On the use of prediction in Passive Optical LANs for healthcare latency-stringent applications

Abstract

Passive Optical Networks (PONs) have been established as the most successful, viable and economic solution for the access part of an optical network. During the last two decades, a great proportion of the academic research in networking has focused on improving the bandwidth allocation algorithms for PONs so as to provide better user services. Recently, PONs have been suggested as the means for providing ultra-low latency for QoS stringent applications, such as the Tactile Internet, for healthcare applications in a hospital environment. Despite their potential, little attention has been given so far to prediction methods for allocating bandwidth in advance so as to decrease end-to-end latency in such environments. This paper aims to address this gap, by proposing a novel framework for Passive Optical LANs. By utilizing a prediction scheme and a Dynamic Bandwidth Allocation (DBA) algorithm that allocates bandwidth based on network feedback information, the proposes approach is shown to significantly contribute to reduction of latency.