Real-time Anomaly Detection at the L1 Trigger of CMS Experiment

TL;DR

This paper demonstrates the deployment of a low-latency autoencoder-based anomaly detection system in the CMS experiment's trigger system during LHC Run 3, enabling real-time identification of potential new physics signatures without disrupting data collection.

Contribution

It introduces a novel integration of a neural network-based anomaly detection into the LHC trigger system, achieving real-time processing within strict latency constraints.

Findings

Successful deployment of autoencoder in FPGA hardware

Real-time detection of anomalous events during proton collisions

Validation of the system's performance in live LHC data

Abstract

We present the preparation, deployment, and testing of an autoencoder trained for unbiased detection of new physics signatures in the CMS experiment Global Trigger (GT) test crate FPGAs during LHC Run 3. The GT makes the final decision whether to readout or discard the data from each LHC collision, which occur at a rate of 40 MHz, within a 50 ns latency. The Neural Network makes a prediction for each event within these constraints, which can be used to select anomalous events for further analysis. The GT test crate is a copy of the main GT system, receiving the same input data, but whose output is not used to trigger the readout of CMS, providing a platform for thorough testing of new trigger algorithms on live data, but without interrupting data taking. We describe the methodology to achieve ultra low latency anomaly detection, and present the integration of the DNN into the GT test crate, as well as the monitoring, testing, and validation of the algorithm during proton collisions.