Quo Vadis, Optical Network Architecture? Towards an Optical-processing-enabled Paradigm

Among various aspects in optical network architectures, handling transit traffic at intermediate nodes represents a defining characteristic for classification. In this context, the transition from the first generation of optical-electrical-optical (O-E-O) mode to the second generation of optical-bypass marked a paradigm shift in redesigning optical transport networks towards greater network efficiency. Optical-bypass operation has then become the \textit{de facto} approach adopted by the majority of carriers in both metro and backbone networks in the last two decades and has remained basically unchanged. However, in optical-bypass network, the fact that in-transit lightpaths crossing a common intermediate node must be separated in either time, frequency or spatial domain to avoid adversarial interference appears to be a critical shortcoming as the interaction of such lightpaths in optical domain may result in efficient computing and/or signal processing operations for saving spectral resources. Inspired by the accelerated progresses in optical signal processing technologies and the integration of computing and communications, we introduce in this paper a new architectural paradigm for future optical networks and highlight how this new architecture has the potential to shatter the \textit{status quo}. Indeed, our proposal is centered on exploiting the superposition of in-transit lightpaths at intermediate nodes to generate more spectrally efficient lightpaths and how to harness this opportunity from network design perspectives. We present two case studies featuring optical aggregation and optical XOR encoding to demonstrate the merit of optical-processing-enabled operation compared to its counterpart, optical-bypass. Numerical results on realistic network typologies are provided, revealing that a spectral saving up to $30\%$ could be achieved thanks to adopting optical-processing network.