About

Ken R. Duffy is a professor at Northeastern University with joint appointments in the Department of Electrical & Computer Engineering and the Department of Mathematics. He has served as interim chair of the Electrical & Computer Engineering department from November 2023 to June 2024 and is currently the chair of the Mathematics department since July 2025. His research focuses on information theory, error correction decoding, communications, and applied probability. He works in collaborative multidisciplinary teams to design, analyze, and implement algorithms using tools from probability, statistics, and machine learning, with applications in digital circuits and DNA. Duffy holds a B.A. in Mathematics (mod) and a PhD in Applied Probability, both from Trinity College Dublin. He previously served as a professor at the National University of Ireland, Maynooth, where he was the Director of the Hamilton Institute, an interdisciplinary research center, from 2016 to 2022. He has also been involved in leading research initiatives funded by organizations such as the Science Foundation Ireland and DARPA, and has received numerous awards for his research contributions, including best paper and demo awards at various conferences. Duffy is an associate editor for IEEE Transactions on Information Theory and IEEE Transactions on Molecular, Biological, and Multi-Scale Communications.

Research topics

• Computer Science
• Artificial Intelligence
• Algorithm
• Arithmetic
• Mathematics
• Speech recognition
• Computer engineering
• Theoretical computer science

Selected publications

• Control of antibody class switch recombination by quantitative integration of antigen signaling

  The Journal of Immunology · 2026-04-01

  articleOpen access

  Immunoglobulin class switch recombination by B lymphocytes requires the simultaneous expression of activation-induced cytidine deaminase (AID) and noncoding germline transcripts (GLT). Expression of both components is controllable through cytokine signalling and arise through stochastic mechanisms linked to cell division that can be predicted in mathematical models. Here we demonstrate that the same principles govern the mathematically predictable antigen-mediated regulation of isotype switching. We report both antigen affinity and antigen concentration regulated rates of AID expression and germline transcription with varying sensitivities and division-linked mechanisms. A model incorporating AID and germline transcript expression accurately predicted the rates of class switching in B cells across varying antigen affinities and dosages. Thus, class switch recombination is governed by a universal molecular logic that allows cytokine and antigen induced signals to be incorporated into a single predictive mathematical model that is robust to variable external signals to which B cells are exposed.

  PublisherDOI

• Guessing random additive noise decoding for digital data communication

  Foundations and Trends® in Integrated Circuits and Systems · 2026-04-21

  article1st authorCorresponding

  This monograph presents guessing random additive noise decoding (GRAND), a novel approach to error correction that has rapidly transitioned from theoretical development to hardware implementation. Unlike traditional error correction paradigms that co-design codes and decoders, GRAND operates on a code-agnostic basis by systematically guessing noise patterns in decreasing order of likelihood and querying whether inverting each guess from a received sequence yields a valid codeword. This approach enables maximum likelihood decoding for any linear or non-linear code with moderate redundancy. A distinctive feature of GRAND algorithms with soft input is their ability to generate accurate blockwise soft output in the form of probabilistic estimates of decoding correctness, even with a single decoding. This capability surpasses conventional approximations and enables critical applications including upgrading CRC codes from error detection to error correction, reducing undetected error rates, enabling approximate query orders and facilitating soft-input soft-output iterative decoding of long, powerful concatenated codes. The monograph systematically explains the theoretical foundations, algorithmic variants and hardware implementation principles that have enabled GRAND’s remarkably fast path from conception to multiple taped-out application-specific integrated circuits. Through an explanation of query ordering strategies for both hard and soft input scenarios, performance evaluation across various codes and applications to iterative decoding, this work elucidates why GRAND’s inherent parallelizability and elegant mathematical properties make it suitable for efficient circuit implementation. The treatment is designed to clarify fundamental principles while providing practical insights for researchers and engineers working in error correction coding with an eye towards VLSI design.

