
Hessam Mahdavifar
VerifiedNortheastern University · Electrical and Energy Engineering
Active 2009–2026
About
Hessam Mahdavifar is an Associate Professor in the Department of Electrical and Computer Engineering at Northeastern University College of Engineering, having joined the faculty in August 2023. His research focuses on coding and information theory, wireless communications, data privacy and security, as well as privacy-preserving computing and learning. He holds a PhD and MS in Electrical and Computer Engineering from the University of California San Diego, and a BS in Electrical Engineering from Sharif University of Technology. Throughout his career, Dr. Mahdavifar has received several honors and awards, including the CAREER Award from the National Science Foundation in 2020, the Qualcomm Innovation Fellowship in 2020, and a Best Paper Award at the IEEE Conference on RFID in 2015. He is recognized as one of the top scientists worldwide, being among the top 2% of most-cited scientists according to Stanford University in 2025. His research contributions include significant work in physical layer secret key generation, analog secret sharing, and coding for non-coherent wireless networks, among other areas.
Research topics
- Computer science
- Algorithm
- Theoretical computer science
- Mathematics
- Discrete mathematics
Selected publications
Efficient Low-Memory Fast Stack Decoding with Variance Polarization for PAC Codes
IEEE Open Journal of the Communications Society · 2026-01-01
articleOpen accessSenior authorPolarization-adjusted convolutional (PAC) codes have recently emerged as a promising class of error-correcting codes, achieving near-capacity performance particularly in the short block-length regime. In this paper, we propose an enhanced stack decoding algorithm for PAC codes that significantly improves parallelization by exploiting specialized bit nodes, such as rate-0 and rate-1 nodes. For a rate-1 node with <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">N<sub>0</sub></i> leaf nodes in its corresponding subtree, conventional stack decoding must either explore all 2<italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><sup>N<sub>0</sub></sup></i> paths, or, same as in fast list decoding, restrict attention to a constant number of candidate paths. In contrast, our approach introduces a pruning technique that removes candidate paths with small path metrics while ensuring that the probability of pruning the correct path decays exponentially with the threshold. Furthermore, we propose a novel approximation method for estimating variance polarization under the binary-input additive white Gaussian noise (BI-AWGN) channel. Leveraging these approximations, we develop an efficient stack-pruning strategy that selectively preserves decoding paths whose bit-metric values align with their expected means. This targeted pruning substantially reduces the number of active paths in the stack, thereby decreasing both decoding latency and computational complexity. Numerical results demonstrate that for a PAC(128, 64) code, our method achieves up to a 70% reduction in the average number of paths without degrading error-correction performance.
IEEE Microwave and Wireless Technology Letters · 2026-01-01
articleFuture integrated sensing and communication (ISAC) systems require simultaneous multibeam operation with low-latency hardware and robust isolation under synchronization error and fading. Conventional code-division multiplexing using Walsh–Hadamard codes is extremely time-sensitive. This letter demonstrates that conventional temporal-only coded multibeam arrays result in inter-beam sidelobe level (SLL) collapsing to within a few decibels of the main lobe and varying by more than 10–20 dB over delay. By embedding moderate-length Gold sequences into a spherical spatial codebook, the proposed Spherical-Gold scheme leverages both <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$1/\sqrt{N}$</tex-math> </inline-formula> temporal and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$1/\sqrt{M}$</tex-math> </inline-formula> spatial bounds, achieving effectively <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$1/$</tex-math> </inline-formula>(<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\sqrt{N} \cdot \sqrt{M}$</tex-math> </inline-formula>) correlation without increasing RF complexity. Measurement results and verification have been performed using an Analog Devices ADAR3002 Ka-band 256-element receiver, where four simultaneous beams demonstrate <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${\ge}15$</tex-math> </inline-formula> dB beam isolation with <<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\pm 2.5$</tex-math> </inline-formula> dB variation under timing error and fading, while temporal-only code-division multiple-access (CDMA) degrades to −5 to −7 dB inter-beam SLL with <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\gt \pm 8$</tex-math> </inline-formula> dB variation under timing error.
