Dr. Hengjie Yang from Qualcomm was invited to give an online academic report on the topic of "Variable-Length Coding over Binary Channels with Full Feedback" in the morning of October 27th, 2022. This report is hosted by Lin Zhou, a professor from the School of Cyber Science and Technology, Beihang University. There are about 20 professors and students in total.
In the report, Dr. Yang first introduced variable length coding with feedback, and then introduced the small-ensive-difference (SED) coding scheme proposed by Naghshvar et al., pointing out that for symmetric binary input channels, The variable-length feedback (VLF) code constructed with the SED encoding scheme can asymptotically achieve the capacity and Burnashev's optimal error index. Based on this, Dr Yang extends the SED encoding scheme to binary asymmetric channels with completely noise-free feedback and obtains a non-asymptotic VLF reachability bound, which is shown to achieve both capacity and the best error index. In addition, for the special case of binary symmetric channels, Dr Yang obtained another non-asymptotic VLF reachability bound using a two-stage analysis method that utilizes both submartingale synthesis and Markov chain analysis. Next, Dr Yang presents simulation results showing that both new VLF reachability bounds exceed the reachability bounds of Polyanskiv.
At the end of the lecture, Dr. Yang answered the questions raised by the professors and students, and had an in-depth exchange and discussion with the professors and students on the ideas and details of SED coding.
【 Brief Introduction of the Speaker 】
Dr. Hengjie Yang is currently a Senior engineer at Qualcomm Inc. He received his bachelor's degree in communication engineering from Xidian University in 2017, followed by his master's degree and doctorate degree in electrical and computer engineering from UCLA in 2018 and 2022, respectively. Dr. Yang's current research interests include information theory and channel coding theory. His research topics include CRC-assisted convolutional code construction and list decoding, and variable length coding with finite code length. Dr. Hengjie Yang was awarded a Dissertation Year Fellowship from UCLA in 2021.