A self-organization scheme for adaptive routing | |
| Author | Aekavute Sujarae |
| Call Number | AIT Diss. no.CS-97-02 |
| Subject(s) | Self-organizing systems Neural networks (Computer Science) |
| Note | A dissertation submitted in partial fulfillment of the requirements for the degree of Doctoral of Engineering |
| Publisher | Asian Institute of Technology |
| Abstract | Upon the increasing diversity of wide-area network applications, offering varying type of service (TOS) such as low delay, high throughput. has become a crucial factor for network design. Since each application demands different performance measures from the network, the routing protocol must be able to determine an appropriate route for each TOS. In order to achieve this, the routing protocol must be capable of distinguishing traffic types and serving them according to their demands and requirements. However, routing algorithms commonly used in wide-area networks do not distinguish traffic either by type of packets or by state of the network. Therefore, the routing decisions are primarily based on a single performance criterion. Thus, a routing model that takes into account the varying performance of the network under a variety of requirements is needed in order to achieve effective routing decisions. In this dissertation, a self-organizing based routing model which incorporates multi-criteria based routing decisions under least-cost path strategy through Winner-Take-All (WTA) self-organizing network and Bellman-Ford algorithm is developed and investigated. To come up with such an approach, essential parameters which influence WTA network performance in terms of convergence rate, speed of retrieval, and time and space complexity are quantitatively studied and a simple approach in constructing binary input vectors for WTA is developed. The effectiveness of the proposed routing scheme is studied and demonstrated through computer simulations with different network environment. Experiments on single class application show that the scheme yields competitive performance against other routing strategies, whereas, significant improvements on both throughput and delay are shown under multiple class application. |
| Year | 1997 |
| Type | Dissertation |
| School | School of Advanced Technologies (SAT) |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Computer Science (CS) |
| Chairperson(s) | Sadananda, Ramakoti |
| Examination Committee(s) | Yulu, Qi ;Johri, Lalit M.;Itoh, Kiyoshi; |
| Degree | Thesis (Ph.D.) - Asian Institute of Technology, 1997 |