Skip to main content

NEB; mean and AC of WSS signal through LTI

Short answer

Noise equivalent bandwidth is B_N = ∫|H(f)|^2 df / |H(0)|^2. For a WSS input, output mean is constant μ_Y = μ_X H(0), and output autocorrelation R_Y(τ) = h(τ)*h(-τ)*R_X(τ) depends only on τ, so output is WSS.

Define noise equivalent bandwidth. Find mean and AC function at the output when a WSSP signal is passed through the LTI system.

Noise equivalent bandwidth

The noise equivalent bandwidth (NEB) BNB_N of a filter is the bandwidth of an ideal rectangular filter, with the same peak (or zero-frequency) gain, that passes the same noise power as the actual filter when both are driven by the same white noise.

If white noise of PSD N0/2N_0/2 is applied, actual output power =N02∫−∞∞∣H(f)∣2df= \frac{N_0}{2}\int_{-\infty}^{\infty}|H(f)|^2df and ideal output power =N02∣H(0)∣2⋅2BN= \frac{N_0}{2}|H(0)|^2 \cdot 2B_N. Equating:

BN=∫0∞∣H(f)∣2 df∣H(0)∣2B_N = \frac{\int_0^{\infty}|H(f)|^2\,df}{|H(0)|^2}
 |H(f)|^2
  |H(0)|^2 +-------+         area of rectangle
           |\      |         = area under |H(f)|^2
           | \     |
           |  \__  |
           |     ~~|~~~___
           +-------+------------> f
           0       BN

Example: RC-LPF has BN=1/(4RC)=1.57f3dBB_N = 1/(4RC) = 1.57 f_{3dB}. Then output noise power is simply N0BN∣H(0)∣2N_0 B_N |H(0)|^2.

Mean and autocorrelation at the output

Let a WSS process X(t)X(t) (mean μX\mu_X, autocorrelation RX(τ)R_X(\tau)) drive a stable LTI system with impulse response h(t)h(t):

Y(t)=∫−∞∞h(α)X(t−α) dαY(t) = \int_{-\infty}^{\infty} h(\alpha)X(t-\alpha)\,d\alpha

Mean

μY=E[Y(t)]=∫−∞∞h(α)E[X(t−α)] dα=μX∫−∞∞h(α) dα=μXH(0)\mu_Y = E[Y(t)] = \int_{-\infty}^{\infty} h(\alpha)E[X(t-\alpha)]\,d\alpha = \mu_X\int_{-\infty}^{\infty}h(\alpha)\,d\alpha = \mu_X H(0)

The output mean is the input mean times the DC gain, a constant.

Autocorrelation

RY(t,t+τ)=E[Y(t)Y(t+τ)]=∫ ⁣ ⁣∫h(α)h(β) E[X(t−α)X(t+τ−β)] dα dβ=∫−∞∞∫−∞∞h(α)h(β) RX(τ+α−β) dα dβ\begin{aligned} R_Y(t, t+\tau) &= E[Y(t)Y(t+\tau)] \\ &= \int\!\!\int h(\alpha)h(\beta)\,E[X(t-\alpha)X(t+\tau-\beta)]\,d\alpha\,d\beta \\ &= \int_{-\infty}^{\infty}\int_{-\infty}^{\infty} h(\alpha)h(\beta)\,R_X(\tau+\alpha-\beta)\,d\alpha\,d\beta \end{aligned}

This depends only on τ\tau, so Y(t)Y(t) is also WSS. In compact form:

RY(τ)=h(τ)∗h(−τ)∗RX(τ)R_Y(\tau) = h(\tau) * h(-\tau) * R_X(\tau)

Power spectral density (Fourier transform, using h(−τ)↔H∗(f)h(-\tau) \leftrightarrow H^*(f)):

SY(f)=H(f)H∗(f)SX(f)=∣H(f)∣2SX(f)S_Y(f) = H(f)H^*(f)S_X(f) = |H(f)|^2 S_X(f)

Output power

E[Y2(t)]=RY(0)=∫−∞∞∣H(f)∣2SX(f) dfE[Y^2(t)] = R_Y(0) = \int_{-\infty}^{\infty}|H(f)|^2S_X(f)\,df

For white input, SX=N0/2S_X = N_0/2, this gives N0∣H(0)∣2BNN_0|H(0)|^2B_N, which links back to the NEB.

✓ Answer checked by Claude Opus 5.5 (high effort)

All Communication Systems chapter 9 questions →