Empirical evaluation of RaptorQ codes for multicast video streaming over IEEE 802.11ac wireless LANs

Abstract

In this dissertation, RaptorQ fountain codes are used to provide reliable video multicasting over IEEE 802.11ac Wi-Fi networks. Even though multicasting is efficient because of its one-to-many video delivery, it results in high packet loss since it offers no acknowledgement or retransmission mechanisms. The aim of this dissertation is to design, implement, and evaluate a complete testbed that integrates video encoding, RaptorQ forward error correction (FEC), Wi-Fi multicast transmission, decoding, and playback. The system combines FFmpeg, RaptorQ encoder and decoder and FFplay into a reproducible end-to-end pipeline. Three RaptorQ configurations were tested, where the number of source symbols K was fixed at 100 and the number of repair symbols H was varied between 0, 50, and 100. The testing was done at the University of Malta faculty of ICT Auditorium, where three different positions were tested. Reliability at packet and block levels, video quality, transmission rate, and end-to-end delay were evaluated through Symbol Error Rate (SER), Block Error Rate (BLER), Mean Square Error (MSE), and the end-to-end delay measurements. When no repair symbols were used, the multicast transmission suffered from large losses with an SER of approximately 1.05%-1.7% and a BLER of approximately 14%-56%. When 50 repair symbols were used, 72%-89% of the symbol losses were recovered, with SER ranging from 0.13%-0.25% and BLER ranging from 0.29%-0.52%. With 100 repair symbols, near-perfect recovery was achieved, with SER ranging from 0.045%-0.16% and BLER from 0.075%-0.24%. Video quality improved as the average MSE dropped from 300-900 (severe degradation) to 1.6-6.6 (nearly lossless). The end-to-end delay remained below the 0.5-1s threshold defined for the auditorium scenario across all configurations. These results confirm that RaptorQ can drastically improve the reliability of Wi-Fi video multicasting because both packet-level reliability and visual quality significantly improved with minimal delay. The implemented testbed serves as a reusable framework for future research on video protected by FEC streamed over wireless networks.