Comparison of two hardware-based hit filtering methods for trackers in high-pileup environments

TL;DR

This paper compares two hardware-based hit filtering methods, Associative Memory and Hough transform, for particle tracking in high pile-up environments, evaluating their efficiency and robustness through simulations.

Contribution

It provides a detailed comparison of AM and Hough transform methods for hit filtering in high energy physics tracking, highlighting the advantages of AM in reducing hit combinations.

Findings

AM produces fewer hit combinations than Hough transform.

Both methods maintain high efficiency despite increased detector support material.

Efficiency drops slightly when detector modules are deactivated, regardless of pile-up level.

Abstract

As experiments in high energy physics aims to measure increasingly rare processes, the experiments continually strive to increase the expected signal yields. In the case of the High Luminosity upgrade of the LHC, the luminosity is raised by increasing the number of simultaneous proton-proton interactions, so-called pile-up. This increases the expected yields of signal and background processes alike. The signal is embedded in a large background of processes that mimic that of signal events. It is therefore imperative for the experiments to develop new triggering methods to effectively distinguish the interesting events from the background. We present a comparison of two methods for filtering detector hits to be used for triggering on particle tracks: one based on a pattern matching technique using Associative Memory (AM) chips and the other based on the Hough transform. Their efficiency and hit rejection are evaluated for proton-proton collisions with varying amounts of pile-up using a simulation of a generic silicon tracking detector. It is found that, while both methods are feasible options for an efficient track trigger, the AM based pattern matching produces a lower number of hit combinations with respect to the Hough transform whilst keeping more of the true signal hits. We also present the effect on the two methods when increasing the amount of support material in the detector and introducing inefficiencies by deactivating detector modules. The increased support material has negligible effects on the efficiency for both methods, while dropping 5% (10%) of the available modules decreases the efficiency to about 95% (87%) for both methods, irrespectively of the amount of pile-up.