USING H-PARAMETER TO REDUCE SELF-SIMILARITY INTCP NETWORKS

ABSTRACT In the twenty first century, many people and machines are talking to each other using several types of communication systems. The greater the number of communication systems, the more varied and complex, the means of having them interconnected. Telecommunication operators need to provide quality of service (QoS) to their subscribers and to achieve that quality of service, operators need to focus on the control and the management of traffic. However, it has been observed that the traffic in a network communication system exhibits a problem called “Self-Similarity”. This is a phenomenon, which occurs when the system is subject to larger queuing delays, higher drop rates and extended period of congestion. Therefore,the objectives of this study is to propose a method of eliminating self-similarity in the network communication system. The procedure used to solve this problem of Self- Similarity is as follows: the H-parameter was used to estimate the degree of self-similarity in the network communication system. To compute this H-parameter, the VTP (Variance Time Plot) of packets sniffed using Wire shark was analysed. Then, the principle of RED (Random Earlier Detection) was used to write an algorithm to reduce the level of self-similarity in TCP traffic. The results showed a reduction in the value of the H-parameter from 0.75 to 0.5 using RED. H-parameter value of 0.5 means that the network is good as the degree of selfsimilarity has been reduced. This implies that the RED application has improved upon the self-similarity in the network communication system Keywords: QoS, self-similarity, H-parameter, variance time plot, random earlier detection,TCP