Artificial intelligence and cognitive methods in elastic optical networks for the future internet

• Grant number: 18/22878-1
• Support Opportunities: Regular Research Grants
• Start date: December 01, 2019
• End date: November 30, 2021
• Field of knowledge: Engineering - Electrical Engineering - Telecommunications
• Agreement: MCTI/MC
• Principal Investigator: Joaquim Ferreira Martins Filho
• Grantee: Joaquim Ferreira Martins Filho
• Host Institution: Centro de Tecnologia e Geociências (CTG). Universidade Federal de Pernambuco (UFPE). Recife , SP, Brazil
• City of the host institution: Recife
• Associated researchers: André Victor Silva Xavier ; Carmelo Jose Albanez Bastos Filho ; Daniel Augusto Ribeiro Chaves ; Danilo Ricardo Barbosa de Araújo ; Erick de Andrade Barboza ; Leonardo Didier Coelho ; Raul Camelo de Andrade Almeida Júnior

Abstract

The traffic generated by the Internet has grown about 100% a year since it began to be measured, and came to dominate the communication networks from the turn of the century. The networks prepared for this with the launch of a large optical infrastructure, initially oversized. Currently, the pressure for investment in telecommunications infrastructure stems mainly from the emergence and adoption of various applications on the Internet, such as video streaming, high definition television, social networks, file transfer, among others. As an example, the amount of traffic handled by wireless networks have increased from 3 exabytes in 2010 to more than 190 exabytes throughout 2018, and it is forecasted to exceed 500 exabytes by 2020. This growth in mobile data traffic has been accelerating networks second / third generation (2G / 3G) wireless to 4G / 4.5G and beyond. Looking to the future, 5G wireless technology is on the horizon, which is characterized by supporting higher data rates, excellent end-to-end performance and ubiquitous user coverage with low latency, low power consumption and low cost. The viable infrastructure solution for 5G mobile communication networks is to connect each cell to a high-capacity optical network for backhaul, which connects to a metropolitan (metro) and long-distance (backbone) optical network. The current demand for applications that require machine-to-machine communication, such as real-time monitoring, smart cities and buildings, smart grids, autonomous and interconnected cars, intelligent prostheses, etc., has contributed to the emergence of a massive number of devices in a so called Internet of Anything (IoA). This will require research and implementation efforts to increase the capacity of existing and future communication networks.Network planning and optimization are real-world problems and have a high level of complexity, as it involves the entire equipment structure, logical network architecture, and management and control mechanisms. In optical elastic networks, the automation of some of these processes by means of optimization techniques has aroused the interest of the industry, as it enables a reduction of costs and the delivery of more efficient networks. The automatic planning and management of this network paradigm presents a great challenge, since the procedures for the allocation of spectrum with heterogeneous demands, the activation, maintenance and deactivation of optical paths, the appropriate choice of the operating point of several elements in the network, as amplifiers, besides the elaboration of projects of lower capital and operational costs are complex problems and require attention.Artificial intelligence techniques for the planning and optimization of operation of high performance optical networks have been pursued by the scientific community as well as the industry, and field experiments have already been reported.The overall objective of the project is to develop solutions for data and control plans of optical networks using computational intelligence techniques, cognitive methods, optimization methods and complex networks. With this, it is intended to contribute to the development of this enabling technology to the Internet, in the sense of making it more efficient to meet the growing demands of data traffic from the internet of the future. (AU)