A Novel Objective Video Quality Assessment Metric For Cloud Gaming Applications

This study introduces a unique method for evaluating the Quality of Experience (QoE) of cloud gaming applications objectively for various network conditions, involving different packet latency and packet loss ratio (PLR) parameters. We use a combination of a perceptual hash (pHash) and a structural similarity index measure (SSIM) to detect the duplication of packets with greater accuracy than either pHash or SSIM alone. We discovered that increased PLR causes more frame duplication in cloud gaming applications, which lowers the end-user QoE. Using the NetEm emulator and the Parsec streaming platform in a controlled network environment, we evaluated our model on three distinct games based on content complexity, genre, and popularity in cloud gaming applications. We observed that different content complexity of games is impacted differently by varying network conditions; for example, slow-moving games are more affected by PLR and frame duplication. We also carried out a subjective QoE evaluation of cloud gaming with 22 volunteers. The results obtained from subjective testing were used to extend our study to compare the accuracy of our model compared with existing Full-Reference (FR) and No-Reference (NR) techniques. We discovered that our model surpasses the majority of the currently available solutions in terms of predicted accuracy, many of which were based on Video on Demand (VoD) applications. This demonstrates how important it is to create a QoE model, particularly for cloud gaming given its distinctive utilisation of network parameters.