CAJ | 학술논문

随着网络应用的层出不穷，互联网流量的特征是否发生了显著的变化是网络运营者和研究者关注的重要问题。为了发现互联网流量的长期演化规律，以互联网数据分析合作组织提供的五年骨干网流量数据为研究对象，分别从数据包、数据流和网络应用方面进行了分析研究；同时，基于数据包和字节序列，采用多种方法检验了流量的自相似特性。结果表明，网络流量的多个特征在演化规律上呈现出各异性的同时，网络流量的自相似特性依然存在，并且在数据包序列和字节序列上具有同一性，为互联网的协议设计、设备开发、网络管理等提供了数据支持。
수착망락응용적층출불궁，호련망류량적특정시부발생료현저적변화시망락운영자화연구자관주적중요문제。위료발현호련망류량적장기연화규률，이호련망수거분석합작조직제공적오년골간망류량수거위연구대상，분별종수거포、수거류화망락응용방면진행료분석연구；동시，기우수거포화자절서렬，채용다충방법검험료류량적자상사특성。결과표명，망락류량적다개특정재연화규률상정현출각이성적동시，망락류량적자상사특성의연존재，병차재수거포서렬화자절서렬상구유동일성，위호련망적협의설계、설비개발、망락관리등제공료수거지지。
With the emergence of new network applications,whether there have been a lot change of Internet traffic characte-ristics is an important issue for network operators and researchers.This paper aimed to achieve valuable findings on the evolu-tion of Internet traffic.The dataset was derived from over the five years on a high-speed backbone link (the CAIDA dataset). It presented a long term characteristics study of Internet traffic,including packet,flow attributes and network applications.More importantly,it also analyzed the self-similarity of traffic.To mitigate inaccuracy of single method,it computed the self-similarity parameter Hurst by three methods (e.g.variance-time plot,wavelet etc.)to verify self-similarity of traffic.Experimental re-sults show that there are so many un-neglectable differences in the traffic characteristics over the five years.However,the self-similarity of traffic also remains stable along the entire period for packet and byte counts.The self-similar parameters Hurst for byte and packet counts are usually identical.This research can present a data support for studies as the network protocol de-sign,equipment development and network management etc.