Quantum Semantic Communication Beyond the Shannon-Wyner Channel Capacity

TL;DR

This work proposes and experimentally validate a quantum semantic communication scheme, applying it to 3D point clouds, and achieves a 46.30-fold efficiency gain over direct transmission, surpassing both Wyner and Shannon capacity limits.

Abstract

Quantum Secure Direct Communication (QSDC), a paradigm-shifting breakthrough in quantum communication, exploits quantum states for unmediated information transmission. Rooted in the inviolable fundamental laws of quantum mechanics, QSDC enables ultrasensitive detection of even the faintest eavesdropping attempts, guaranteeing true communication security solely when no interference exists. If eavesdropping or intrusion is detected mid-transmission, the system instantly alerts users and severs data flow, shielding them from unauthorized tracking and mitigating hacker threats. Over two decades, QSDC has seen extraordinary advancements, currently attaining kilobit-per-second transmission over 100 km of commercial optical fiber. However, its practical scalability remains constrained by insufficient transmission rates, a critical bottleneck. Semantic communication, which drastically boosts transmission efficiency by extracting core information features, nevertheless stays vulnerable to malicious intrusions. Integrating these paradigms promises to simultaneously enhance equivalent data rate and security. Herein, we propose and experimentally validate a quantum semantic communication scheme, applying it to 3D point clouds. It achieves a 46.30-fold efficiency gain over direct transmission, surpassing both Wyner and Shannon capacity limits. This breakthrough not only clears the path for large-scale QSDC deployment but also marks a pivotal milestone in quantum information science.

Figures (3)

• Figure 1: General framework for quantum semantic communication.
• Figure 2: Framework of the quantum semantic communication of point clouds.
• Figure 3: (a) Communication rate and QBER versus time of the QSDC system transmitting raw bits over a 50 km fiber link during 3 hours. (b) Transmission time and CD under different Numbers when $n=50$. Each Number represents 100 transmission rounds. In each round, three point clouds are selected from the dataset, encoded into a $3\times n$ semantic feature, and transmitted once. The transmission time, measured from the start of encoding to the completion of decoding. (c) Average encode time and average decode time under different Numbers when $n=50$. (d) Comparison of transmission time, encode and decode, and CD under different $n$ and batch sizes. Transmission time denotes the total time to complete all transmission tasks corresponding to each $n$. For batch size 3, all $n$ are tested; for batch size 32, only $n=10,20,50$ are tested. In each transmission, a batch of point clouds is transmitted, and the process is repeated $10261/\text{batch size}$ times. Encode and decode times and CD values are computed for $n=10,20,50,100,200,300$. Note that $n=6144$ corresponds to transmitting the original point clouds without encoding. (e) Transmission efficiency for different $n$ with the Shannon–Wyner channel capacity as reference. At a fiber distance of 50 km, the first dashed line at 560.20 kbps corresponds to the secrecy capacity of the QSDC system derived from the Shannon-Wyner entropy, while the second dashed line at 1496.53 kbps marks the mutual information between the quantum transmitter and quantum receiver, namely the Shannon capacity. (f) Visualization of reconstructed point clouds for different $n$.