Adaptive Virtual Reality Streaming: A Case for TCP

Virtual reality (VR) is one of the applications with the most strict requirements in the performance of next generation networks, since it requires both high throughput, low delay, and packet loss. As the performance of networks, and the level of congestion varies over time, a need for adaptation in the stream’s data rate, in order to maintain reasonable packet loss, while using the available bandwidth, emerges. Motivated by this, in this contribution, we present an adaptation algorithm for VR applications, that exploits fuzzy logic, and transmission control protocol (TCP) transport in order to maintain the optimal data rate of the VR stream. In this direction, we perform a performance assessment of VR networks in the network simulator 3, that reveals that adaptation of the data rate is indeed necessary to provide the best possible VR data at the client. Moreover, it becomes evident that TCP transport, in combination with a data rate adaptation algorithm significantly reduces both the packet loss and the network delay, while maintaining high throughput. Finally, the proposed fuzzy algorithm outperforms well known adaptation algorithms in terms of throughput, delay and fairness.