ICTON 2010 We.C1.3 978-1-4244-7797-5/10/$26.00 ©2010 IEEE 1 Cognitive Optical Networks: Need, Requirements and Architecture Georgios S. Zervas, Dimitra Simeonidou High-Performance Networks Group, School of CSEE, University of Essex, Colchester, UK Tel: (+44) 1206 874233, Fax: (+44) 1206 872900, e-mail: gzerva@essex.ac.uk ABSTRACT This paper is proposing cognitive optical networks (COGNITION) as a new paradigm to address the increasing complexity of optical networks and provide an efficient autonomic as well as user-controlled network. Cognitive optical networks aim to introduce cognition on multiple planes (e.g. data plane, control plane, management plane, service plane, application layer) in order to perceive current network conditions, and then plan, decide and act on those conditions. The network can learn from these adaptations and use them to make future decisions, all while taking into account end-to-end goals. The ultimate goal of COGNITION is first to enhance optical network infrastructure by providing cognition on devices, systems and layers. The COGNITION architecture and its processes will be discussed throughout the paper. Keywords: Cognitive optical networks, cognitive cycle, self-x optical networks. 1. INTRODUCTION Today, networks comprise of several discrete network platforms, based on numerous technology waves (ATM, IP, SDH, OTN, etc.), and designed to support specific services. This increase of complexity has made it very difficult for resource and service providers to manage the different service and operational scenarios. At the same time, a new generation applications range from increasingly demanding end users services (3D TV, online gaming, etc.), to high-end distributed computing applications (e.g. cloud computing). These are increasing the burden on the available network resources and their operational management, making network management, operation and administration as well as users experience cumbersome. As a result, Future Internet should manage this ever-increasing complexity without adding more complexity and in the meanwhile minimizing the network operators’ and users’ time and effort on network operations and management. Future optical networks should be easily maintainable and their capabilities should be continuously improved and upgraded by relying as little as possible on human intervention. This paper proposes the concept of cognitive optical networks (COGNITION) that is aware of its environment, reflects on internal and external knowledge, reasons over just knowledge and acts on a decision or plan while constantly learning from the experience. The paper describes how optical networks can evolve from being aware, adaptive and finally cognitive as well as proposes a COGNITION network and node architecture that covers multiple layers and aspects of cognitive optical networking. 2. COGNITIVE OPTICAL NETWORK (COGNITION) AND NODE ARCHITECTURE 2.1 From aware to adaptive to cognitive optical networks In recent times, cognition has been identified and applied to different networking and communications systems. Originally has been introduced to wireless systems and networks such as cognitive radios and cognitive wireless networks [1]. Cognitive technology has been a requirement for self-aware networks so as to make configuration decisions based on a mission and a specific environment [2]. Another study [3] specifies the term cognition as the capability of the network to respond to conditions or events based on reasoning and prior knowledge to deliver self-adaptation and end-to-end performance delivery. Furthermore, Boscovic [4] defines cognitive networks as a network that can adapt its topology and/or operational parameters to respond to user needs while optimizing the overall network performance and enforcing operating and regulatory policies. However cognitive technologies can change the view on designing and deploying optical networks in order to provide a more simplistic, scalable and future-proof solution. Cognitive optical networks are promising to be the major step towards efficient autonomic and well as user-controlled management of the increasing complexity of optical networks. In order to deliver cognition, optical networks should be initially aware (perceive current conditions), adaptive (plan, decide and act on those conditions) and then learn from those adaptations and use them to make future decisions, while taking into account end-to-end goals. As such, the creation of aware optical networks is based on gathering partial or complete environmental information (at any or multiple layers). Such information might trigger a simple protocol decision or to maintain the awareness of network/service status. For example, a network system is aware of the services’ QoS parameters and makes resource discovery and reservation decisions to maintain or even improve QoS. Such modification should be autonomous to be considered adaptive, which can be accomplished via a protocol, mechanism or pre-programmed response. Finally, when an optical network is aware, adaptive and learns it becomes cognitive.