Prerequisites & Notation

Prerequisites for Chapter 15

This chapter combines LTI system theory with the spectral analysis of random processes developed in Chapters 13--14. The reader should be comfortable with convolution (time and frequency domain), the Wiener-Khinchin theorem, and the notion of wide-sense stationarity. We will also use Cauchy-Schwarz and basic optimization.

  • Wide-sense stationarity, autocorrelation, cross-correlation(Review ch13)

    Self-check: Can you state the WSS conditions and compute rxx(Ο„)r_{xx}(\tau) for a given process?

  • Power spectral density and the Wiener-Khinchin theorem(Review ch14)

    Self-check: Can you go from rxx(Ο„)r_{xx}(\tau) to Px(f)P_x(f) via Fourier transform and back?

  • LTI systems, impulse response, frequency response, convolution

    Self-check: Can you compute y(t)=h(t)βˆ—x(t)y(t) = h(t) * x(t) and relate it to yΛ‡(f)=hΛ‡(f)xΛ‡(f)\check{y}(f) = \check{h}(f)\check{x}(f)?

  • Cauchy-Schwarz inequality

    Self-check: Can you state and apply Cauchy-Schwarz for integrals: ∣∫fgβˆ—βˆ£2β‰€βˆ«βˆ£f∣2∫∣g∣2|\int f g^*|^2 \leq \int |f|^2 \int |g|^2?

  • White noise and its PSD(Review ch14)

    Self-check: Do you know that ideal white noise has Px(f)=N0/2P_x(f) = N_0/2 for all ff?

Notation for This Chapter

The following notation is used throughout Chapter 15. Symbols follow the FSP convention: lowercase subscripts for PSD arguments, boldface for vectors/matrices.

SymbolMeaningIntroduced
h(t)h(t), h[n]h[n]Impulse response of an LTI system (CT and DT)
hˇ(f)\check{h}(f)Frequency response (transfer function) of the LTI system
Px(f)P_x(f)Power spectral density of input process XX
Py(f)P_y(f)Power spectral density of output process YY
Pxy(f)P_{xy}(f)Cross-power spectral density
rxx(Ο„)r_{xx}(\tau)Autocorrelation of a CT WSS process
rxx[m]r_{xx}[m]Autocorrelation of a DT WSS process
N0N_0One-sided noise PSD (W/Hz)
SNR\text{SNR}Signal-to-noise ratio
Οƒ2\sigma^2Noise variance / noise power
EsE_sSignal energy
BNB_NNoise equivalent bandwidth
s(t)s(t)Known deterministic signal (matched filter context)
N(ΞΌ,Οƒ2)\mathcal{N}(\mu, \sigma^2)Gaussian distribution with mean ΞΌ\mu and variance Οƒ2\sigma^2