TU Delft
Year
NEDERLANDSENGLISH
Organization
Education Type
Education print this page print this page     
2016/2017 Electrical Engineering, Mathematics and Computer Science Bachelor Computer Science and Engineering
TI2716-A
Signal Processing
ECTS: 5
Responsible Instructor
Name E-mail
Prof.dr.ir. R.L. Lagendijk    R.L.Lagendijk@tudelft.nl
C.C.S. Liem    C.C.S.Liem@tudelft.nl
Contact Hours / Week x/x/x/x
2/0/0/0 hc; 2/0/0/0 wc; 4/0/0/0 lab
Education Period
1
Start Education
1
Exam Period
1
2
Course Language
English
Required for
Image Processing. Multimedia analysis. Highly recommended for Computer Science M.Sc. track Data Science and Cybersecurity specialisation.
Expected prior knowledge
Calculus: series, integration, complex numbers, complex exponents, Euler's formula. Linear algebra: vector and matrix manipulations. Special attention is asked for complex numbers. Self-check exercises and study material is provided via Blackboard.
Course Contents
Digital signal processing is used in many modern computer sciences systems and applications. Examples are content recommenders in information systems; speech, music and image content based retrieval and searching; music and video compression; sensor data processing in embedded systems; bioinformatics and medical data analysis. This course deals with the foundations and principles of digital signal processing. The first part concentrates on acquiring digital signals (sampling) and the basic linear filtering operations and convolution. The second part of the course introduces the concept of frequency or Fourier description of signals and systems. This concept is the foundation of many of today's computer science-based systems and applications, and has found wide applications in the processing of a variety of sound, music, sensor, image, video and other multimedia information.
Study Goals
1. Properties of digital signals and systems. The student is able to explain, compute and apply the properties of time-discrete linear time-invariant systems and the input-output relation; explain and use the principle of superposition; write arbitrary signals in terms of the sum of impulse signals.

2. Linear time-invariant (LTI) filters and convolution. The student is able to explain, determine and apply the impulse response of a time-discrete LTI system such as FIR or IIR filter; explain the relation between impulse response, input-output relation and convolution; manually (for short signals) or numerically (using Matlab, for long signals) convolve the input signal and the impulse response to compute the output signal; explain and apply the properties of convolution, including the difference between linear and circular convolution; recursively compute the output of an IIR filter; implement and visualize convolution via Matlab; implement and apply FIR filters in practical exercises and assignment, and visualize and interpret the results.

3. Sampling. The student is able to explain and apply the process of data acquisition (sampling) and Shannon's sampling theorem; explain the concepts of oversampling, undersampling and aliasing; explain how to avoid aliasing; use of zero-order or linear interpolation for the reconstruction of time-continuous signals; make audible (using Matlab) original and processed audio signals and interpret the results.

4. Discrete Fourier Transform (DFT). The student is able to write a periodic signal as the sum of complex exponentials; explain the concepts and differences of Fourier series, DTFT and DFT; explain and interpret the complex spectrum, magnitude spectrum and phase spectrum ; compute and visualize manually (for short signals) or numerically (for longer signals) the magnitude and phase of DTFT and DFT coefficients, and interpret the results; explain the difference between the numerical results computed using the DFT for finite-length signals and the theoretical results computed using the DTFT for infinite-length signals; explain the principles of the fast Fourier transform (FFT); apply the DTFT and DFT to practical exercises and assignments.

5. Frequency response of LTI filters. The student is able to compute manually (for short impulse responses) or numerically (for longer impulse responses) the frequency response of an LTI system using DTFT and DFT; visualize and explain the magnitude and phase response of an LTI system; evaluate convolution via product of frequency response of the filter and Fourier transform of the input signal; compute the frequency response of cascaded filters; numerically determine the impulse response of an FIR filter with desired frequency response using the inverse-DFT method.

