better538website
Processed class transcripts: original summaries of the first five available recordings
These are original summaries of locally available transcripts, not transcript copies. Later meetings will appear here only if a transcript becomes available.
Lecture 1 — Course orientation and DSP applications
Topics discussed
- Course systems, communication channels, assessments, and MATLAB-based work.
- The semester's application-driven scope: speech, radar, GPS, wireless systems, OFDM, filtering, compression, and spectral estimation.
- Why discrete-time convolution, correlation, sampling, and Fourier analysis recur across these applications.
- A speech-signal/MATLAB demonstration connecting sampled values, waveforms, spectra, spectrograms, and reconstructed audio.
Associated official material
Course outline · Course syllabus · speech and MATLAB resources
Lecture 2 — Discrete-time signals, frequency, and LTI setup
Topics discussed
- Sequences as integer-indexed samples; sampling interval and sampling rate.
- Sampled sinusoids, discrete-time angular frequency, frequency equivalence modulo 2π, and aliasing intuition.
- Discrete-time periodicity; unit step, Kronecker delta, finite rectangles, and geometric sequences.
- Linearity, time invariance, LTI systems, and the role of the impulse response.
Associated official material
Lecture 3 — Discrete-time convolution
Topics discussed
- Derivation of convolution from shifted Kronecker deltas, linearity, and time invariance.
- Weighted shifted impulse responses and flip–slide–multiply–sum as complementary views.
- Finite-sequence output length, overlap, index roles, and convolution properties.
- Convolution matrices, MATLAB computation, FFT computation, and the distinction between linear and circular convolution.
Associated official material
Lecture 4 — LTI consequences, FIR filters, and correlation preview
Topics discussed
- Convolution matrices and MATLAB conventions; shifted/scaled inputs and outputs.
- Cascade and parallel LTI systems; causality and BIBO stability through the impulse response.
- Difference equations, FIR filters, and impulse-response coefficients.
- Correlation, matched filtering, radar, GPS, and delay detection.
Associated official material
Lecture 5 — Deterministic correlation, matched filters, and Barker codes
Topics discussed
- Deterministic autocorrelation and cross-correlation as tools for locating a known finite sequence within sampled, noisy data.
- Matched filtering as convolution with the time-reversed signal (and conjugation for complex signals), with cross-correlation as the corresponding sliding comparison.
- Radar-style delay detection: each returned copy produces a scaled autocorrelation peak at its delay; multiple echoes give multiple peaks.
- Finite-length autocorrelation: a length-N sequence produces 2N−1 lag values; for real sequences it is even-symmetric, and its magnitude is largest at zero lag.
- Why constant-envelope phase-coded sequences are useful with nonlinear power amplifiers; ±1 chip sequences, Barker codes, and their low sidelobes.
- Doppler-shifted radar returns and searching a grid of candidate Doppler frequencies to form a delay–Doppler profile.
- MATLAB demonstrations of Barker-code autocorrelations and the invariance of autocorrelation under time shifts and time reversal.
Associated official material
Week 2 file group · Basics of Autocorrelation · Autocorrelation Properties · BarkerCodesClass.m