• Version
• Download 14
• File Size 556.63 KB
• File Count 1
• Create Date 10 December 2025
• Last Updated 10 December 2025

Review on Vision Transformer Based Semantic Communications for Next Generation Wireless Networks

 

 

Akhilesh1, Mr Pradeep Nayak2, Adithi Shetty3, Ankitha K N4 ,  Anushree K5

Faculty, Department of Information Science and Engineering2

Students, Department of Information Science and Engineering1,3,4,5

Alva’s Institute of Engineering and Technology, Mijar, Mangalore, Karnataka, India.

Email: anukiran959@gmail.com

 

 

Abstract: Semantic communications, which prioritize the transfer of semantic meaning above raw data, have the potential to completely transform data transmission in the developing 6G network landscape. data precision. This research introduces a semantic communication system based on Vision Transformers (ViTs) that has been specifically built to minimize bandwidth consumption while achieving high semantic similarity during image transmission. The suggested architecture may effectively encode images into a high semantic content at the transmitter and accurately reconstruct the images at the receiver while taking noise and real-world fading into account by using ViT as the encoder-decoder framework. Our approach beats Convolution Neural Networks (CNNs) and Generative Adversarial Networks (GANs) designed for producing such images, building on the attention processes built into ViTs. With a Peak Signal-to-noise Ratio (PSNR) of 38 dB, the architecture based on the suggested ViT network outperforms existing Deep Learning (DL) techniques in preserving semantic similarity across various communication settings. These results prove that our ViT-based method is a major advancement in semantic communications. Vision Transformer (ViT), Deep Learning (DL), 6G, semantic communication, bandwidth efficiency, and Peak Signal to Noise Ratio (PSNR) are the index terms.

Keywords: Semantic Communication, Deep Learning (DL), Image Transmission, Bandwidth Efficiency

 

Download