Suggested Model Improvements

 Apart from modelling a more commonly used phaselock loop by assuming a non unit choice of transfer function, it is believed that one major improvement that could be made is in modelling the filtering of the noise by each clock in the synchronisation network. This could be achieved by assuming that the input reference signal is filtered by each PLL's closed loop transfer function before being propagated to the subsequent clock. This would involve an extra noise term at the model output

$$S_o(f)=S_i(f)\vert H(2\pi fj)\vert^2$$

where S_i(f) is the input time error noise power spectral density and $H(2\pi fj)$ is the closed loop transfer function.