首页|期刊导航|网络安全与数据治理|评价加壳软件识别工具稳定性的新方法

评价加壳软件识别工具稳定性的新方法OA

A new method for evaluating stability of packed software identification tools

中文摘要英文摘要

针对现有加壳软件识别工具存在的体系化评价标准缺失,特别是稳定性量化分析不足的问题,提出了一种根据现有的准确性指标来评价识别工具稳定性的 LK 方法:通过构建检测准确性与样本变化度的回归模型,创新性提出抗干扰度指标用于量化识别工具的稳定性特征.研究采用对比实验范式,对主流工具进行横向测试:首先通过干扰样本集获取检测性能变化数据,继而基于回归模型计算抗干扰度参数,最终构建涵盖准确性、稳定性等多方面指标的综合评价方案.实证结果表明,研究提出的抗干扰度指标能有效揭示并量化各识别工具的识别准确性、稳定性与样本变化程度间的关联特征.该研究不仅为优化加壳软件检测流程提供了理论支撑,通过结合稳定性的全新分析评价与现有的准确性分析,更为新型加壳方案的识别算法及对抗性测试优化提供了可量化的效果验证框架.

In view of the problem that the current packed software identification tools are missing comprehensive evaluation standard,especial-ly lack of quantified analysis of stability of the identification tools,this paper proposes a new method called LK method,which evaluates the stability of the identification tools based on their accuracy.The method constructs a linear regression model on the degree of variation on the samples and accuracy,innovatively proposes a measurable index called"Disturbance Resistance Index"or DRI,to quantify the stability of the identification tools.The research takes a comparative model to test several common identification tools:firstly,it obtains the performance vari-ation data by using interfered samples,then utilizes the data to compute the DRI based on the linear regression model,and finally constructs a multi-dimensional evaluation system covering accuracy,stability and so on.The analysis on the result shows that the DRI index proposed by the research can effectively disclose and quantify the associated features between the variation on samples and the accuracy and stability of the identification tools.This method not only provides theoretical support for optimizing the detection process of packed software,but also,by com-bining a brand-new stability analysis and evaluation with current accuracy analysis,offers a quantifiable effect verification framework for the i-dentification algorithms of new packaging schemes and the optimization of adversarial testing.

李可;宋栋;俞楚翘

华北计算机系统工程研究所,北京 100083华北计算机系统工程研究所,北京 100083中国人民解放军 93216 部队,北京 100085

信息技术与安全科学

加壳软件检测算法评价方法

packed softwaredetection algorithmevaluation method

《网络安全与数据治理》 2026 (9)

60-65,6

10.19358/j.issn.2097-1788.2026.09.008

评论