Updating the OSPF routing protocol for communication networks by optimal decision-making over the k-shortest path algorithm

Internet routing protocols such as Routing Information Protocol (RIP) pre-compute all the shortest paths by Dijkstra's algorithm (shortest path first, SPF) based on the number of hops between one node and another. Every time any communication is intended, RIP looks-up for the optimal choice in a routing table. This is a high speed method in the decision-making process but not necessary fast for data traffic as it does not take into account any real-time measure of route congestion. Open Shortest Path First (OSPF) presents a dynamic version of this problem by computing the shortest paths taking into account network features such as bandwidth, delay and load. OSPF thereby maintains link-state databases updated at near real-time at every router. Although OSPF protocol is widely used, the Enhanced Interior Gateway Routing Protocol (EIGRP) presents another option for taking the optimal routing. At EIGRP, each node independently runs an algorithm to determine the shortest path (Dijkstra's algorithm) from itself to every other node in the network. In addition to the routing table, EIGRP uses neighbour information (neighbour nodes on node j are those having directly connection nodej) and a table of favourite (commonly used) routes. In this context, a multi-criteria decision-making (MCDM) approach may represent an alternative perspective for the automatic selection of an optimal path among the k-shortest paths. MCDM methods effectively support a plethora of decision problems, their crucial role being widely acknowledged (Kumar et al., 2017). With respect to routing protocols in communication networks, a final decision on selecting the optimal path depends on various evaluation criteria. These sometimes are mutually dependent and conflicting with each other. This is the case of criteria such as traffic density, path length, data type, and key performance indicators. Under such criteria, the objective is to ultimately select one path among the k-shortest paths. MCDM methods have the ability of going towards the solution that represents a best trade-off for the selected network intends and satisfies their multiple aspects regarding their mutual importance. MCDM are capable of managing both qualitative and quantitative aspects when it is required an evaluation concerning a set of alternatives (Mulliner et al., 2016). Several MCDM methods have been proposed in the existing literature, each one being characterised by specific procedures and objectives. However, common points for MCDM methods are that they can be mainly aimed at: selecting the best option among various alternatives, ranking alternatives to establish their weights and/or to draw up a list of priorities (Vargas et al., 2017), and clustering alternatives into different groups on the basis of their common features (Certa et al., 2017). Among MCDM methods, the fuzzy evolution of the Technique for Order Preference by Similarity to Ideal Solutions (TOPSIS), that is the FTOPSIS (Chen, 2000) method, is herein proposed to get the ranking of the possible k-paths related to a communication network according to the evaluation of suitable criteria and to simultaneously take into account uncertainty affecting input data as it is, for instance, the traffic density.