Digital Signal Processing - Concise Notes
UNIT I: Discrete-Time Signals and Systems
• Discrete-time sequences: Sequences indexed in discrete time (x[n]), examples include
unit impulse, unit step, exponential, and sinusoidal signals.
• Properties: Linearity, time-invariance, causality, stability, memoryless, invertibility.
• Interpolation and Decimation: Resampling techniques—interpolation increases the
sampling rate, decimation reduces it.
• LTI Systems: Characterized by impulse response h[n], system output via convolution
y[n] = x[n] * h[n].
• System Responses: ZIR (zero-input), ZSR (zero-state), total response = ZIR + ZSR.
• LCCDE: Linear Constant Coefficient Difference Equations used to describe discrete
systems.
• Sampling Theorem: Perfect reconstruction possible if sampling rate > 2× highest
frequency (Nyquist rate).
UNIT II: Z-Transform
• Z-Transform: X(z) = Σ x[n]z⁻ⁿ, used for analyzing discrete-time systems.
• Properties: Linearity, time shifting, scaling, convolution in time ↔ multiplication in Z-
domain.
• Inverse Z-transform: Partial fraction expansion, power series, long division methods.
• Transfer Function H(z): Ratio of output to input in Z-domain, describes system
behavior.
• Stability: System is stable if all poles lie inside the unit circle in Z-plane.
UNIT III: DFT and FFT
• DFT: Converts finite-duration sequences to frequency domain, X[k] = Σ x[n]e^(-
j2πkn/N).
• Properties: Periodicity, symmetry, linearity, circular convolution.
• Spectral Leakage: Due to non-integer cycles in windowed data.
• Zero Padding: Adds zeros to increase frequency resolution.
• Windowing: Reduces spectral leakage; common windows—Hamming, Hanning,
Blackman.
• FFT: Efficient algorithm for DFT computation. Radix-2 DIT and DIF algorithms reduce
complexity from N² to NlogN.
UNIT IV: FIR Filters
• FIR: Finite Impulse Response, always stable, potentially linear phase.
• Linear Phase: Ensures no phase distortion, achievable in FIR using symmetry.
• Window Method: Design LP, HP, BP, BR filters using window functions.
• Frequency Sampling: Places frequency response at specific points to interpolate filter.
• Optimal Filters: Minimizes error between desired and actual response (intro only).
UNIT V: IIR Filters
• IIR: Infinite Impulse Response, may be unstable, recursive structure.
• Design Methods: From analog filters using Impulse Invariance or Bilinear Transform.
• Finite Word Length Effects: Quantization can cause overflow, limit cycles.
• Notch Filter: Removes a narrow frequency band, useful in noise rejection.
• Time-Frequency Analysis: Short-time Fourier transform and wavelet (intro only).
• Implementation: Practical aspects using DSP processors or tools like
MATLAB/LabVIEW.
