In-situ Data Analytics In Cyber-Physical Systems

摘要: Cyber-Physical System (CPS) is an engineered system in which sensing, networking, and computing are tightly coupled with the control of physical entities. To enable security, scalability resiliency, new data analytics methodologies required for computing, monitoring optimization CPS. This work investigates related challenges CPS through two study cases: Smart Grid Seismic Imaging System. For smart grid, this provides a complete solution management based on novel in-situ designs. We first propose important tasks power monitoring: grid topology change power-line outage detection. address issue low measurement redundancy identification, particularly low-level distribution network, we develop maximum posterior mechanism, capable embedding prior information breakers status to enhance identification accuracy. In detection, existing approaches suffer from high computational complexity security issues raised centralized implementation. Instead, presents distributed framework, carries out in-network processing invokes complexity, requiring only simple matrix-vector multiplications. functionality, also restoration strategy involving reconfiguration resource planning after faults or changes. seismic imaging system, several innovative schemes each sensor node computes tomography its partial gossip local neighbors. The generated fashion originally. Different conventional approach collection then order, our proposed methodology much more efficient. underlying mechanisms avoid bottleneck problem bandwidth since all processed nature limited decisional communicated. Furthermore, algorithms can deliver quicker insights than state-of-arts imaging. Hence they promising solutions real-time analytics, highly demanded disaster applications. Through extensive experiments, demonstrate that methods able achieve near-optimal quality tomography, retain communication cost, provide analytics. INDEX WORDS: System, Grid, In-Network Processing, In-Situ Analytics, Distributed Computing. IN-SITU DATA ANALYTICS IN CYBER-PHYSICAL SYSTEMS