On Network Design and Planning 2.0 for Optical-computing-enabled Networks
Dao Thanh Hai, Isaac Woungang
TL;DR
The paper addresses the growing demand for scalable optical networks by proposing optical-computing-enabled networks, where optical nodes perform in-network computations through controlled lightpath interference to boost capacity and efficiency. It introduces two enabling operations—optical aggregation/de-aggregation and optical XOR encoding/decoding—and formulates a routing-wavelength-network coding assignment problem, solved via an ILP and a scalable heuristic. The study demonstrates potential spectral efficiency gains in both small-scale and realistic COST239/NSFNET topologies, highlighting the added complexity and the need for new design algorithms. This work lays the groundwork for optical network design and planning 2.0, promising improved capital and operational efficiency as enabling photonic technologies mature.
Abstract
In accommodating the continued explosive growth in Internet traffic, optical core networks have been evolving accordingly thanks to numerous technological and architectural innovations. From an architectural perspective, the adoption of optical-bypass networking in the last two decades has resulted in substantial cost savings, owning to the elimination of massive optical-electrical optical interfaces. In optical-bypass framework, the basic functions of optical nodes include adding (dropping) and cross-connecting transitional lightpaths. Moreover, in the process of cross-connecting transiting lightpaths through an intermediate node, these lightpaths must be separated from each other in either time, frequency or spatial domain, to avoid unwanted interference which deems to deteriorate the signal qualities. In light of recently enormous advances in photonic signal processing / computing technologies enabling the precisely controlled interference of optical channels for various computing functions, we propose a new architectural paradigm for future optical networks, namely, optical-computing-enabled networks. Our proposal is defined by the added capability of optical nodes permitting the superposition of transitional lightpaths for computing purposes to achieve greater capacity efficiency. Specifically, we present two illustrative examples highlighting the potential benefits of bringing about in-network optical computing functions which are relied on optical aggregation and optical XOR gate. The new optical computing capabilities armed at optical nodes therefore call for a radical change in formulating networking problems and designing accompanying algorithms, which are collectively referred to as optical network design and planning 2.0 so that the capital and operational efficiency could be fully unlocked.