6. Applications of filters and DFT. The student is able to apply theory and practice to (simplified) versions of relevant computer science-based signal processing problems and assignments, such as music compression and recognition.
Education Method
The course consists of
- 7 plenary lectures in which the main concepts are explained and discussed;
- 7 working group in which theory and practical (Matlab-based) exercises are solved and discussed;
- 6 hands-on sessions in which practical (Matlab-based)assignments need to be solved and approved by the instructor.
Computer Use
The course includes a compulsory set of computer (Matlab-based) assignments that need to be solved and approved by the instructor during the hands-on sessions. Matlab software is pre-installed on the computers available during the hands-on sessions. Students can also install Matlab software on their own computer.
Literature and Study Materials
The course loosely follows the book 'Signal Processing First' James H. McClellan, Ronald W. Schafer and Mark A. Yoder ISBN 0-13-120265-0 Prentice Hall. Alternatively, also on-line material may be used, such as 'The Scientist and Engineering´s Guide to Digital Signal Processing', S.W. Smith ( http://www.dspguide.com/pdfbook.htm ). Slides used during the lectures are made available via Blackboard. Theory and practical exercises, as well as the Matlab assignments are available for download via Blackboard. During the course several pointers to on-line MOOCs and video lectures will be given.
Practical Guide
Study material and the hands-on assignments can be downloaded via Blackboard.
Assessment
The first partial examination is in week 4, and deals with the materials covered in plenary lectures, working groups and Matlab assignments in week 1-3. The result of this partial exam counts for 30%.

The second partial examination is scheduled at the end of the quarter, and covers all course material with emphasis on the materials covered in plenary lectures, working groups and Matlab assignments in week 5-8. The result of this partial exam counts for 70%.

A resit exam, covering all materials of the plenary lectures, working groups and Matlab assignments, is scheduled at the end of the second quarter.

The Matlab hands-on assignments are integral part of the course and must be approved during the hands-on sessions, even if the student solves the assignments at home (using an own computer). Completion of the hands-on assignments are a prerequisite for entering the second partial exam or the resit exam.

The completed Matlab hands-on assignments are valid for an entire academic (study) year, but expire at the end of the academic (study) year. Failing to pass the course in one year implies redoing all Matlab hands-on assignments.
Permitted Materials during Tests
The lecturers have prepared one page with key equations (formuleblad). This page is permitted during the examination. It is also allowed to make notes on this one printed page.
Enrolment / Application
Enrolment is necessary for both partial examinations. Due to the cut-off dates, students are advised to immediately sign-up for the first partial examination in week 4. It is not necessary to enroll for the Matlab hand-on assignments.
Tags
Image processing
Information & Communication
Mathematics
Modelling
Project
Signals and Systems
Judgement
The final grade at the end of the quarter in which the course is lectured is determined as follows.
1. The grade (rounded to one decimal) of the first partial examination counts for 30%.
2. The grade (rounded to one decimal) of the second partial examination counts for 70%.
3. The Matlab hands-on assignments have to be completed and approved by the instructor. The approval of each assignment must be done within the week set for that assignment. In case of unforeseen circumstances (like illness) the instructor can allow for a delay of maximally one week, and only if such approval is obtained at the latest in the week the assignment should have been completed.

Validity of partial examination results and completed Matlab hands-on assignments.
* The results of the partial examinations expire at the end of the quarter in which the course was lectured if the Matlab hands-on assignments have not been completed and approved by the instructor before the second partial examination.
* The completed Matlab hands-on assignments are valid for an entire academic (study) year, but expire at the end of the academic (study) year.

The grade of the resit examination is determined as follows.
1. The grade of the resit examination.
2. The Matlab hands-on assignments completed before the resit examination. Students who had not yet completed the hands-on assignments during the regular sessions are invited to contact the instructor to make a single appointment to verify and approve all (remaining) assignments. These appointments take place in the weeks before the resit examination.
Self test
Fourier transforms, DTFT and DFT rely heavily on the description of signals as a sum of complex exponentials. It is important that the student brings operational knowledge of complex numbers (see first years Calculus). A set of self-tests is available as well as self-study materials. Complex exponents are assumed background knowledge and not discussed during the lectures.