Cybernetics and Information Technologies
Volume 2, No 2. Sofia, 2002, Bulgarian Academy of Sciences

Soft Computing Agents for MPLS Networks

Georgi Kirov, Dimitar Lakov
Institute of Computer and Communication Systems, 1113 Sofia

The paper underlines the main advantages and drawbacks of IP routing. Some possibilities for resolving of routing problems are analyzed. A motivation for the use of Soft Computing Technologies (SCT) and Multiple Protocol Label Switching (MPLS) protocol for solving network routing problems is shown. The basic concept of the MPLS networks are described. Some definitions of Soft Computing Agents (SCAs) are briefly considered. It is demonstrated that the combination of the advantages of Soft Computing Technology with Software Agents paradigm is universal optimal resource allocation techniques. Optimal and Information SCA for MPLS network routing in accordance with users' preferences are developed. The ability of the Optimal SCA for processing and analysis of large amount of network information and generating of Label Switch Paths (LSP) is demonstrated. A special attention is paid to Information SCA that improve and support Optimal SCA in choosing the best path in respect to all available routing data.

Keywords: IP routing, MPLS protocol, soft computing agents