Latency-aware service provisioning in survivable multilayer IP-over-elastic optical networks to support multi-class of service transmission

Abstract

To meet the requirements of the emerging applications, future networks must transmit multiple class of services traffic with diverse quality of service (QoS) requirements. In this paper, we formulate an integer linear programming (ILP) to design multilayer internet protocol (IP) over elastic optical networks (EONs), IP-over-EONs to support multi-class services with different latency and availability requirements. The objective of the ILP is to minimize capital expenditure (CAPEX) and spectrum usage while satisfying all QoS constraints in terms of latency and survivability mechanism. Furthermore, a heuristic algorithm is proposed to solve this problem in large-scale networks. We compare the results of the ILP formulation and the proposed heuristic algorithm in a 5-node network, where our results reveal that the performance gaps between the ILP and heuristic algorithm are on average 2.4% and 9.8% in terms of CAPEX and occupied spectrum, respectively, while the heuristic algorithm has a lower run time. Moreover, we quantitatively analyze the effects of end-to-end latency on the CAPEX of multilayer IP-over-EONs. Our results reveal that serving latency-sensitive demands increases the CAPEX of the network. This is because all-optical connections (i.e., single-hop IP connections) must be established to meet the delay constraints, which in turn leads to a higher CAPEX and occupied spectrum. Furthermore, we compare the effects of using different types of transponders in three practical topologies and evaluate the impacts of the length of the topology links on CAPEX and spectrum usage. Finally, we propose the best transponder types that could be used for each topology based on the gathered results.