Ponente
Descripción
The Tile Calorimeter (TileCal), a central component of the ATLAS detector at the Large Hadron Collider (LHC), plays a crucial role in measuring the energy of hadronic particles produced in high-energy collisions. As the Large Hadron Collider (LHC) enters its High Luminosity phase (HL-LHC), the Tile Calorimeter (TileCal) will face substantial challenges arising from elevated radiation levels, increased data throughput, and unprecedented pile-up conditions. These conditions demand more efficient and robust signal processing techniques to ensure accurate energy reconstruction.
This study explores strategies to optimize the trade-off between FPGA resource usage and latency in the implementation of signal reconstruction algorithms. Accurate reconstruction of the calorimeter pulse is essential, as it directly reflects the energy deposited by particles. Currently, the Optimal Filtering (OF) algorithm is employed to extract pulse amplitudes from digitized samples. While OF has proven effective under nominal conditions, its performance deteriorates in high pile-up environments.
To address this limitation, the study investigates the potential of Neural Network-based approaches, specifically Single Layer Perceptrons (SLP) and Multilayer Perceptrons (MLP), trained on simulated TileCal pulse data, as alternatives to traditional methods.
