FPGA-Based Data Compression for Large-Scale Nuclear Physics Experiments

Abstract

In nuclear physics experiments, FPGAs control a large number of front-end electron ics boards for signal processing. This project provides a generic code for adaptive downsampling of signal traces. The performance of the code is characterized quanti tatively in sampling frequency and resource usage for different FPGAs. The process of the data compression is based on the noise estimate and a trigger level to decide the compression ratio so that smaller sample groups are used to store the signal data in regions dominated by critical signal information and larger lengths of sam ple groups are averaged in areas containing noise. The method and implementation of the compression are described in detail with both VHDL and Python.