Multiple-model coding scheme for electrical signal compression

Abstract

This paper proposes a low-latency Multiple-Model Coding (MMC) approach to compress sampled electrical signal waveforms under encoding rate constraints. The approach is window-based. Several parametric waveform models are put in competition to obtain a first coarse representation of the signal in each considered window. Then, different residual compression techniques are compared to minimize the residual reconstruction error. The model parameters are quantized, and the allocation of the rate budget among the two steps is optimized. Simulation results show that the proposed MMC approach achieves a higher signal-to-noise ratio than state-of-the-art solutions on periodic and transients signal waveforms.