CAJ | 학술논문

在无线传感器网络下，进行分布式系统下数据流时间序列的离群检测研究，对实际生活中的火灾检测、欺诈、入侵检测和金融分析等诸多领域都有非常重要的作用。先对序列进行标准化，再利用动态时间弯曲路径进行序列间离群检测，较大地提高了检测的精度。同时针对DTW的计算复杂度瓶颈问题，加入了提前终止的思想，并利用多级判断来进一步提速。通过NS2仿真实验，验证了所提出的分布式系统下离群序列的加速检测方法既节省了能量消耗，又保持了较高的检测准确率和速度。
재무선전감기망락하，진행분포식계통하수거류시간서렬적리군검측연구，대실제생활중적화재검측、기사、입침검측화금융분석등제다영역도유비상중요적작용。선대서렬진행표준화，재이용동태시간만곡로경진행서렬간리군검측，교대지제고료검측적정도。동시침대DTW적계산복잡도병경문제，가입료제전종지적사상，병이용다급판단래진일보제속。통과NS2방진실험，험증료소제출적분포식계통하리군서렬적가속검측방법기절성료능량소모，우보지료교고적검측준학솔화속도。
In wireless sensor network,abnormal time series sequence detection under distributed system plays a very important role in many areas in real life,such as fire monitoring,fraud,intrusion detection and financial analysis. This paper firstly does Z-normalization for the sequences,then uses dynamic time warping to detect outliers,which largely improves the accuracy of detection. Meanwhile,to solve the bottleneck of computational complexity of DTW, the early abandoning and multistage judgment is added to accelerate the DTW algorithm. The algorithm is verified by NS2 simulator. The results demonstrate that this algorithm not only decreases the energy consumption,but also keeps a high precision and time efficiency.