The landscape of mobile gaming has evolved significantly over the years, with profound changes in network reliability and traffic patterns. In the early 2010s, mobile games faced challenges due to unreliable networks and primarily featured asynchronous gameplay. However, in the current era, modern mobile games benefit from robust network connectivity, mirroring PC gaming experiences by relying on persistent connections to game servers. This shift prompted us to conduct an in-depth traffic analysis of two mobile games that represent opposite ends of the genre spectrum: a massively multiplayer game resembling PC MMORPGs with tightly synchronized gameplay, and a single-player puzzle game that incorporates asynchronous social interactions. Surprisingly, both games exhibited remarkably similar traffic footprints; small packets with short inter-packet arrival times, indicating their high expectations for network reliability. This suggests that game developers now prioritize network quality similarly to their PC gaming counterparts. Additionally, our analysis of packet lengths unveiled that recent mobile games predominantly employ short packets dominated by a few key packet types closely tied to player actions, which conforms to observations from PC online games. However, the self-similarity in traffic patterns, a notable feature in PC online games, only partially explains the traffic in mobile games, varying across genres. These findings shed light on the evolving traffic patterns in mobile games and emphasize the need for further research in this dynamic domain.

Audio auto-tagging is the process of assigning labels to audio clips for better categorization and management of audio file databases. With the advent of advanced artificial intelligence technologies, there has been increasing interest in directly using raw audio data as input for deep learning models in order to perform tagging and eliminate the need for preprocessing. Unfortunately, most current studies of audio auto-tagging cannot effectively reflect the semantic relationships between tags—for instance, the connection between “classical music” and “cello”. In this paper, we propose a novel method that can enhance audio auto-tagging performance via joint embedding. Our model has been carefully designed and architected to recognize the semantic information within the tag domains. In our experiments using the MagnaTagATune (MTAT) dataset, which has high inter-tag correlations, and the Speech Commands dataset, which has no inter-tag correlations, we showed that our approach improves the performance of existing models when there are strong inter-tag correlations.