A Novel Sensor Dynamic Reliability Evaluation Method and its Application in Multi-Sensor Information Fusion

Abstract

The improvement in recognition accuracy of a multi-sensor system is inseparable from the sensors' dynamic reliability, especially in the case of highly conflicting information. Recent research on multi-sensor system has shown that a difficult problem for the system designer is to figure out what factors are crucial in determining the reliability of the sensor. However, the acquisition of sensor reliability is affected by multi-factors in practical applications. In this paper, a novel method for sensor dynamic reliability evaluation based on evidence theory and belief entropy has been developed, aimed at obtaining a reasonable reliability parameter. Through experimental analysis and practical applications, the results indicate that sensor weight distribution and information entropy value could be used as efficient elements for higher recognition rate to improve the robustness and recognition rate of the multi-sensor system.

Keywords

• Sensor reliability
• belief entropy
• evidence theory
• target recognition
• evidence credibility weight
• Deng entropy