Efficient policing for screen mirroring traffic

Orestis-Stavros Loizides,Polychronis Koutsakis,Mohammed N Smadi,David Murray,Ghada Badawy,Marios Kastrinakis

doi:10.1186/s13673-018-0127-8

Abstract

The greediness of multimedia applications in terms of their bandwidth demands calls for new and efficient network traffic control mechanisms, especially in wireless networks where the bandwidth is limited. In an enterprise-like environment, an additional burden is expected to be added to the network by screen mirroring traffic. Smart mobile devices are displacing personal computers in many daily applications but at the same time users still need to use a large display, keyboard and mouse. Hence, the transmission of low-latency, high fidelity video over a Wi-Fi link can lead to significant unfairness among users in terms of the bandwidth that is available to them, if this wireless video traffic is not accurately policed. In this work, we focus on the problem of policing screen mirroring traffic. We evaluate various classic and new traffic policing mechanisms, and we propose a new mechanism which is shown to clearly outperform all other mechanisms, including the widely used token bucket policer.

Highlights

Smart mobile devices such as smartphones and tablets are becoming more powerful every day with the advancement of mobile computing chips, while at the same time major software and operating system companies develop their products in a single code base suitable for many platforms [1]
Screen mirroring traffic can be classified and treated in a manner similar to the one described in [37], i.e., packets are classified according to an identifier which is left on the data packets by an application and is used for identifying the generating application, marking the class of the data packets and correspondingly policing the data packets according to the classification marker, reducing the traffic rate to force the traffic to follow the allocation limit
Frame Size Aware Token Bucket (FSA-token bucket (TB)) was shown to outperform all mechanisms for every video of Dataset 1 used in our study, followed by Jaccard index-infused Markovian clustering policer (JIMC-P)

Summary

Introduction

Smart mobile devices such as smartphones and tablets are becoming more powerful every day with the advancement of mobile computing chips, while at the same time major software and operating system companies develop their products in a single code base suitable for many platforms [1]. In our work in [26] we proposed a new traffic policing mechanism, the Frame Size Aware Token Bucket (FSA-TB), which was shown to outperform all of the widely used mechanisms in the literature.

Results

Conclusion