Paper on Efficient Information Compression for MIMO Published in IEEE Wireless Communications Letters
The paper “CSI Compression Method With Dual Differential Feedback for Next-Generation Wi-Fi Networks” written by the members of the Wireless Networks Lab Andrey Barannikov, Ilya Levitsky, Vyacheslav Loginov, and Evgeny Khorov has been recently published in IEEE Wireless Communications Letters.
The paper explores a potential enhancement for the MIMO feature in Wi-Fi – channel sounding reduction. To enjoy the data rate acceleration thanks to MIMO, a Wi-Fi device needs to tell its access point the state of the wireless channel. Such a message has a lot of complex matrices, which, even when compressed using standardized 802.11 methods, can make up hundreds or thousands of bytes. This unwanted overhead can take considerable time to transmit.
The authors of the paper propose a new method to further compress the channel state information (CSI) in a lossless way. The insight is that CSI is highly correlated across the frequency subcarriers and consecutive (in time) feedbacks. So, to minimize its volume, only the differences in two dimensions can be used, which can further be compacted with a simple Huffman-like encoder. The paper reveals that the new method saves about 50% of feedback size on average based on real experimental data. The benefit of the solution its low complexity compared to most analogues and potential compatibility with the 802.11 standard.
Since our previous paper, our research has deepened considerably. We have set up a series of new experiments, held many discussions, reproduced existing methods for channel information and created new ones. We had to address related topics such as coding theory, statistics and digital signal processing. This allowed us to obtain new valuable data that will be useful in improving the quality of wireless information transmission systems. I am very glad that this colossal work has been appreciated and published in a top-ranking journal. Such recognition of our team efforts inspires to do more!
the first author of the article Andrey Barannikov comments.
The paper is published in IEEE Wireless Communications Letters, a first quartile (Q1) journal. It is the third work recently published by the Wireless Networks Laboratory members in a highly cited scientific journal. Read about the previous work below.
Wireless Networks Lab is a ‘Megagrant’ lab established in 2017 around the project on Cloudified Wireless Networks for 5G and beyond, led by Prof. Ian Akyildiz. Currently the WNL members, led by Evgeny Khorov, regularly present their results at leading scientific conferences, cooperate with Russian and international companies, and contribute to the standardization of wireless networks.