Parameter evolution for quality of service in multimedia networking

  • Authors:
  • Ji Lu;Tao Li;Xun Gong

  • Affiliations:
  • Department of Computer Science, Sichuan University, Sichuan Province, China;Department of Computer Science, Sichuan University, Sichuan Province, China;Department of Computer Science, Sichuan University, Sichuan Province, China

  • Venue:
  • PRIMA'06 Proceedings of the 9th Pacific Rim international conference on Agent Computing and Multi-Agent Systems
  • Year:
  • 2006

Quantified Score

Hi-index 0.00

Visualization

Abstract

Multi-path routing is an important mechanism for Quality of Service of multimedia communication in current networks. The key point of multi-path routing is to develop an efficient method to distribute the traffic over these multiple paths. This paper proposes a novel approach, QMP (QoS Multi-path Routing), to dynamically evolve the traffic distribution parameters using natural computation and make them adaptive to the changes of network flow and topology. First, a kind of function that reflects the changes of the network characteristics is coded using a novel natural computation technique. Then, regarding it as an antibody, QMP uses an artificial immune idea to evolve the function. Using the evolved function, some parameters in the networks can be predicted. After that, this paper proposes a traffic distribution approach that uses the parameters obtained to forward the real-time traffic, and in the meantime, adjusts the traffic parameters to adapt to the new network changes. The simulations show that this approach performs better than the classic optimal routing algorithms on feasibility and is also comparable to the optimal algorithms in terms of performance. Furthermore, QMP is better than the previous excellent multi-path and single-path routing approaches in both aspects of the delays and network resource usage.