SAFERIDES: Application of decentralized control edge-computing to ridesharing monitoring services

Abstract

Edge computing changed the face of many industries and services. Common edge computing models offload computing which is prone to security risks and privacy breach. However, advances in deep learning enabled Internet of Things (IoTs) to onload tasks and run cognitive tasks locally. This research introduces a decentralized-control edge model where computation and decision-making are moved to the IoT level. The model aims at decreasing communication and computation dependance on the edge which affect efficiency and latency. The model also limits data transfer to the edge to avoid security and privacy risks. Decentralized control is a key to many business applications that prioritizes safety, real-time response, and privacy such as ridesharing monitoring and industrial operations. To examine the model, we developed SAFERIDES, a scene-aware ridesharing monitoring system where smart phones are detecting violations at the runtime. Current monitoring systems require costly infrastructure and continuous network connectivity. However, SAFRIDES uses optimized deep learning models that run locally on IoTs to detect and record violations in ridesharing. The system achieved the lowest latency among current solution, while minimizing data sharing and maintaining privacy. Moreover, decentralized edge computing empowers IoTs and upgrades their functionality from sensing to independent decision-making.