Electrical and Computer Engineering Assistant Professor Imtiaz Ahmed, Ph.D., recently received the Excellent Paper Award at the 2021 International Conference on Artificial Intelligence in Information and Communication (ICAIIC 2021) for his paper titled “Joint Demodulation and Decoding with Multi-Label Classification Using Deep Neural Networks”. ICAIIC, co-sponsored by IEEE Communications Society, aims at addressing advances in research on information and communications technology, covering topics ranging from ICT technology issues to emerging AI applications for ICTs. The paper is co-authored by Wenjie Xu, Ramesh Annavajjala, Ph.D., and Wook-Sung Yoo, Ph.D.
“In this paper, we propose a deep learning-based receiver design that enhances the performance of conventional wireless communication receivers,” said Dr. Ahmed.
Dr. Ahmed and his research team at Howard University work on the design, optimization, and performance analysis of current (5G-NR) and next generation (beyond 5G) cellular communication systems. The team applies model-driven signal processing tools and data-driven artificial intelligence techniques in the transmitter and receiver design to push the performance boundary and alleviate the real-time computational complexity. Both fundamental and applied research works are conducted via rigorous simulations and in practical testbed while conforming the (3GPP) standardization requirement for cellular communication systems. The outcome of their research activities will open the door to developing cost-efficient, low-latency, and high-performance communication network design for beyond 5G cellular communication systems.
Current projects supervised by Dr. Ahmed include artificial intelligence (AI) based physical layer algorithm design, integrated aerial-terrestrial (drone, high altitude platform station) communication network design, hybrid free-space optics and radio frequency (RF) system design with the aerial network, intelligent reflecting surface (IRS) aided wireless network design and optimization, and resource control for cell-free massive multiple input multiple output (MIMO).
In his award-winning paper, Dr. Ahmed summarizes that joint baseband demodulation and decoding can be accomplished by developing a deep neural network (DNN) based multi-label classification algorithm in the communication receiver. The DNN can be trained offline with a large dataset over a wide range of signal quality. Once trained, the network can be applied online for real-time applications. Simulation results reveal that the designed algorithm outperforms the conventional scheme while limiting the real-time computational complexity.