Impact of collocated regeneration and differential delay compensation in optical transport networks

Exploiting inverse-multiplexing in optical transport networks (OTN) leads to a cost-effective network design based on legacy 10 Gbps or 40 Gbps equipment to support next-generation services, e.g., 100 Gbps Ethernet. The inclusion of multipath routing permits to better balance the load and avoid problematic capacity bottlenecks. However, with diverse-path routing the inverse-multiplexed connections may arrive at the destination node with differential delay. Such effect needs to be mitigated by buffering the early-arriving connections until synchronization is achieved and the original signal can be adequately reconstructed. One of the main design issues on inverse-multiplexed networking consists in routing the demands in such way that the buffer sizes are minimized and the load balancing maximized. In this context, centralized or distributed solutions may be adopted. The latter option considers that buffering at intermediate path nodes results in smaller buffer dimensions per node. However, existing literature in this topic has systematically neglected the limited availability of expensive optical-to-electrical and electrical-to-optical converters necessary to buffer the transit signals. OTN architectures typically dispose of such elements when signal regeneration is performed. Hence, it makes sense to promote the buffering-regeneration collocation to limit network expenditures. For this reason, this work proposes a set of integer linear programming formulations for planning OTN networks using inverse-multiplexed traffic in a scenario where the distribution of the differential delay compensation takes into account the presence of regeneration equipment. As a result, we observe that our methodologies are capable of reaching a satisfactory compromise by minimizing the regenerator count while attaining small buffer sizes.