Nonlinear filters for pulse amplitude extraction in FERMI

Linear-nonlinear hybrid filters for pulse amplitude extraction and their optimization using a training approach are presented. A simulated detector signal, corrupted by thermal and pile-up noise, sample timing jitter and the time-domain superposition of pulses is applied to the hybrid filters for performance evaluation. The optimal filtering in high-energy physics data acquisition is also discussed and the problem of simultaneously obtaining good time resolution and high amplitude precision is reviewed. The FIR-order statistic hybrid filter structure with three subfilters is implemented as part of the FERMI readout microsystem for energy extraction from high-resolution calorimeters