  PublisherDOI

• Salience, legitimacy and credibility of single-cell sperm data and their evaluation

  Forensic Science International Genetics · 2026-02-28

  articleOpen access
  PublisherOA PDFDOI

• Practical AI-based cell extraction and spatial statistics for large 3D bone marrow tissue images

  Cell Reports Methods · 2026-03-01

  articleOpen access

  Although the molecular regulation of hematopoiesis is well characterized, the spatial organization of hematopoietic cells within bone marrow (BM) remains unclear. Advances in microscopy have produced increasingly detailed images of murine BM, yet accurate and scalable methods to extract and analyze these complex datasets are limited. We present PACESS, a computational workflow for BM analysis that combines convolutional neural networks for 2D cell detection and classification with an automated method to extrapolate into 3D, spatial statistical analyses to define tissue regions based on local cell-type densities, and logistic regression to assess whether the relative abundances of cell types reflect reciprocal dependencies. Using PACESS, we investigate the spatial organization of T cells, megakaryocytes, and leukemic cells, revealing that distinct leukemic clusters generate diverse, previously unrecognized neighborhoods within the same BM cavity. PACESS, thus, provides a powerful tool to dissect BM architecture.

  PublisherDOI

• Efficient Soft-Output Guessing for Enhanced Quantum Tanner Code Decoding

  arXiv (Cornell University) · 2026-03-18

  preprintOpen accessSenior author

  We introduce a generalized low-density parity-check decoding framework for quantum Tanner codes utilizing soft-output guessing random additive noise decoding (SOGRAND). By soft-output decoding entire component codes, we mitigate trapping sets and cycles, resulting in improved convergence. SOGRAND, combined with ordered statistic decoding (OSD) post-processing, outperforms the standard belief propagation plus OSD baseline by up to three orders of magnitude in logical error rate, providing a way forward for scalable decoding of the emerging class of Tanner-code-based quantum codes.

  PublisherDOI

• Efficient Soft-Output Guessing for Enhanced Quantum Tanner Code Decoding

  ArXiv.org · 2026-03-18

  articleOpen accessSenior author

  We introduce a generalized low-density parity-check decoding framework for quantum Tanner codes utilizing soft-output guessing random additive noise decoding (SOGRAND). By soft-output decoding entire component codes, we mitigate trapping sets and cycles, resulting in improved convergence. SOGRAND, combined with ordered statistic decoding (OSD) post-processing, outperforms the standard belief propagation plus OSD baseline by up to three orders of magnitude in logical error rate, providing a way forward for scalable decoding of the emerging class of Tanner-code-based quantum codes.

  PublisherOA PDF

• An Optimal Modulation Bits-to-RF Digital Transmitter Using Time-Interleaved Multi-Subharmonic Switching

  IEEE Journal of Solid-State Circuits · 2026-02-18

  article

  This article presents a fully integrated bits-to-RF transmitter (Tx) featuring deep power back-off (PBO) enhancements, leveraging a multi-subharmonic switching (multi-SHS) digital power amplifier (DPA) with time-interleaving and a harmonic-rejection digital-to-phase converter (DPC). This work employs a nonuniform optimal modulation (OM) constellation, where symbol probability is inversely related to its amplitude, and symbols transmitted more frequently are assigned to shorter bit lengths, collectively reducing the transmission error rate. The system, operating from 2.6- to 1.3-V VDDs, achieves 22.7-dBm peak output power with 58.1% peak power-added efficiency (PAE) and 52% peak system efficiency (SE), implemented in 65-nm CMOS technology. The system is able to support both QAM and OM modes of operation with high efficiency. Dynamic measurements using 64-/256-QAM constellations achieved 23.2%/19.7% PAE and 19.7%/16.8% SE at 17.1/16.4-dBm average output power, while maintaining an error vector magnitude (EVM) of −29.7/−30.75 dB at 1.5-GHz carrier frequency. In addition, the transmission of 64-/256-point OM constellations with 23.1%/19.6% PAE and 19.4%/16.5% SE at 16.9/16.2-dBm average output power, with an EVM of −29.9/−31.37 dB was demonstrated. In comparison of standard QAM, the proposed 64-point OM scheme reduces the symbol error rate (SER) by 12.4–<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$15.4{\times }$</tex-math> </inline-formula> and the bit error rate (BER) by 5.4–<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$7.3{\times }$</tex-math> </inline-formula> through the use of padding bits and a Guessing random additive noise decoder (GRAND)-assisted demodulator, across 22–26-dB receiver (Rx) signal-to-noise ratio (SNR) levels. In addition, 256-point measurements across Rx SNR levels of 30–34.6 dB demonstrated 6–<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$8.5{\times }$</tex-math> </inline-formula> lower SER and 1.4–<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$2{\times }$</tex-math> </inline-formula> lower BER, showcasing the ability to deliver highly efficient and reliable communication across various modulation orders.

