Beirut · 2015 to 2019
Where the message starts
I studied Electrical and Computer Engineering at the American University of Beirut and graduated with High Distinction. Along the way I helped run the robotics club's regional design challenge and TEDxAUB, and worked on street lighting for a village in south Lebanon.
Montreal · 2019 to now
A long way north
I came to McGill for a master's, then stayed for a PhD in Prof. Warren Gross's Integrated Systems for Information Processing Lab. Winters are long, which is good for writing papers.
The problem
Every channel is noisy
On the way, interference and fading flip bits. A 1 leaves the transmitter and a 0 arrives. Error-correcting codes add a little structure to every message so that the receiver can work out what was sent anyway.
Research · universal decoding
Guess the noise, not the message
Guessing random additive noise decoding (GRAND) tries the most likely noise patterns first and stops as soon as one gives a valid codeword. It works with any code. I co-wrote the Springer book on building it in hardware, and I also work on ordered statistics decoding.
Read the bookResearch · polar codes and HARQ
Bridges made of butterflies
Polar codes protect the control channels of 5G. My recent work lets fast successive-cancellation and list decoders handle incremental-redundancy HARQ, so a failed packet can be fixed by sending a few more bits instead of the whole thing again.
See the papersResearch · decoder hardware
It has to fit on a chip
An algorithm only matters if it can run at gigabits per second on a few square millimetres of silicon. I design decoder architectures for ASICs and FPGAs, and I co-invented an optimization-based polar decoder (US Patent 12,463,747).
Compare decoders belowMcGill · since 2020
Passing it on
I run the labs and tutorials for Computer Organization and Digital Systems, where students write ARM assembly and build hardware on FPGA boards. Across 370 course evaluations, students rate my help with the material at 4.2 out of 5.
Teaching evaluationsPlay
Decoder playground
Pick decoders and a code to compare frame error rate against area, throughput and energy, scaled to a common technology node.
Add simulated curves (), grouped by code
Metrics and scaling rules
For a design with throughput \(T\) (Gb/s), area \(A\) (mm²) and power \(P\) (mW), built in node \(n_1\) at supply \(V_1\):
$$\mathrm{AE} = \frac{T}{A}\ \left[\tfrac{\mathrm{Gb/s}}{\mathrm{mm}^2}\right], \qquad \mathrm{EE} = \frac{P}{T}\ \left[\tfrac{\mathrm{pJ}}{\mathrm{bit}}\right]$$Scaling to node \(n_2\) and supply \(V_2\) uses the area and energy factors
$$S_A = \left(\frac{n_2}{n_1}\right)^{2}, \qquad S_E = \left(\frac{V_2}{V_1}\right)^{2}\left(\frac{n_2}{n_1}\right)^{2}$$so that
$$A' = S_A\,A, \qquad \mathrm{AE}' = \frac{\mathrm{AE}}{S_A}, \qquad \mathrm{EE}' = S_E\,\mathrm{EE}, \qquad P' = S_E\,P, \qquad T' = T.$$Throughput is not scaled. Energy needs \(V_1\), so designs without a reported supply voltage drop out of energy and power plots while scaling is on. These are first-order CMOS rules (C.-H. Lin et al., IEEE TCAS-I, 2009). Treat scaled values as rough estimates.
Diamonds have a full FER curve. When only the Eb/N0 at FER = 10-5 is known, it is drawn as a cross. Each selected design gets a card under the plots with its full details and notes, and the last column of the table holds the notes too (click a note to expand it). The page address keeps your selection, so you can share a comparison.
Read
Selected papers
Lately
News
| Nov 04, 2025 | US patent granted: “Polar decoder and associated method” (US 12,463,747 B2), an optimization-based polar decoder developed with Huayi Zhou, Ryan Seah, and Warren J. Gross. |
|---|---|
| Jun 01, 2025 | Paper accepted at ICC 2025: “Hardware-friendly IR-HARQ for Polar SCL Decoders” |
| Sep 15, 2024 | Published in IEEE TCOMM: “Reduced-Complexity Projection-Aggregation List Decoder for Reed-Muller Codes” |
| Jul 01, 2024 | Published in IEEE JSAC: “Decoding of Polar Codes Using Quadratic Unconstrained Binary Optimization” |