RTLMarker: Protecting LLM-Generated RTL Copyright via a Hardware Watermarking Framework
Kun Wang, Kaiyan Chang, Mengdi Wang, Xinqi Zou, Haobo Xu, Yinhe Han, Ying Wang
TL;DR
RTLMarker introduces a hardware watermarking framework to protect LLM-generated RTL copyright by embedding watermarks into RTL code and the synthesized gate-level netlist via semantic-preserving Verilog transformations. It combines a rule-based transformation set, a transformer-based feature representation, and a watermark detection network to jointly optimize watermark transparency and effectiveness, with additional netlist-level detection using synthesis traces. The approach demonstrates superior RTT and netlist watermarking performance against baselines, along with resilience to variable-name attacks and an improved transparency profile. This work offers a practical IP-protection solution for RTL design in the era of LLM-assisted chip development, bridging RTL validation and hardware watermarking at the gate level.
Abstract
Recent advances of large language models in the field of Verilog generation have raised several ethical and security concerns, such as code copyright protection and dissemination of malicious code. Researchers have employed watermarking techniques to identify codes generated by large language models. However, the existing watermarking works fail to protect RTL code copyright due to the significant syntactic and semantic differences between RTL code and software code in languages such as Python. This paper proposes a hardware watermarking framework RTLMarker that embeds watermarks into RTL code and deeper into the synthesized netlist. We propose a set of rule-based Verilog code transformations , ensuring the watermarked RTL code's syntactic and semantic correctness. In addition, we consider an inherent tradeoff between watermark transparency and watermark effectiveness and jointly optimize them. The results demonstrate RTLMarker's superiority over the baseline in RTL code watermarking.