A Stacking Learning-Based QoE Model for Cloud Gaming

Enhancing Cloud Gaming QoE Estimation by Stacking Learning

“…The introduction of these novel features was facilitated through a questionnaire administered to the participants. Noteworthy differences and intriguing discoveries between these two datasets, particularly in the second dataset, have been detailed in the study [9].…”

“…In our other works [9,10], we utilized a wired cloud gaming testbed, since at the time GeForce Now had just being launched [20] and most cloud gaming services only supported PCs. This First testbed that was used to gather the original data had a server operating on Windows 10 and GeForce Experience and Gamestream protocol enabled, equipped with an i7-3770 processor, 16GB of DDR3 RAM, and a GTX 1050Ti graphics card.…”

“…Numerous studies explore QoE in cloud gaming or propose platform-specific enhancements, as evident in works such as [7,8]. As of now, to our knowledge, just a few QoE models have been devised solely based on human-provided rankings, and most are our own [9,10].…”

“…Building upon our earlier publication [9], this article extends its scope by addressing gaming streamed via a wireless network and to mobile devices, constituting a mobile cloud gaming service. In our previous work, we considered a wired cloud gaming service where the gaming is streamed through a cable network to a PC.…”

“…The first model, created on our first work [9], was developed through a wired network, with tree games, 81 players and 2472 instances. The new model uses the original dataset and expands it with 856 new instances, with two games (one equal to the first and one new game) and 9 players, being 7 new players.…”