UniC-Vision: A 14.4Gb/s 7.3pJ/b Universal Vision Transformer OFDM Channel Estimation Accelerator for B5G/6G AI-RAN

Digest of technical papers/Digest of technical papers - IEEE International Solid-State Circuits Conference 2026

View at publisher
Authors: Sangbu Yun, Chanhee Lee, Soonhyun Kwon, Jian-Fei Chang, Zhengya Zhang, Youngjoo Lee
DOI: 10.1109/isscc49663.2026.11409283
Cited by: 0
References: 8

Abstract

This work presents an AI-RAN channel estimation accelerator for next-generation communications, offering hyper reliability, low latency, and universal frequency range coverage, thereby meeting B5G/6G requirements. Utilizing algorithm-hardware co- optimization and architecture optimizations, the prototype achieves a throughput of 14.4 Gb/s at 7.3 pJ/b in 28 nm CMOS, surpassing the state-of-the-art channel estimation systems by 7.7 to 18.9 dB in reliability and 16 -to-$39 \times$in throughput.

Fields of Study & Subjects

Advanced Wireless Communication Techniques Advanced Neural Network Applications Radio Frequency Integrated Circuit Design

Publication details

Journal / source
Digest of technical papers/Digest of technical papers - IEEE International Solid-State Circuits Conference
Pages
48–50
Publication date
15 February 2026
Language
EN