Load-balanced routing with selective even traffic splitting ∗

Abstract

This paper proposes an even-split Smart-OSPF (S-OSPF) scheme to reduce network congestion more than the conventional non-split S-OSPF and to distribute traffic more easily than the conventional split S-OSPF. In split S-OSPF, source edge nodes distribute traffic unevenly to their neighbor nodes, but the implementation becomes involved to split traffic with different distribution. In non-split S-OSPF, source edge nodes transmit traffic to only one neighbor so that network congestion can be minimized, where non-split S-OSPF distributes traffic more simply than split S-OSPF. In the proposed scheme, source edge nodes transmit traffic evenly to selected neighbor nodes to minimize network congestion. The optimization problem to select a suitable set of neighbor nodes for even traffic distribution raised by the proposed scheme is formulated as an Integer Linear Programming (ILP) problem. The difficulty of solving the ILP problem in a practical time leads us to introduce a heuristic algorithm. The performances of our developed heuristic algorithm are evaluated via simulation developed in terms of network size. Numerical results show that even-split S-OSPF offers better routing performance than non-split S-OSPF for small-size networks and matches that of split S-OSPF for large-size networks.