Abelian Group Codes for Classical-Quantum Channels: One-Shot and Asymptotic Rate Bounds
IEEE Transactions on Information Theory · 2026-04-23
articleSenior authorWe study the problem of transmission of information over classical-quantum (CQ) channels in the one-shot regime where the underlying codes are constrained to be shifted <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">group codes</i>. Given a group <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">G</i>, a group code of length <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">n</i> is a subgroup of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">G<sup>n</sup></i>. In the achievability part, we introduce a new collection of input probability distributions that incorporates the encoding homomorphism and the underlying channel law. Using a random coding argument, we characterize the performance of group codes in terms of hypothesis testing relative-entropic quantities. In the converse part, we establish bounds by leveraging a hypothesis testing-based approach. Furthermore, we apply the one-shot result to the asymptotic stationary memoryless setting, and establish a single-letter lower bound on the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">group capacity</i> of a CQ channel. Moreover, we derive a matching upper bound on the asymptotic group capacity.
ArXiv.org · 2026-03-20
articleOpen accessFuture integrated sensing and communication (ISAC) systems require simultaneous multibeam operation with low-latency hardware and robust isolation under synchronization error and fading. Conventional code-division multiplexing using Walsh-Hadamard codes is extremely time-sensitive. This paper demonstrates that conventional temporal-only coded multibeam arrays suffer from inter-beam sidelobe level (SLL) collapse to within a few dB of the main lobe, with variations exceeding 10-20 dB over delay. By embedding moderate-length Gold sequences into a spherical spatial codebook, the proposed Spherical-Gold scheme leverages both temporal and spatial correlation bounds, achieving effective inter-beam isolation without increasing RF complexity. Measurement results and verifications are performed using an Analog Devices ADAR3002 Ka-band 256-element receiver with four simultaneous beams. The proposed scheme demonstrates at least 15 dB rejection with less than 2.5 dB variation in SLL under time error and fading, whereas temporal-only CDMA degrades to approximately -5 to -7 dB SLL with nearly 8 dB variation under time delay.
arXiv (Cornell University) · 2026-03-20
preprintOpen accessFuture integrated sensing and communication (ISAC) systems require simultaneous multibeam operation with low-latency hardware and robust isolation under synchronization error and fading. Conventional code-division multiplexing using Walsh-Hadamard codes is extremely time-sensitive. This paper demonstrates that conventional temporal-only coded multibeam arrays suffer from inter-beam sidelobe level (SLL) collapse to within a few dB of the main lobe, with variations exceeding 10-20 dB over delay. By embedding moderate-length Gold sequences into a spherical spatial codebook, the proposed Spherical-Gold scheme leverages both temporal and spatial correlation bounds, achieving effective inter-beam isolation without increasing RF complexity. Measurement results and verifications are performed using an Analog Devices ADAR3002 Ka-band 256-element receiver with four simultaneous beams. The proposed scheme demonstrates at least 15 dB rejection with less than 2.5 dB variation in SLL under time error and fading, whereas temporal-only CDMA degrades to approximately -5 to -7 dB SLL with nearly 8 dB variation under time delay.
Cost-Aware Neural Early Stopping for Local Constraint OSD Decoders
ArXiv.org · 2026-03-20
articleOpen accessSenior authorLocal constraint ordered statistics decoding (LC-OSD) provides strong soft decision performance for short block length linear codes, but its practical cost is dominated by the number of tested error patterns (TEPs). This paper proposes a neural early stopping (NES) protocol for LC-OSD with explicit cost control through one trade-off parameter balancing frame error risk and search effort. The proposed approach is trained with frame error rate (FER)-aligned supervision at predefined checkpoints, and learns if additional search is still likely to improve the current best candidate. Later, stopping is decided by comparing predicted continuation need with a cost measured in TEPs. Experimental results across multiple code families show that the proposed protocol significantly reduces average TEP count with only marginal FER degradation, using a single global model for the range of all operating signal-to-noise ratios (SNRs).
Univariate Bicycle Quantum LDPC Codes: Explicit Logical Structure and Distance Bounds
arXiv (Cornell University) · 2026-05-13
preprintOpen accessSenior authorWe introduce univariate bicycle (UB) codes, a structured subclass of generalized bicycle (GB) quantum low-density parity-check (LDPC) codes obtained via a Frobenius relation. This construction reduces the code design space from a two-polynomial search in GB codes to a single-polynomial search, while preserving sparsity. We provide an explicit algebraic characterization of the logical coset spaces by constructing a basis for the logical quotient space, yielding a complete parametrization of logical operators. Leveraging this structure, we derive upper bounds on the minimum distance by relating structured logical representatives to cycle-density properties of associated circulant matrices. Finally, simulation results for short- to medium-length UB codes (block lengths ranging from a few hundred to approximately $10^3$) demonstrate competitive performance relative to existing GB and bivariate bicycle (BB) codes despite the additional algebraic restriction.
