Biography

I earned my Masters of Applied Science degree from the Electrical and Computer Engineering Department from the University of Toronto in November 2022. Prior to that, I got my Bachelors of Science degree in Electrical Engineering with a minors in Mathematics from Sharif University of Technology, Tehran, Iran.

During my masters studies, I’ve been doing research under the supervision of Professor Stark Draper at the intersection of Optimization methods and Coding Theory, specifically using Alternating Direction Method of Multipliers (ADMM) in linear programming (LP) decoding of LDPC (Low-Density Parity-Check) codes. My thesis focused on reducing the algorithmic complexity of the so-called iterative ADMM-LP decoder through two approaches.

• Reducing per-iteration complexity by proposing a sparse affine projection algorithm (SAPA):
  In each iteration of the algorithm, a parity-polytope projection must be applied which is computationally intensive, with a time complexity of $O(d \log(d))$, where d is the dimension of the polytope. The proposed Sparse Affine Projection Algorithm (SAPA) approximates the projection step with great precision, such that the overall performance of the decoder is not significantly impacted while achieving a linear time complexity of $O(d)$.
• Reducing the number of iterations by proposing a randomized layered scheduling for node updates:
  Previous research in the field of iterative LDPC decoders has demonstrated the efficacy of scheduling node updates to improve convergence. While prior works have proposed deterministic strategies for scheduling updates, we proposed a randomized approach by introducing a probability mass function over the node indices. This randomized schedule allows the decoder to take advantage of the information provided by the Signal-to-Noise (SNR) ratio and adapt the schedule to fall somewhere between a completely greedy and completely agnostic schedule.