Phase Noise & Jitter: Understanding Cyclostationary Processes

119 Views Asked by At

My question relates to trying to understand the ways to characterize cyclostationary processes. Reference to the literature would be helpful!

My question has the following parts:

Part I: Phase Noise

Given the following definitions:

$$ \mathcal{L}(f) = {1 \over A^2} \int_{-\infty}^{\infty} d \tau R_{vv} (\tau) e^{-j (2 \pi f- \omega_0) \tau} $$ where $A$ is the normalization for dBc per Hz. $$ \mathcal{S}_\phi (f) = \int_{-\infty}^{\infty} d \tau R_{\phi \phi} (\tau) e^{-j 2 \pi f \tau} $$ $$v(t) = \cos( \omega_0 t + \phi(t))$$ $$R_{vv}(\tau) = \frac{1}{T}\int_{0}^T dt E[ v(t+ \tau) v(t)]$$ where $T = {2 \pi \over \omega_0}$ $$R_{\phi \phi} (\tau) = E[ \phi(t+ \tau) \phi(t)]$$

What is the relationship between $\mathcal{L}(f)$ and $S_\phi(f)$? Under what conditions are they similar, and under what conditions do they differ?

For this part at least, we can assume $\phi$ is Wide-Sense Stationary, although I am curious how this generalizes for cyclostationary $\phi$.

Part II: Jitter

Let us define the time series $ t_0, t_1,...$ as the deviation of the 0 crossing points times of the waveform v(t) from an ideal one with $\phi = 0$. Remember that v(t) is modulated by the stationary (or maybe even cyclostationary with period ${2 \pi \over \omega_0}$) process $\phi(t)$. We are interested in calculating the value $\sigma_t$, which is the standard deviation of the time series.

What is the expression that relates $\mathcal{L}(f)$ to $\sigma_t$? What is the expression that relates $\mathcal{S_\phi}(f)$ to $\sigma_t$? Under what conditions do these expressions break down?

Wiener Khintchine seems to suggest that $\sigma_t \propto \frac{1}{ \omega_0} \sqrt{\int_0^\infty \mathcal{S_\phi}(f) df}$. However I am not sure the integration limits if I replace $S_\phi$ with $\mathcal{L}(f)$

Also, what is the analog to Wiener Khintchine for cyclostationary $\phi(t)$?