your Matlab-files to Take a look at and apply our homework guidelines to avoid unecessary point penalties.

Problem Classes material

Important: Please read and apply our SPSC homework guidelines

Lesson Handout Slides Files Sample Solutions
1 - Median Filtering  handout_1  slides_1  matlab_1  sample_solution_1
2 - Static Nonlinearities


 slides_2  matlab_2 sample_solution_2 
3 - Fading Memory Nonlinearities   handout_3  slides_3  matlab_3  sample_solution_3
4 - Infinite Memory Nonlinearities  handout_4  slides_4  matlab_4  sample_solution_4


Education Level: 
Master Level