Tactile-capable optical cloud distribution networks -

Abstract

Over the past decade, Cloud computing has become the most beneficiary paradigm to provide optimized and efficient computing resources for modern Internet applications. However, cloud networks can be a burden for latency-sensitive applications, such as Tactile Internet, which can't afford the high delay due to communication with remote cloud servers, which typically reside in the core network. In this work, we propose OCLDN, an agile, programmable and scalable Optical CLoud Distribution Network, which exploits content distribution networks, to deliver killer cloud and tactile services to end-users at very high-speeds, ultra-low latency, and low cost. With OCLDN, a software defined networking (SDN)-based mini-cloud data center is installed at the central office of a next-generation passive optical network, which includes a cloud-based tactile steering server, and supports all types of cloud applications. Meeting the stringent quality-of-service (QoS) requirements for these applications requires a fast, adaptive and effective bandwidth allocation mechanism. We mathematically formulate this problem using Mixed Integer Linear Programming (MILP), and then introduce a new Dynamic Wavelength and Bandwidth Allocation (DWBA) scheme to overcome the limitations of the MILP model. The proposed DWBA scheme enables the protection of tactile services by dedicating it upstream wavelength(s), and enables inter-channel statistical multiplexing so as to maximize the network throughput without impairing the QoS requirements for all types of services. It also employs a modified version of the Water-Filling technique and advanced intra-ONU scheduling. Extensive simulations are performed; results demonstrate the ability of the proposed DWBA to outperform existing schemes, and highlight its effectiveness in meeting the QoS demands for all types of services.