Skip to main content

Engineering Probability and Random Processes · Correlation, power spectra, and linear systems

Work with two-sided angular frequency

Problem

Engineering Probability and Random Processes practice problem

Work with two-sided angular frequency. On the band ∣ω∣≤1|\omega|\le 1 the real mean-zero jointly wide-sense stationary processes S,N1,N2S,N_1,N_2 are mutually uncorrelated and have constant spectraSS(ω)=1,SN1(ω)=1,SN2(ω)=4,S_S(\omega)=1,\qquad S_{N_1}(\omega)=1,\qquad S_{N_2}(\omega)=4,and all three spectra vanish for ∣ω∣>1|\omega|>1. Define the observationsX=S+N1,Y=2S+N2.X=S+N_1,\qquad Y=2S+N_2.Mutual uncorrelatedness means that every cross-spectrum among distinct letters {S,N1,N2}\{S,N_1,N_2\} is identically zero. Magnitude-squared coherence is∣γXY(ω)∣2=∣SXY(ω)∣2SXX(ω) SYY(ω)|\gamma_{XY}(\omega)|^2 =\frac{|S_{XY}(\omega)|^2}{S_{XX}(\omega)\,S_{YY}(\omega)}at frequencies where the denominator is positive. Convert a spectrum to a lag-zero moment by the standing conventionR(0)=12π∫−∞∞S(ω) dω.R(0)=\frac1{2\pi}\int_{-\infty}^{\infty}S(\omega)\,d\omega.1. Compute the three spectra SXXS_{XX}, SYYS_{YY}, and SXYS_{XY} on
∣ω∣≤1|\omega|\le 1. Then evaluate the magnitude-squared coherence on that band.

2. Using the displayed R(0)R(0) convention, compute Var⁡(X)\operatorname{Var}(X),
Var⁡(Y)\operatorname{Var}(Y), and Cov⁡(X,Y)\operatorname{Cov}(X,Y).

3. Compute the ordinary time-domain correlation
Corr⁡(X(t),Y(t))\operatorname{Corr}(X(t),Y(t)).

4. Audit both sentences: (i) “magnitude-squared coherence equals the
time-domain correlation 1/21/2, or else equals the unsquared coherence ∣SXY∣/SXXSYY|S_{XY}|/\sqrt{S_{XX}S_{YY}}”; (ii) “R(0)R(0) is the raw integral of SS with no factor 1/(2π)1/(2\pi), so the variances are 44 and 1616 (or 22 and 88).”

Do not replace the two-channel observation by a single-process geometric covariance, and do not invoke a telegraph spectrum.

Hint

Hint

Uncorrelated sources add in power: SXX=1+1=2S_{XX}=1+1=2, SYY=4⋅1+4=8S_{YY}=4\cdot 1+4=8, SXY=2⋅1=2S_{XY}=2\cdot 1=2.

Check your work

Work the problem yourself first. Then open it in Training to check your answer and read the full worked solution.

The answer check and full solution for this problem come with ProofAnvil Practice membership ($19 USD monthly). See membership. Or start with the free Engineering Probability and Random Processes sample problem: Try the free sample problem.

More Engineering Probability and Random Processes practice problems