Real-time video analysis at the edge of the network can reduce response latency and cost of transmission, which is very promising to significantly improve public safety. The Detroit Police Department initiated a crime-fighting project called "Project Green Light Detroit", which leverages the real-time video streams between police headquarters and partners who have installed more than hundred high-quality video cameras. The collected video data will be the input of the EVAPS. In the EVAPS, it will analyze the live and archived video data simultaneously at edge nodes that are computing resources closing to the video data sources, to identify potential dangerous events/objects (e.g., gunshot in a crime, the kidnapper's vehicle of an AMBER alert).
Connected vehicle technology is standard in many new cars, where hundreds of sensors are installed with advanced driver-assistance system and up to 4TB of data are generated. Only by analyzing the data can one reveal meaningful connections, trends and patterns that can help provide a better driving experience and improve vehicle quality and reliability. In this project, we intend to build an open vehicular data analytics platform (OpenVDAP) based on Firework for real-time on-vehicle data analytics, including telematics, vehicle status diagnostics, driving assistant, and battery management for electric vehicles (EVs), etc.