Smart caching for live 360$^o$ video streaming in mobile networks

Despite the advances of 5G systems, the delivery of 360$^o$ video content in mobile networks remains challenging because of the size of 360$^o$ video files. Recently, edge caching has been shown to bring large performance gains to 360$^o$ Video on Demand (VoD) delivery systems, however existing systems cannot be straightforwardly applied to live 360$^o$ video streaming. To address this issue, we investigate edge cache-assisted live 360$^o$ video streaming. As videos' and tiles' popularities vary with time, our framework employs a Long Short-Term Memory (LSTM) network to determine the optimal cache placement/evictions strategies that optimize the quality of the videos rendered by the users. To further enhance the delivered video quality, users located in the overlap of the coverage areas of multiple SBSs are allowed to receive their data from any of these SBSs. We evaluate and compare the performance of our method with that of state-of-the-art systems. The results show the superiority of the proposed method against its counterparts, and make clear the benefits of accurate tiles' popularity prediction by the LSTM networks and users association with multiple SBSs in terms of the delivered quality.