Abstract: Simulation testing is a critical technology for verifying the safety and reliability of autonomous driving systems. To address the data leakage caused by complete exposure of scenario data during this process, a data obfuscation protection method for simulation test scenarios was proposed, along with a corresponding three-tier obfuscation strategy. This method incorporates a series of obfuscation techniques, including data re-encoding, name replacement, sequence shuffling, label reconstruction, trigger condition obfuscation, and event obfuscation. Furthermore, this method defines a three-level obfuscation strategy based on obfuscation intensity, significantly enhancing scenario data security. Experimental results demonstrated that the simulation outcomes for obfuscated scenario data were consistent with those of the original data. As the obfuscation level increased, the degree of data protection also improved. The first and second-tier obfuscation methods had no significant impact on simulation efficiency, whereas the third-tier method introduced a slight delay in simulation execution time. Overall, all levels of the obfuscation methods maintained reasonable simulation performance while effectively preventing data leakage, providing a viable solution for the protection of autonomous driving simulation scenario data.

Key words: autonomous driving, simulation testing, OpenSCENARIO, data security, data obfuscation

摘要： 仿真测试是验证自动驾驶系统安全性和可靠性的核心技术，针对该过程中由于场景数据明文共享使用导致的数据泄露问题，提出了一种针对场景数据的混淆保护方法。该方法覆盖了数据重编码、命名替换、顺序扰乱、标签重构、触发条件混淆及事件混淆等混淆方法，并按混淆强度划分为三级，在不影响仿真测试结果的情况下提高了场景数据的安全性。实验结果表明，混淆后的场景数据与原始仿真结果一致，且随着混淆等级的提高，数据保护程度逐渐增强。一级和二级混淆方法对仿真效率无显著影响，而三级混淆方法略微增加了仿真执行时间。整体来看，所有混淆方法均能够保持合理的仿真性能，且有效防止数据泄露，为自动驾驶仿真测试场景数据保护提供了可行的解决方案。

关键词: 自动驾驶, 仿真测试, OpenSCENARIO, 数据安全, 数据混淆