并行多路径传输过程数据相似性检测仿真  

Simulation of Data Similarity Detection in Parallel Multipath Transmission

作　　者：赵敏[1] ZHAO Min(Xidian University,Xi'an Shanxi 710071,China)

机构地区：[1]西安电子科技大学,陕西西安710071

出　　处：《计算机仿真》2021年第9期181-184,375,共5页Computer Simulation

摘　　要：为提高数据传输过程的挖掘效率和原始数据的质量,提出了一种并行多路径传输过程数据相似性检测方法。首先,通过位码代替法计算并行传输过程数据架构的相似度,然后利用优化编译剔除数值化后的公共子代表式,并将其转化成汇编数据。这一过程中,由于传输偏移地址、函数地址、部分跳转指令与立即数存在大量噪声,因此需通过反汇编过程对噪声滤波进行归一化处理,从而构成汇编数据集合,再利用下三角矩阵将传输数据之间的相似度对比转换为储存矩阵之间的相似度对比,并计算矩阵内架构相似度与语义相似度,实现数据相似度检测。仿真结果证明,相比于传统方法,上述方法的数据召回率和检测损耗较小,且相似性检测误差较小,有效性更高。In order to improve the mining efficiency in data transmission and the quality of original data, a data similarity detection method for parallel multipath transmission was proposed.Firstly, the similarity of data architecture in parallel transmission was calculated by bit code substitution, and then the common subexpression after numerical conversion was eliminated by the optimized compilation, and it was transformed into the compiled data.In this process, the transmission shift address, function address, some jump instructions and immediate data, contained a lot of noise, so it was necessary to normalize the noise filtering by disassembly process, so as to constitute a set of compiled data.Moreover, the lower triangular matrix was used to transfer the similarity between transmission data into the similarity between storage matrices.Meanwhile, the architecture similarity and semantic similarity in matrix were calculated to realize data similarty detection.Following conclusions can be drawn from simulation results: the data recall rate and detection loss of the proposed method are lower than the traditional method.Meanwhile, the detection error is smaller and the effectiveness is higher.

关 键 词：并行多路径 数据传输 相似性检测 数值 反汇编 噪声 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]