Visually Secure Deep Joint Source-Channel Coding With Chaotic Map Against Deep Known-Plaintext Attack

In recent years, deep learning-based joint source-channel coding (DJSCC) has gained significant attention for its impressive performance in image transmission. Unlike traditional separate source-channel coding (SSCC) methods, DJSCC performs particularly well in low signal-to-noise ratio (SNR) and li...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:IEEE open journal of the Communications Society Ročník 6; s. 1847 - 1858
Hlavní autoři: Fu, Yuyang, Suto, Katsuya
Médium: Journal Article
Jazyk:angličtina
Vydáno: New York IEEE 2025
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Témata:
Channel coding
Chaos theory
Chaotic communication
Chaotic map encryption
Coding
Decoding
deep joint source-channel coding
Deep learning
Encoding
Encryption
Image coding
Image communication
Image quality
Image reconstruction
Image transmission
Incompatibility
known-plaintext attack
Receivers
Security
Signal to noise ratio
Tensors
visual security
ISSN:2644-125X, 2644-125X
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:In recent years, deep learning-based joint source-channel coding (DJSCC) has gained significant attention for its impressive performance in image transmission. Unlike traditional separate source-channel coding (SSCC) methods, DJSCC performs particularly well in low signal-to-noise ratio (SNR) and limited bandwidth environments. However, ensuring the security of private information during transmission remains a critical concern. A notable limitation of DJSCC is its incompatibility with traditional encryption methods used for secure communications, making it vulnerable to eavesdropping attacks. To address this issue, we propose integrating a chaotic map encryption method into the DJSCC framework for secure wireless image transmission. This approach leverages chaotic sequence to shuffle the position of the elements in latent space without altering the values of the latent tensor. This allows the encryption process to be designed independently of DJSCC, eliminating the need for re-training the end-to-end model. Our proposed method preserves DJSCC's superior transmission characteristics, ensuring high-quality image reconstruction at the receiver, while effectively ensuring the security against deep learning-based known plaintext attacks (Deep KPA).
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:2644-125X
2644-125X
DOI:10.1109/OJCOMS.2025.3548079