Real-Time Cross-Layer Semantic Error Correction Using Language Models and Software-Defined Radio
2026-04-09 • Networking and Internet Architecture
Networking and Internet Architecture
AI summaryⓘ
The authors worked on a way to fix mistakes in messages by using both the signal's physical details and the meaning of the words together. They improved an earlier method called CL-SEC to work in real-time on actual radio hardware. To do this, they created software that can quickly get detailed signal info from the hardware and connect it to modern language models. Their tests showed that combining signal data with language understanding works better than using each by itself.
Language ModelsSemantic Error CorrectionLog-Likelihood RatiosSoftware-Defined RadioFPGAEncoder-Decoder ModelsCross-Layer FusionReal-Time InferenceMiddleware
Authors
Yuchen Pan, Yuyang Du, Yirun Wang, Shiqi Xu, Lihao Zhang, Soung Chang Liew
Abstract
As Language Models (LMs) advance, Semantic Error Correction (SEC) has emerged as a promising approach for reliable network designs. Yet existing methods prioritize intent over accuracy, falling short of verbatim recovery. Our recent work, Cross-Layer SEC (CL-SEC), addressed this by fusing physical-layer Log-Likelihood Ratios (LLRs) with semantic context, but its real-time feasibility remained unvalidated. This paper demonstrates CL-SEC on a live Software-Defined Radio (SDR) testbed, resolving implementation barriers with: 1) an SDR middleware enabling real-time LLR extraction from FPGA hardware, and 2) a generalized inference interface supporting modern encoder-decoder LMs. Real-world experiments confirm that the cross-layer fusion significantly outperforms either source alone.