Performance evaluation of resources management in WebRTC for a scalable communication

Abstract

Web Real-Time Communication (WebRTC) offers peer-to-peer communications without any plug-ins. However, WebRTC cannot provide scalability because of its method that depends on a single server or due to the resource limitations and network topology in the architectural of the WebRTC. This paper aims to design a real environment using MATLAB simulation tools to specify the limitations of resources in WebRTC for bi-directional video conferencing, such as CPU performance, bandwidth consumption and Quality of Experience (QoE) using different topologies such as mesh, star and hybrid (a combination of unidirectional/star & bi-directional/mesh). Moreover, several CPU cores like i3, i5, i7, Xeon, i9 and Xeon Phi, as well as bandwidths: 0.5, 1, 5, 10, 30, 50, 100, 500 and 1000 (Mb/s) were considered to achieve and expand the scalability. In this implementation, the factors of real-time implementation were used. Thus, the utilized measurements were already validated while MATLAB presents coefficient with 95% confidence bound. Additionally, this paper highlights the obstructions are preventing scalability in WebRTC using a centralized server. This illustration is beneficial for interested developers who intend to use WebRTC duplex video conferencing among undefined users and different topologies. Furthermore, our simulation-based’ performance evaluation shows the efficiency of the hybrid topology in decreasing the bandwidth overhead and CPU load in WebRTC.