H.264 Compress-Then-Analyze Transmission in Edge-Assisted Visual SLAM
Konferenz: European Wireless 2023 - 28th European Wireless Conference
02.10.2023-04.10.2023 in Rome, Italy
Tagungsband: European Wireless 2023
Seiten: 6Sprache: EnglischTyp: PDF
Autoren:
Hofer, Johannes; Steinke, Jannek (Deutsche Telekom Chair of Communication Networks, TU Dresden, Germany)
Sossalla, Peter (Deutsche Telekom Chair of Communication Networks, TU Dresden, Germany & Audi AG, Ingolstadt, Germany)
Nguyen, Giang T. (Haptic Communication Systems, TU Dresden, Germany & Centre for Tactile Internet with Human-in-the-Loop (CeTI), Germany)
Fitzek, Frank H. P. (Deutsche Telekom Chair of Communication Networks, TU Dresden, Germany & Centre for Tactile Internet with Human-in-the-Loop (CeTI), Germany)
Inhalt:
Edge-assisted Visual Simultaneous Localization and Mapping (V-SLAM) systems offload complex analysis modules from computationally constrained mobile devices to cloud servers at the network edge. As visual data must travel through networks to reach the edge cloud, it is crucial to minimize transferred data while maintaining reliable and accurate localization and map generation. State-of-the-art systems perform pre-analysis of SLAM modules on the mobile device, causing high processing delays and limiting functionality. To minimise both throughput demand and latency, we are investigating video compression techniques in Compress-Then-Analyze (CTA) transmission systems. Evaluations employing the widely used H.264 video compression standard demonstrate that CTA achieves lower throughputs than its counterparts, as well as significantly reduced processing delays and transmission latencies. The results open up new possibilities to explore this research direction on edge-assisted V-SLAM further.