University: Akita University
Name of sponsoring professor: Ken-ichi Yukimatsu
Department: Computer Science & Engineering Department

From the year 2006, we have been using OPNET in our labs for the research titled, "A study on next-generation photonic networks with buffer-less photonic packet routers and traffic engineering mechanism". Its research theme is to investigate the feasibility of high speed and high quality, brand-new networks utilizing photonic-based principle and technologies

We chose this research theme as we noted the current situation being the following: current IP network is still on its way of rapid and dramatic evolution, and it is finding revolutionary application in an inexperienced manner. That kind of application, without saying, is supported undoubtedly by the realization of sufficiently high speed of the back-bone communication hardware. The evolving network application, however, still requires much more of the hardware capability. In order to fulfill the heavy demand of the current development, the network configuration itself is required to be changed, or to be evolved. The center conception or the focus of our research resides how to solve the demanding problem above by cleverly utilizing mechanism, so called as traffic engineering. The idea has not yet been proposed anywhere and a conceptual investigation is very much useful.

In our research activities in 2006 to 2007, we worked on the buffer-less routing networks that uses rerouting control, and traffic engineering scheme based upon application flows for QoS enhancement.

In the first research, we studied novel network architecture based on buffer-less packet routing, and discussed the feasibility of the architecture from the view point of packet traffic throughput characteristics. In order to reduce the packet loss at buffer-less packet routers due to packet contention on outgoing links, rerouting control method was expected to be effective. Two rerouting control schemes using expending routing tables in the existing IP packet network and node coordinated in grid-type networks were studied and their traffic characteristics were examined by computer simulation.

In the second research, we studied a traffic control scheme that uses application flows as identified by application information in the packets being transferred. Our proposed scheme increased flexibility in terms of IP packet routing. First, it was confirmed that individual application flows could be routed through their own minimum-cost paths. Cost was determined, for each flow, from one or more QoS parameters, for example, delay and availability. Second, it facilitated traffic engineering, namely the efficient utilization of network resources. In this situation, one source-destination traffic demand was satisfied by setting one or more flows as needed for traffic grooming. Numerical results showed that the proposal allowed the network to accommodate more traffic even if the cost-routed traffic created a bottleneck link.

As stated above, the network of our conception does not exist. Therefore, to estimate the validity of the idea proposed, a competent network simulator is indispensable. This is why we have been using the "OPNET" as a basic research tool. Two principal purposes are as follows;

1. To realize virtually the network with our proposal, namely the optimal network architecture with the photonic "Buffer-less Packet Router", the use of OPNET is essential.

2. To evaluate the effectiveness of the MPLS-based traffic engineering scheme on various network topologies, the OPNET is considered to be the most suitable simulation tool.

To propose effective re-routing methods as well as the topology is also important in the research. In particular, we are evaluating various control methods, using the OPNET MPLS module with the Modeler.