Hamid Saber

About Hamid Saber

Wireless communications•Machine learning, Artificial intelligence, deep learning •Channel coding: Classical and modern: Polar, LDPC, Turbo• Mathematics, algebra, stochastic processes and probability theory, coding theory, information theory• Programming and debugging: Matlab, C, C++• 3GPP NR standardization RAN1

Experience

1. Senior Staff Engineer

   Samsung Semiconductor

   Jan 2019 — Present · San Diego, CA, US

   AI/ML in 5G/6G RAN– Channel Coding∗ Designed, trained and evaluated the performance of aa AI/ML channelauto-encoder, based on NR QC-LDPC codes for QAM modulation· With only 44 power scaling learnable weights it provides between.5to.75 dB gain over 5G NR QC-LDPC codes∗ Won ICC 2022 best paper award for Product-AE· A new loss function, adaptive scaled norm, for BLER minimization∗ polar/PAC code design via RL. This provided up to.3 dB gain over5G NR polar codes under SCL decoding– CSI∗ Designed Auto-encoder for CSI compression and prediction– QAM Detection∗ Trained and evaluated a learnable LLR scaling QAM detector with end-to-end LDPC coded BLER minimization objective– Federated Learning∗ Designed and analyzed multi-step Meta/personalized FL for heteroge-neous FL for fast online-training for 6G application· This improved classification accuracy on MNIST, CIFAR data setcompared to FedAvg and Per-FedAvg– Generative AI Application to Communication∗ Diffusion model. Flow-Matching (FM) Generative model with OT conditional VF/path. • 3GPP 5G/6G3GPP 5G/6G RAN1 delegate, attending RAN1 meeting, 100+ system designpatents; applications and granted, and 3GPP contributions on agenda items:– AI/ML use case on RAN1 including CSI prediction, CSI compression andbeam prediction in spatial and time domain, joint source and channel coding– 6GR channel coding and modulation and AI/ML use cases– ultra-reliable low latency communication (uRLLC) Release 16 and 17– Carrier aggregation (CA) enhancement Release 16– MIMO Multi-TRP PDCCH, PDSCH, PUSCH, PUCCH enhancement release16 and 17• UE Modem Legacy PHY Algorithm– Designed and implemented a simplified SCL (SSCL) decoder for polar codeswith new special nodes identical performance to SCL decoder– Designed a simplified low latency SCL decoder for PAC codes with identicalperformance to SCL decoder – Implemented a linear programming (LP) decoding of polar codes under onthe fundamental polytope

Education

• Amirkabir University of Technology - Tehran Polytechnic

  Bachelor of Science (B.Sc.), Electrical, Electronics and Communications Engineering

  2003 — 2007

• Carleton University

  Doctor of Philosophy (PhD), Electrical Engineering

  2011 — 2016

• University of Tehran

  Master's degree, Electrical engineering

• University of Tehran

  Master of Science (M.Sc.), Electrical, Electronics and Communications Engineering

  2007 — 2010

Skills

• C++
• Multi User Detection: Cdma Tdma, Ofdm
• Wireless Communications Systems
• Matlab
• Digital Signal Processors
• Latex
• Fpga
• Cryptography
• C
• Programming
• Signal Processing
• Simulink
• Simulations
• Algorithms
• Linux
• Modelsim
• Theory of Detection and Estimation
• Digital Signal Processing
• Spread Spectrum
• Vlsi
• Verilog
• Pspice
• Error Control Coding and Information Theory