Our paper “Secure Point Cloud Streaming with Learning-Based Super-Resolution” has been accepted for publication in the International Journal of Semantic Computing.
The paper presents a system for real-time AR/VR point cloud streaming that downsamples and partially encrypts content at the origin server, then decrypts and upscales it at the client using an ML-based super-resolution model integrated into an adaptive bitrate (ABR) streaming pipeline. Evaluated on a full CloudLab deployment, the SR-aided approach raises average delivered resolution from about 18% to 31% of full density under constrained network conditions — a 76% relative improvement over an ABR-only baseline — while cutting bandwidth and encryption/decryption overhead.
Authors: Mohammad Waquas Usmani (University of Massachusetts Amherst), Sankalpa Timilsina (Tennessee Technological University), Michael Zink (University of Massachusetts Amherst), and Susmit Shannigrahi (Tennessee Technological University).