Back to search

Aristotle University Of Thessaloniki (AUTH)

Optimization of 360° video transmission

Abstract

dc:description

Recent advancements in immersive technologies like augmented and virtual reality (VR) have driven the adoption of 360° omnidirectional videos. However, streaming 360° video presents massive challenges, requiring high bitrates and ultra-low latency to maintain a satisfactory quality of experience (QoE). This thesis addresses these challenges by proposing novel optimization methodologies for tile-based 360° video streaming across three core pillars. The first pillar establishes a foundation through user behavior analysis and video compression evaluation. By analyzing VR users' head movements and tile-viewing patterns, this work identifies significant potential for bitrate savings. Deep learning viewport prediction is developed to classify tiles as watched or unwatched based on historical traces; a Long Short-Term Memory (LSTM) network is found to achieve the highest classification accuracy at 94.62% and is adopted for proactive streaming. Additionally, an evaluation of video compression standards reveals the superiority of VVC/H.266 over other existing schemes. Testing various tiling granularities shows that an 8x8 grid offers the optimal trade-off between streaming flexibility and compression efficiency. The second pillar focuses on the tile quality selection problem for multi-user transmission in bandwidth-constrained environments. Using a viewport-aware 360° video quality assessment (360° VQA) metric as the objective function, the problem is formulated as a multiple-choice knapsack problem (MCKP). A scalable greedy algorithm assigns optimal tile quality levels by leveraging aggregated tile weights across user viewports. By prioritizing central viewport tiles over edge tiles, this approach demonstrates strong resilience to viewport prediction errors. In simulations with eight concurrent users, the proposed method improves the delivered viewport weighted-to-spherical peak signal-to-noise ratio (VWS-PSNR) by 0.67 to 1.28 dB over baseline methods. The third pillar addresses quality assessment for multi-quality, tile-based 360° videos. A subjective experiment with 27 users across six videos encoded with twelve distinct tile encoding patterns resulted in the STAV360 dataset, which is the first 360° VQA dataset for mixed-quality tile-based streaming. Leveraging the collected Mean Opinion Scores (MOS) and viewing directions, a set of objective 360° VQA metrics considering spatial quality variations was developed. The proposed tile-based and viewport-tile-based VQA models significantly outperform full-frame metrics, with the top-performing model achieving a Pearson’s correlation coefficient (PCC) of 0.880 and a Spearman’s rank-order correlation coefficient (SROCC) of 0.883. Together, these contributions advance immersive content delivery by modeling VR user behavior and perception, facilitating intelligent streaming strategies, and enabling seamless, high-quality VR experiences.

Degree

thesis:*
Grantor dc:publisher
Aristotle University Of Thessaloniki (AUTH)
Year dc:date
2026

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Mahmoud, Moatasim-Tariq-Rashad
  • Μαχμούντ, Μουτασίμ-Ταρίκ-Ρασάντ

Subjects

dc:subject × 14

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/62343

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Mahmoud, Moatasim-Tariq-Rashad; Μαχμούντ, Μουτασίμ-Ταρίκ-Ρασάντ. Optimization of 360° video transmission. Aristotle University Of Thessaloniki (AUTH), 2026. http://hdl.handle.net/10442/hedi/62343