On Minimizing Cost of Reliable Network Topological Design using a Practical DP Approach

Abstract

This paper addresses an NP-hard problem called NTD-CR that using dynamic programming (DP) scheme. This problem referred to as Network Topology Design with minimum Cost subject to a Reliability constraint. It designs a minimal-cost communication network topology that satisfies a pre-defined (s, t) terminal reliability constraint. Our DP approach, called DPCR-L, mainly select the set of possible links to be deleted from the original network to generate an optimal Network Topology. The NTD-CR design problem aims to find a Network Topology that has minimal cost with the required reliability for the network. Five link-ordering criteria are proposed to improve the performance of DPCR-L. Each greedy heuristic order allows DPCR-L to enumerate the selected deleted links, which improves the time complexity while producing near optimal topology. Extensive Simulations based on different benchmark networks of various sizes are used to compare DPCR-L with existing state-of-the-art techniques and show the merits of using the ordering methods, and the effectiveness of our algorithm. Our simulations show that DPCR-L produces 93% optimal results. Interestingly, of the non-optimal results, DPCR-L produces a network with reliability no worse than 5.38% off optimal, and most of the non-optimal results have a lower cost than that for optimal up to 0.17%. Typically, for the most of the network topologies, DPCR-L generates NT with the same or better 2-terminal reliability measure and speeds up its running time up to 31.71%. Furthermore, simulation results on large size networks show that DPCR-L speeds up the process with up to (47.28%) compared to the recent existing approach. Finally, the results present the applicability of DPCR-L on networks containing a large number of links and demonstrated better performance and computationally more efficient than other existing algorithms.