Adaptive network-based fuzzy inference model on CPS for large scale intelligent and cooperative surveillance

Abstract

This paper proposes Cyber-Physical System (CPS) architectures for Large Scale Intelligent and Cooperative Surveillance (ICS) and presents an Adaptive Network-based Fuzzy Inference Model (ANFIM) for reducing the false accident-event report rates of Controller Manager (CM) systems among CPS components. In ICS, sensor nodes detect traffic accident events from general roads and crossroads and transmit event attribute values (car and accident types), time attribute values (date, month, day, year), environmental variable values (weather, temperature, luminous intensity, humidity and sound), and accident event videos to the Controller through WSN mesh networks. Using these pieces of information, the Controller conducts image processing and delivers all sensing data to the CM together with accident-event probability values. The CM conducts data filtering with multi regression analysis utilizing the past accident event database. Centering on the resultant values obtained through the foregoing process, this paper proposes a mathematical model for influence analysis of environmental variables that affect accident events. Based on the model's results, the authors establish a Sugeno Type fuzzy rule base for reducing false accident-event report rates to propose an ANFIM.