AppShot: A Conditional Deep Generative Model for Synthesizing Service-Level Mobile Traffic Snapshots at City Scale
Date:
Conference talk at SCONE – the SCOttish Networking Event, Edinburgh, UK
Abstract: Service-level mobile traffic data enables research studies and innovative applications with a potential to shape future service-oriented communication systems and beyond. However, real-world datasets reporting measurements at the individual service level are hard to come by due to privacy or industrial secrecy reasons. APPSHOT is a model for producing synthetic high-fidelity snapshots of the traffic generated by mobile services. It can operate in any geographical region for which contextual information typically found in public repositories is available, thus allowing the generation of new and open traffic datasets for academic research. APPSHOT stems from an original characterization of service-level mobile traffic data, which underpins a novel conditional GAN design instantiated by a convolutional neural network generator and two discriminators. The model features multiple innovative mechanisms including multi-channel generation and use of FiLM layer to address the unique challenges involved in generating mobile service traffic snapshots. Experiments with ground-truth data collected by a major operator in multiple metropolitan areas show that APPSHOT can produce realistic network loads at service level for areas where it has no prior traffic knowledge, and that such data reliably supports service- oriented networking studies.