Recovery of varying-bandwidth signal for level-crossing sampling

In the paper, a method of reconstruction of a bandlimited signal based on a local bandwidth that can vary in time is proposed. Since the local signal bandwidth in many applications fluctuates, and over long time intervals might be significantly lower than the total signal bandwidth, the proposed approach allows to reconstruct the original signal at sub-Nyquist rate if the latter is related to the total bandwidth. It is shown that the local bandwidth may be estimated dynamically by monitoring the actual mean rate of level-crossing sampling. The estimation algorithm of local bandwidth for Gaussian bandlimited random process is presented. Furthermore, the estimation of local parameters of non-stationary random Gaussian process using sliding window is reported. The simulation results validate the proposed approach.