Improving OSPF Protocol based Latency : A new algorithm based on Dijkstra by using OSPF existing Metrics in SDN networks

Resumo

SDN (software defined networking)-based networks may be defined as a new generation of networks using virtual layers and switches and central controller which try to handle a few controlling and managerial tasks of switches and rotors of networks in upper layers on a software basis. In fact, it reduced dependence to hardware and increases software capabilities. These networks face challenges such as quality of relationship between controller and devices existing in the network and delay in network that is subject of this thesis. According to the applied researches, one of offered solutions for reduction of delay is using path finding algorithms. Rotors’ task is transfer of information. Algorithms must be implemented on these rotors to choose the best path for data transfer in the network. Path finding table is used in rotor. According to the data available in path finding table, the best path is found. Each path finder must have complete information of network’s communication infrastructure and calculate and identify all other paths of communications between them and their costs. Later, data collection forms the data structure related to network infrastructure graph. In these conditions, to find the best path between path finders, the shortest path algorithms (SPT) are used such as Dijkstra. Since rotors receive the sent update massages due to network changes, path finding table amends itself and identifies the new probability path. Selection of best path is made by massages metric. Upon processor fastening and hardware cheapening, a standard protocol in the name of OSPF was presented that manufactured by CISCO, particularly in a network that its equipment are not necessarily made by CISCO is based on Dijkstra and uses cost and band broadness as metric, transfer the data related to connected network and rotors connected to network between adjacent rotors and records all of its information in the table. Later, Dijkstra's algorithm is implemented and the best paths led to different destinations are inserted in the path finding table. The objective of this study was presenting an algorithm based on Dijkstra there in addition to cost Metric, another metric to be used that highly reduces traffic in the network and improves delay time in the network.