5G-Ready Multi-Failure Resilient and Cost-Effective Transport Networks

Data explosion and higher service level agreements (SLA) are common terminologies in modern communication networks. Recent advancements toward 5G networks add to the complexity where billions of devices, both static and mobile, are connected generating trillions of megabits of data flooding the network and where service providers scuffle daily to deliver key performance indicators: Lower latency, longer battery life, higher data rates, ultra-high reliability, and more connected devices. The optical-fiber world simultaneously witnesses a denser, fiber-rich network infrastructure which is critically challenged in fast economically developing countries by heavy construction of roads and other infrastructures that inflict repeated and unpredictable cuts to the extent that traffic is simultaneously interrupted in up to 10% of the optical links. Thus, network resiliency to fiber failures is already and will continue to be a top priority for operators. Depending on the SLA, operators choose from various resiliency options, e.g., at single network layers or across multiple ones, with dedicated protection or shared restoration, depending on decisive factors like network availability and total cost of ownership (TCO). With traditional network architectures, high resiliency comes with significantly increased TCO involving optical transport network (OTN) switches, L1 protection switching, and L1 restoration. This translates into high CAPEX and excessive space and power consumption, hampering the desirable seamless scalability with the continuously increasing demand for more capacity. This paper explores an alternative hyperscale network architecture, which is adaptable and resilient to multiple fiber failures, thus delivering committed service levels to the end users and which, at the same time, significantly reduces network cost compared to a traditional all OTN-switched network. The hyperscale architecture exploits low footprint carrier grade, data center interconnect type high-density transponders and cost-effective optical protection switching and restoration via software defined networking control. The paper also describes a routing and grooming algorithm to dimension a shared resource pool for optical restoration and a statistical simulation method for time effective execution, simulating failures over a carefully selected sub-set of failure scenarios to guarantee a high degree of network availability. Via network simulation, the paper demonstrates that multiple failure resiliency and cost effectiveness do not necessarily represent a contradiction in transport networks as long as an appropriate network architecture is chosen and that efficient network planning and optimization are applied.