摘要
为研究5G业务发包模型对时延的影响,针对工业场景中常见的心跳包业务,测量无干扰的室内小基站5G网络下心跳业务在真实物理链路中的环回时延,分析心跳包时延时序包括平稳性在内的统计特性;利用OCSVM算法进行异常检测,分别提取其正常态和异常态的特征参数,得到一种时延曲线的相似度判别方法并验证了其可行性,为平稳时序的相似性研究提供一种新的思路;利用提出的相似度方法检验不同发包模型下的时延相似度,检验结果可为5G工业场景下发包模型的调整、布网等提供参考。
In order to study the influence of 5G packet sending model on the delay,the loop delay of heartbeat packet in real physical link in the 5G network of small indoor base station without interference is measured for heartbeat packet service common in industrial scenarios,and the statistical characteristics of heartbeat packet delay sequence including stationarity are analyzed in this paper.OCSVM algorithm is used for burst detection, and the characteristic parameters of normal and abnormal states are extracted respectively.A similarity discrimination method of time delay curve is obtained and its feasibility is verified,which provided a new idea for the similarity study of stationary time series.The proposed similarity method is used to verify the delay similarity of different packet sending models.
出处
《工业控制计算机》
2022年第4期56-58,60,共4页
Industrial Control Computer
关键词
网络时延
赫斯特参数估计
异常检测
相似性
network delay
hurst parameter estimation
anomaly detection
similarity
