Sparse deconvolution methods for online energy estimation in calorimeters operating in high luminosity conditions

Abstract

Energy reconstruction in calorimeters operating in high-luminosity particle colliders has become a significant challenge. In this scenario, pulses from the calorimeter front-end output overlap (pile-up effect), compromising energy estimation when no preprocessing for signal separation is applied. Recently, signal deconvolution-based methods have been proposed for both online and offline reconstruction. For online processing, constraints such as processing speed, memory requirements, and implementation cost limit overall performance. In contrast, offline reconstruction allows the use of Sparse Representation theory to implement more sophisticated iterative deconvolution methods. This paper presents iterative deconvolution approaches based on sparse representation algorithms with computational costs suitable for online implementation. Using simulated data, the proposed methods are compared with current techniques to evaluate performance in online environments. The analysis shows that, despite higher computational cost compared to standard methods, the performance gains may justify their use—particularly the Separable Surrogate Functional approach, which proves to be feasible for implementation on modern FPGAs.