Cost feasibility analysis of translucent optical networks with shared wavelength converters

Translucent optical networks have emerged as potential yet feasible candidates to bridge the gap between the opaque and transparent network architectures. By allowing electrical 3R signal regeneration only at selected points in the network, translucent architectures represent a cost-effective, power-efficient solution. Concurrently, forecasts predicting highly dynamic traffic patterns make it crucial for next-generation transport networks to engage highly agile technologies that include sub-wavelength switching (SWS). In translucent SWS networks, contention resolution is achieved through the still technologically immature all-optical wavelength converters (WCs). Since WCs are expected to be expensive, power-consuming devices, there has been significant research effort on devising WC-sharing architectures, which aim at minimizing the number of these devices in the network. WC sharing, however, requires complex switching fabrics that involve a much higher number of optical gates and stronger degradation due to physical layer impairments (more electrical 3R regenerators). It is clear, then, that the technological interest in WC-sharing architectures mainly depends on the cost trade-offs existing between these three components. For this reason, in this work, we carry out a comprehensive cost feasibility analysis of translucent networks based on asynchronous WC-sharing packet switches. After modeling a set of translucent WC-sharing switching fabrics, we assess their performance in an isolated node scenario in terms of the number of WCs and optical gates required. Given the results obtained, we select the shared-per-node (SPN) architecture to compare its hardware requirements with those of a network based on dedicated WC nodes (i.e., one WC per wavelength and input port). To this end, an iterative simulation algorithm is used to dimension translucent SWS networks considering a broad range of continental-scale topologies. The results are first analyzed using relative cost- values, and finally the viability/feasibility of WC-sharing schemes is discussed considering state-of-the-art technology. Our main conclusion is that, for SPN-based architectures to become cost effective, the cost of WCs has to be at least two orders of magnitude higher than that of the optical gate and similar to or lower than that of the electrical 3R regenerator.