Title: Time-Based Quantization for FRI and Bandlimited signals

Abstract: We consider the problem of quantizing samples of finite-rate-of-innovation (FRI) and bandlimited (BL) signals by using an integrate-and-fire time encoding machine (IF-TEM). We propose a uniform design of the quantization levels and show by numerical simulations that quantization using IF-TEM improves the recovery of FRI and BL signals in comparison with classical uniform sampling in the Fourier-domain and Nyquist methods, respectively. In terms of mean square error (MSE), the reduction reaches at least 5 dB for both classes of signals. Our numerical evaluations also demonstrate that the MSE further decreases when the number of pulses comprising the FRI signal increases. A similar observation is demonstrated for BL signals. In particular, we show that, in contrast to the classical method, increasing the frequency of the IF-TEM input decreases the quantization step size, which can reduce the MSE.