1 AIT Asian Institute of Technology

Measurement and analysis of traffic in an IP-based network : a case study of Ministry of Defense, Thailand

AuthorBooncherd Nakornrum
Call NumberAIT RSPR no.ICT-09-08
Subject(s)Telecommunication--Thailand--Traffic

NoteA research study submitted in partial fulfillment of the requirements for the degree of Master of Engineering in Information and Communications Technologies
PublisherAsian Institute of Technology
AbstractMeasurement and analysis of real traffic is decisive to obtain traffic characteristics which are useful for understanding the network situations and user behaviors. Understanding the network situations and user behaviors are important for capacity dimensioning and network planning. The real traffic in an IP-based network of Ministry of Defense (MOD), Thailand is studied in this research. Three internal links of MOD network; SDH (Synchronous Digital Hierarchy), IPStar (IPS), and Frame relay (FRR) links are measured by a measurement tool named Paessler Router Traffic Grapher (PRTG) to obtain the bandwidth usage of both inbound and outbound traffic rate from May 1, 2009 to July 31, 2009. The bandwidth usages of all three links are then aggregated as the total traffic of the external link. To find out their characteristics and user behaviors, these statistical traffic data are analyzed in various aspects. Traffic profiles over daily, weekly, and throughout measurement duration are studied to understand the traffic patterns and user behaviors. Busy hour concept is examined to find out the values and periods of busy hour in different methods and the most appropriate method for this traffic is also proposed. Distributions of bandwidth usage are revealed by histogram. The self-similarity property is studied on busy hour traffic by variance time plot technique. The long-term of traffic rate on IPS link is studied to find out its movements. The traffic volume is also studied to obtain its patterns and trends. The analyzed results show that download traffic rate is much higher than upload traffic rate on all links and the total traffic rate in the system is mostly generated by SDH link, backbone of MOD network. User behaviors on almost all links are associated with working days and office time except on IPS link which user behaviors are not different even though on workdays or weekends. However, user behaviors show no pattern on any link since the traffic profiles of all links have day-to-day variations. The most appropriate approach for determining busy hour traffic on this network is average daily peak hour (ADPH)since it gives the most precise value and the traffic profile is quite unstable. The values of time consistent busy hour (TCBH) and fixed daily measurement hour (FDMH) are quite similar and lower than ADPH. The appropriated period that offers the greatest average traffic rate for TCBH is 10am-11amwhile the appropriated period for FDMH is 12:00-1pm. The distribution of traffic rate at 1-minute resolution throughout the measurement duration shows that capacities of all links are not critical at that time since the maximum rate do not reach the link capacity and mostly traffic rates (90 percentile) are much less than the link capacity. The analysis of variance time plot shows that download rates in all daily busy hours on aggregated link throughout the measurement duration are self-similar since all Hurst parameters (H) are greater than 0.5, almost all Hare greater than 0.7 and especially the mean value of H is 0.814.The variance time plot is used to backward estimate the variance of several finer time scales than the original data for both self-similar and Poisson processes. The result shows that the variances of 1-second time scale are approximately 5 and 50 times to the variance of 1-minute time scale according to self-similar and Poisson processes respectively. This indicates that the secondly rates in daily busy hour deviate not far from the mean. The long-term trend of the traffic rate on IPS link could not be obtained since the traffic rate has month-to-month variation. There is also no pattern and trend of traffic volume occurred on any link since the traffic volume is associated with traffic rate that has day-to-day variation.
Year2009
TypeResearch Study Project Report (RSPR)
SchoolSchool of Engineering and Technology (SET)
DepartmentDepartment of Information and Communications Technologies (DICT)
Academic Program/FoSInformation and Communication Technology (ICT)
Chairperson(s)Teerapat Sanguankotchakorn;Erke, Tapio J.;
Examination Committee(s)Rajatheva, R. M. A. P.;
Scholarship Donor(s)Royal Thai Government;
DegreeResearch Studies Project Report (M.Eng.) - Asian Institute of Technology, 2009


Usage Metrics
View Detail0
Read PDF0
Download PDF0