An Adaptive Approach for Network Traffic Load Balancing by Using One-Way Delay

As the Internet becomes increasingly popular and mature, constant increase of demands and complexities for network traffic has been an big issue. A concentration of large traffic demands to one routing node or one communication line makes a network congestion. This congestion causes increment of total network delay and the total amount of packet losses. These losses grow retransmission of packets, so the congestion goes from bad to worse and the nodes and the lines around the congestion are negatively affected. In order to avoid such congestion and to maintain link quality, various traffic engineering (TE) technologies, which utilize processing ability of routing nodes and the lines, are being researched and developed. However, existing ``offline-methods of TE'' lack reaction Sensitivity and adaptability in topology changes. On the other hand, many of ``online-methods of TE'' contain single point of failure (SPOF) issues and adjustability issues in a complex topology network. This paper proposes NREI (Network adaptive Routing algorithm for Environmental Intelligence) which is based on an one-way delay algorithm. NREI is developed from the delay time based adaptive routing algorithm which authors proposed in the past. To compare with static routing, NREI is tested in an IP network. Results of the test show that proposed routing algorithm has better adaptability in congested path avoidance and network load balancing.