Compressed IF-TEM: Time Encoding Analog-To-Digital Compression

TL;DR

This paper introduces a novel analog compression method for IF-TEM based on signal stationarity, significantly improving reconstruction quality or reducing bit usage in time encoding of bandlimited signals.

Contribution

It presents a new analog compression technique for IF-TEM that enhances low-bit-rate reconstruction by exploiting signal stationarity before quantization.

Findings

Achieves 5-20dB MSE improvement with same samples and slightly more bits.

Enables 1-2 fewer bits for fixed MSE and sample count compared to classical IF-TEM.

Demonstrates empirical effectiveness of the compression method.

Abstract

An integrate-and-fire time-encoding-machine (IF-TEM) is an energy-efficient asynchronous sampler. Utilizing the IF-TEM sampler for bandlimited signals, we introduce designs for time encoding and decoding with analog compression prior to the quantization phase. Before the quantizer, efficient analog compression is conducted based on the stationarity of the encoded signal, which is a fundamental characteristic of IF-TEM processing. Low-bit-rate reconstruction is achieved by subdividing the known IF-TEM dynamic range into tighter windows, which can be either fixed size or dynamically changed, and detecting in which window the sample resides. We demonstrate empirically that employing the same number of samples and up to 7% additional bits than the conventional IF-TEM results in a 5-20dB improvement in MSE. Fixing the reconstruction MSE target and the number of samples, using the compressed IF-TEM enables the use of 1-2 fewer bits compared to the classical IF-TEM.