A Reconfigurable Neural Network ASIC for Detector Front-End Data Compression at the HL-LHC

@article{Guglielmo2021ARN,
  title={A Reconfigurable Neural Network ASIC for Detector Front-End Data Compression at the HL-LHC},
  author={Giuseppe Di Guglielmo and Farah Fahim and Christian Herwig and Manuel Blanco Valent{\'i}n and Javier Mauricio Duarte and Cristian V. Gingu and Philip C. Harris and Jim Hirschauer and Martin Kwok and Vladimir Loncar and Yingyi Luo and Llovizna Miranda and Jennifer Ngadiuba and Dan Noonan and Seda Ogrenci-Memik and Maurizio Pierini and Sioni Summers and Nhan Viet Tran},
  journal={IEEE Transactions on Nuclear Science},
  year={2021},
  volume={68},
  pages={2179-2186}
}
Despite advances in the programmable logic capabilities of modern trigger systems, a significant bottleneck remains in the amount of data to be transported from the detector to off-detector logic where trigger decisions are made. We demonstrate that a neural network (NN) autoencoder model can be implemented in a radiation-tolerant application-specific integrated circuit (ASIC) to perform lossy data compression alleviating the data transmission problem while preserving critical information of… 

Figures and Tables from this paper

Waveform processing using neural network algorithms on the front-end electronics
TLDR
This work aims to predict peak amplitude from a single waveform snippet whose rising and falling edges containing only 3 to 4 samples, and thoroughly studied two well-accepted neural network algorithms, Multi-Layer Perceptron (MLP) and Convolutional Neural Network (CNN) by varying their model sizes.
Smart sensors using artificial intelligence for on-detector electronics and ASICs
TLDR
The motivations and potential applications for on-detector AI, and a number of areas of opportunity where machine learning techniques, codesign workflows, and future microelectronics technologies which will accelerate design, performance, and implementations for next generation experiments are discussed.
Data-driven detector signal characterization with constrained bottleneck autoencoders
TLDR
This article presents how deep learning in the form of constrained bottleneck autoencoders can be used to learn the underlying unknown detector response model directly from data using a waveform toy model.
Physics Community Needs, Tools, and Resources for Machine Learning
Machine learning (ML) is becoming an increasingly important component of cutting-edge physics research, but its computational requirements present significant challenges. In this white paper, we
Electronics for Fast Timing
Picosecond-level timing will be an important component of the next generation of particle physics detectors. The ability to add a 4 th dimension to our measurements will help address the increasing
Frontiers in computing for artificial intelligence
TLDR
These technologies offer promising approaches to address some of the leading concerns of future computing that may impact the Electron-Ion Collider but they will require further development and testing in order to support future planning efforts.
A Living Review of Machine Learning for Particle Physics
TLDR
This living review is a nearly comprehensive list of citations for those developing and applying deep learning approaches to experimental, phenomenological, or theoretical analyses, and will be updated as often as possible to incorporate the latest developments.
Which metric on the space of collider events?
Which is the best metric for the space of collider events? Motivated by the success of the Energy Mover’s Distance in characterizing collider events, we explore the larger space of unbalanced optimal

References

SHOWING 1-10 OF 31 REFERENCES
Automatic deep heterogeneous quantization of Deep Neural Networks for ultra low-area, low-latency inference on the edge at particle colliders
TLDR
A novel method for designing optimally heterogeneously quantized versions of deep neural network models for minimum-energy, high-accuracy, nanosecond inference and fully automated deployment on chip is introduced.
Fast inference of deep neural networks in FPGAs for particle physics
TLDR
A case study for neural network inference in FPGAs focusing on a classifier for jet substructure which would enable, among many other physics scenarios, searches for new dark sector particles and novel measurements of the Higgs boson.
Early-Stage Neural Network Hardware Performance Analysis
TLDR
A hardware performance analysis framework for identifying bottlenecks in the early stages of CNN hardware design and how the proposed method can help in evaluating different architecture alternatives of resource-restricted CNN accelerators early in design stages and, thus, prevent making design mistakes is introduced.
High-Level Synthesis for FPGAs: From Prototyping to Deployment
TLDR
AutoESL's AutoPilot HLS tool coupled with domain-specific system-level implementation platforms developed by Xilinx are used as an example to demonstrate the effectiveness of state-of-art C-to-FPGA synthesis solutions targeting multiple application domains.
Dual-Interlocked Logic for Single-Event Transient Mitigation
TLDR
Three exemplar radiation-hardened-by-design synchronous systems using DIL, spatial triple modular redundancy (TMR), and temporal TMR are compared across area, power, delay, and hardness relative to an unhardened baseline system and the system utilizing DIL exhibits desirable tradeoffs compared to spatial andporal TMRs.
Characterization of radiation effects in 65 nm digital circuits with the DRAD digital radiation test chip
A Digital RADiation (DRAD) test chip has been specifically designed to study the impact of Total Ionizing Dose (TID) (<1 Grad) and Single Event Upset (SEU) on digital logic gates in a 65 nm CMOS
Current and Future Challenges in Radiation Effects on CMOS Electronics
Advances in microelectronics performance and density continue to be fueled by the engine of Moore's law. Although lately this engine appears to be running out of steam, recent developments in
The CMS experiment at the CERN LHC
The Compact Muon Solenoid (CMS) detector is described. The detector operates at the Large Hadron Collider (LHC) at CERN. It was conceived to study proton-proton (and lead-lead) collisions at a
LHC and HL-LHC: Present and Future Radiation Environment in the High-Luminosity Collision Points and RHA Implications
The high-luminosity large hadron collider (HL-LHC) is a novel machine configuration which will rely on a number of key innovative technologies to enhance the performance of the present LHC machine as
A New Approach to Single Event Effect Tolerance Based on Asynchronous Circuit Technique
TLDR
The results of fault injection experiments indicate that DMR has a relatively low overhead on both area and latency than TMR, while tolerates SEU in sequential logic, while TSTMR-D provides tolerance for both SEU and SET with reasonable area andLatency overhead.
...
...