  PublisherDOI

• Soft-Output Threshold-Guided CRC Decoding with SOGRAND in 40nm CMOS

  2026-04-19

  article

  We extend Cyclic Redundancy Checks (CRCs) beyond error detection by enabling threshold-guided error correction with the SOGRAND chip. Traditionally, CRCs treat any codeword that passes the CRC check as valid, and request retransmission whenever the check fails. GRAND-family decoders use a noise-guessing approach to recover packets that fail the CRC check; however, undetected errors can occur when an incorrect codeword satisfies the CRC check, resulting in a penalty in the undetected error rate. SOGRAND decodes CRCs and computes soft-outputs to derive an accurate a posteriori probability for each decoded result. Applying a threshold on the a posteriori probabilities enables control over retransmission requests and the undetected error rate. Fabricated in 40nm CMOS, SOGRAND achieves 0.84pJ/bit energy and 11.3ns latency, outperforming the state-of-the-art decoders while enabling confidence-aware CRC error correction through accurate soft-output computation.

  PublisherDOI

• Almost Sure One-Endedness of a Random Graph Model of Distributed Ledgers

  Stochastic Systems · 2025-08-14 · 1 citations

  articleOpen accessSenior author

  Distributed ledgers (DLs) are modern decentralized databases where trusted network members can transparently update data while maintaining security. IOTA is an exemplar alternative to the well-known Bitcoin protocol, and such alternatives, which aim to overcome shortcomings, have gained increasing attention recently. These systems warrant deeper theoretical analysis using directed acyclic graph (DAG) models. One essential property of a properly functioning DL is that all network members holding a copy of the database agree on the sequence of information added, which is referred to as consensus and is known to be related to a structural property of DAGs called one-endedness. In this paper, we consider a model of a DL with sequential stochastic arrivals that mimic IOTA’s attachment rules. Although the resulting DAG model is more complex than Bitcoin, we demonstrate that as time goes to infinity, the IOTA DAG almost surely achieves one-endedness.

  PublisherDOI

• ORBGRAND with Prog-ROCC for High-Bandwidth Channels

  2025-10-26

  article

  We propose an efficient iterative channel decoder based on ORBGRAND for high-bandwidth communications systems. In communications systems with high bandwidth, such as THz or optics, or in extremely large-scale MIMO systems with many receive elements, receiver output symbol rate can outpace the interface bandwidth of a channel decoder. Channel decoding according to a reliable subset of the received symbols is a means of reliable communication without overwhelming the decoder interface. The proposed decoder intakes the minimum amount of redundancy required to obtain a target block error rate. The block error rate remains fixed over a large SNR interval, and the redundancy level required for reliable decoding decreases as SNR increases. This behavior supports low-cost, reliable decoding in links where transmission bandwidth outstrips the interface bandwidth of available decoding hardware.

  PublisherDOI

Frequent coauthors

• Muriel Médard

  Massachusetts Institute of Technology

  144 shared

• Leïla Perié

  46 shared

• Giulio Prevedello

  28 shared

• Philip D. Hodgkin

  Walter and Eliza Hall Institute of Medical Research

  24 shared

• David Malone

  National University of Ireland, Maynooth

  23 shared

• Douglas J. Leith

  22 shared

• Mark M. Christiansen

  National University of Ireland, Maynooth

  20 shared

• Flávio P. Calmon

  Harvard University

  20 shared

Education

• Ph.D. Probability Theory, Mathematics

  University of Dublin Trinity College

  2000

• B.A.(mod) Mathematics, Mathematics

  University of Dublin Trinity College

  1996

Awards & honors

• Best paper award, IEEE Military Communications Conference (2…
• Best demo award, COMSNETS (2023)
• Best research demo award, COMSNETS (2022)
• Best paper award, IEEE Transactions on Network Science and E…
• Cover article, Trends in Cell Biology (2012)