Task-oriented semantic variable compression communication scheme with salient region awareness over hollow-core...

Area of Science:

• Optical Communications
• Image Processing
• Signal Compression

Background:

• High-capacity optical fiber communication systems require efficient data compression.
• Existing methods often lack adaptability to semantic content importance.
• Semantic variable compression is crucial for optimizing data transmission.

Purpose of the Study:

• To propose a task-oriented semantic variable compression communication scheme.
• To leverage histogram contrast (HC) and compressive sensing (CS) for adaptive compression.
• To evaluate the scheme's performance in high-speed optical signal transmission.

Main Methods:

• Utilizing saliency detection (SD) to assess image block importance dynamically.
• Implementing adaptive compressive sensing (CS) sampling with variable compression ratios.
• Transmitting 8 Gb/s and 16 Gb/s orthogonal frequency division multiplexing (OFDM) signals over hollow-core fiber (HCF).

Main Results:

• Achieved bit error rate (BER) below the 7% HD-FEC threshold (3.8 × 10⁻³).
• Maintained high clarity in demodulated constellation diagrams and reconstructed images (PSNR > 30 dB, MSSIM > 0.9).
• Demonstrated up to 0.5 performance gain over fixed-ratio schemes using the semantic-MSSIM (S-MSSIM) metric.

Conclusions:

• The proposed semantic variable compression scheme effectively enhances optical communication quality.
• The adaptive approach based on saliency detection improves compression efficiency.
• The scheme shows significant potential for future high-capacity semantic optical fiber communication systems.