Univariate Bicycle Quantum LDPC Codes: Explicit Logical Structure and Distance Bounds
ArXiv.org · 2026-05-13
articleOpen accessSenior authorWe introduce univariate bicycle (UB) codes, a structured subclass of generalized bicycle (GB) quantum low-density parity-check (LDPC) codes obtained via a Frobenius relation. This construction reduces the code design space from a two-polynomial search in GB codes to a single-polynomial search, while preserving sparsity. We provide an explicit algebraic characterization of the logical coset spaces by constructing a basis for the logical quotient space, yielding a complete parametrization of logical operators. Leveraging this structure, we derive upper bounds on the minimum distance by relating structured logical representatives to cycle-density properties of associated circulant matrices. Finally, simulation results for short- to medium-length UB codes (block lengths ranging from a few hundred to approximately $10^3$) demonstrate competitive performance relative to existing GB and bivariate bicycle (BB) codes despite the additional algebraic restriction.
On the Security of Linear Secret Sharing with General Noisy Side-Channel Leakage
Lecture notes in computer science · 2026-01-01
book-chapterSenior authorMultiple-Bases Belief Propagation List Decoding for Quantum LDPC Codes
ArXiv.org · 2026-05-13
articleOpen accessSenior authorIn this paper, we propose a belief-propagation (BP)-based decoder, termed the Multiple-Bases Belief-Propagation List Decoder (MBBP-LD), for quantum low-density parity-check (QLDPC) codes. The key idea is to generate \emph{structured decoding diversity} by constructing multiple redundant parity-check representations via cycle-free subtree decompositions of the Tanner graph, and running BP decoding in parallel across these representations. This extends the classical Multiple-Bases Belief-Propagation (MBBP) framework to the quantum setting while preserving the linear-time complexity and efficiency of standard BP decoding, and avoids the need for super-linear post-processing. Simulation results demonstrate that MBBP-LD improves upon existing BP-based decoders, including BP with ordered statistics decoding (BP-OSD) and belief propagation with guided decimation (BPGD) across several QLDPC codes, while requiring substantially fewer total BP iterations. For bivariate bicycle codes $[[144,12,12]]$ and $[[288,12,18]]$, MBBP-LD achieves up to $20\%$ reduction in error rate compared to BPGD and up to $30\%$ compared to BP-OSD in the low- and moderate-error regimes. For the larger B1 code $[[882, 24, 18 \leq d \leq 24]]$, MBBP-LD attains comparable or improved performance relative to BPGD while maintaining BP-like decoding latency under parallel implementation.
Recent grants
CAREER: Coding Subspaces: Error Correction, Compression and Applications
NSF · $467k · 2024–2027
CIF: Medium: Collaborative Research: New Frontiers in Polar Coding: 5G and Beyond
NSF · $575k · 2018–2023
Collaborative Research: CIF: Small: Designing Plotkin Transform Codes via Machine Learning
NSF · $300k · 2023–2025
CAREER: Coding Subspaces: Error Correction, Compression and Applications
NSF · $516k · 2020–2024
NSF · $250k · 2019–2023
Frequent coauthors
- 37 shared
Mohammad Vahid Jamali
University of Michigan–Ann Arbor
- 33 shared
Mahdi Soleymani
University of California, San Diego
- 26 shared
Alexander Vardy
University of California, San Diego
- 19 shared
Inyup Kang
Samsung (South Korea)
- 19 shared
Mostafa El‐Khamy
- 16 shared
Jungwon Lee
Seoul National University
- 15 shared
James Chin-Jen Pang
- 14 shared
A. Salman Avestimehr
University of Southern California
Education
- 2014
Ph.D., Electrical and Computer Engineering
Northeastern University
- 2009
M.S., Electrical Engineering
University of Tehran
- 2007
B.S., Electrical Engineering
University of Tehran
Awards & honors
- CAREER Award, National Science Foundation (2020)
- Qualcomm Innovation Fellowship (2020)
- Best Paper Award, IEEE Conference on RFID (2015)
- Silver Medal, International Mathematical Olympiad (2002 and…
- Resume-aware match score
- Save to shortlist
- AI-drafted outreach
See your match with Hessam Mahdavifar
PhdFit ranks faculty by your research interests, methods, and publications — grounded in their actual work, not templates.
- Free to start
- No credit card
- 30-second signup