Signal Processing - July 2016 - 73

MIT residential introductory graduate subject 6.341, the
textbook for which is Discrete-Time Signal Processing [17].
The residential course carries as a prerequisite the undergraduate subject 6.011, the textbook for which is Signals,
Systems, and Inference [18], which itself has as a prerequisite of the basic background of Signals and Systems.
The residential course 6.341 has a lineage and evolution
at MIT that go back to the 1970s, and a key goal in creating 6.341x has been to extend that lineage and evolution forward into the online realm. It is commonly expressed that
the MOOC and online teaching and learning landscape is
currently "the wild west." The territory is rapidly evolving
and currently in a highly experimental stage. With this in
mind, our overall goals with 6.341x included experimenting with online teaching and the edX platform with primary
emphasis on 1) having a positive impact on the residential
course at MIT and potentially residential courses elsewhere,
2) exercising and pushing the boundaries of the online edX
infrastructure and platform, and 3) making the content of
and experience with the MIT graduate subject more widely
accessible worldwide at the level at which it is presented residentially at MIT. To accomplish these goals, the development
of 6.341x evolved in four phases on which we elaborate shortly, after first describing the basic organization of the course.

Course organization
Outline
The course 6.341x is an 11-week graduate-level class divided
into 18 units. Each unit consists of several topic segments, outlined in Table 6. In a typical week of the spring 2015 MOOC,
students were provided with the following:
■■ Multiple courseware topics. Each topic consisted of a combination of brief exercises, text comments, and video segments
recorded in a "chalk and talk" format, extracted and edited from
lectures of the residential course at MIT, and with slides that
were digitally animated specifically for the MOOC. The exercises were interspersed among the video segments specifically
to allow students to verify their progress before moving on to
the next segment. Consistent with the MIT residential course,
lectures contained both mathematically oriented discussions and
live demonstrations of signal processing concepts.
■■ An overview video from the staff each week. Videos were
recorded in "talking head" style, providing an outline of the
week's topics, in addition to brief, high-level audio signal
processing demonstrations illustrating associated concepts.
■■ A set of homework problems. Problems included project-style
numerical signal processing problems that students completed using in-browser tools developed specifically for 6.341x.
In addition, three exams were given to evaluate performance. Accompanying the courseware was an online discussion forum on which we comment in more detail below.

Style and format
Elements of the courseware are depicted in Figure 5. Figure 5(b)
shows still frames captured from the lecture video segments,

Table 6. MIT 6.341x course outline.
Course Unit
Unit 1: Signals and systems in the time and frequency
domains
Unit 2: Allpass and minimum-phase systems
Unit 3: Discrete-time processing of continuous-time
signals
Unit 4: Sampling rate conversion
Unit 5: Quantization and oversampling
Unit 6: Signal-flow graph implementations of LCCDEs(a)
Unit 7: Lattice structures
Unit 8: IIR filter design
Unit 9: FIR filter design
Unit 10: Parametric signal modeling
Unit 11: The Levinson recursion
Unit 12: Multirate systems and polyphase structures
Unit 13: The DFT
Unit 14: Computation of the DFT
Unit 15: Spectral analysis
Unit 16: The TDDTFT(b) and modulated filter banks
Unit 17: Multirate and critically sampled filter banks
Unit E: Enrichment lectures

Release Date
Week 1
Week 2
Week 2
Week 3
Week 3
Week 4
Week 4
Week 5
Week 5
Week 6
Week 6
Week 7
Week 8
Week 9
Week 10
Week 11
Week 11
Weeks 7
and 10

(a)
LCCDEs: Linear constant-coefficient difference equations; (b)TDDTFT:
time-dependent discrete-time Fourier transform.

featuring chalk and talk-style video clips interspersed with lecture slides animated specifically for 6.341x, as well as in-class
signal processing demos. The figure also depicts an example
exercise that would fall between video segments, designed for
students to verify their understanding. Weekly introductory videos were recorded in a talking head format, shown in Figure 5(a).
Figure 5(c) and (d) illustrate interactive elements of two homework problems, in which students were prompted to choose
spectral analysis parameters and enter block diagrams.
The basic staff-student interaction model used in the spring
2015 MOOC is depicted in Figure 6. Referring to this figure,
the lowest-latency method for staff-student interaction was
through the online discussion forum. From the perspective of
students, homework problems and exercises provided instant
feedback about performance, although aggregate results about
student performance were viewable by the staff on a delayed
basis. There was also typically a one-week delay between filming and deploying week overview videos, due to the time associated with editing and audio transcription.

Course evolution
The development of 6.341x consisted of four phases that began
in the spring of 2013 and continued through the fall of 2015.
■■ Phase I. Work began in spring 2013 on the translation of significant content of 6.341, as used residentially, into digital
form to be used initially as an online augmentation to the residential course. The online platform for digital content was initially Open edX, the open-source platform used residentially
by MIT and other schools that mirrors the edX.org infrastructure. It was recognized at the outset that because the residential course covers a graduate-level subject, it would inherently

IEEE Signal Processing Magazine

July 2016

http://www.edX.org

Table of Contents for the Digital Edition of Signal Processing - July 2016