A sensor fusion system-oriented attack detection method, system and medium

Abstract

The application provides a kind of attack detection method, system and medium for sensor fusion system, it is related to communication security technical field, the method includes obtaining control input and each sensor measurement information, according to this, divide credible and suspicious sensor set and establish attack hypothesis set.By injecting bounded exposure input to control input, construct sensor layer, hypothesis layer and attack output set, use rolling time optimization to increase the separation trend between hypothesis layer and attack output set, while carrying out sensor layer detection.Based on the detection result and separation trend value, iteratively update the attack hypothesis set, repeat the active exposure detection until the stop condition is met, and finally identify the attack and the attacked sensor.The method expands the gap between normal output and attack output by actively injecting bounded exposure input and rolling time optimization, thereby achieving active exposure of covert deception attack and accurate identification of attacked sensor without modifying sensor interface and communication link.

Images