CAJ | 학술논문

重复数据作为大量冗余数据的一种,对其有效及时地滤除能保证单芯片多处理器(CMP)数据库稳定运行.传统方法采用Fourier域变换方法,在干扰数据较多,信噪比较低时滤除效果较差,提出一种基于微分盲卷积的异构CMP数据库重复数据滤除算法,提取出满足固有模态函数的微分盲卷积分量,根据重复数据的丢失信息流特征,进行特征分解,得到重复数据的去均值量,实现对基于微分盲卷积的异构CMP数据库重复数据滤除算法改进.仿真实验表明,该算法能有效提高CPU利用率,提高系统数据处理的性能,各个节点信息处理效率较高,系统的占用率有所降低,具有数据访问和处理方面的优越性.
중복수거작위대량용여수거적일충,대기유효급시지려제능보증단심편다처리기(CMP)수거고은정운행.전통방법채용Fourier역변환방법,재간우수거교다,신조비교저시려제효과교차,제출일충기우미분맹권적적이구CMP수거고중복수거려제산법,제취출만족고유모태함수적미분맹권적분량,근거중복수거적주실신식류특정,진행특정분해,득도중복수거적거균치량,실현대기우미분맹권적적이구CMP수거고중복수거려제산법개진.방진실험표명,해산법능유효제고CPU이용솔,제고계통수거처리적성능,각개절점신식처리효솔교고,계통적점용솔유소강저,구유수거방문화처리방면적우월성.
Repetitive data as a large number of redundant data, the effective and timely filter can guarantee the single chip multi processors (CMP) database and stable operation. The traditional method using Fourier transform domain method, in-terference in more data, low SNR filtering effect is poor, proposed one kind of duplicated data filtering algorithm of differen-tial blind deconvolution of heterogeneous database based on CMP, on CMP heterogeneous database to repeat the data flow information model and differential blind deconvolution processing, adjust the resolution of frequency by using the window function of variable width method. According to the loss of information flow characteristics of duplicate data, eigenvalue de-composition, repeated data to the mean volume, the realization of the improved algorithm of repeated data filtering differen-tial blind deconvolution of heterogeneous database based on CMP. Simulation results show that the algorithm can effective-ly improve the utilization ratio of CPU. To improve the performance of the system of data processing, the information of all the nodes higher processing efficiency, system occupancy rate decreased, has advantages of data access and processing.