Developing a Certifiable UAS reliability Assessment Approach Through Algorithmic Redundancy

Manned aircraft, civilian or military, are required to meet certain reliability standards specified by the FAA in order to operate in the US national airspace. These reliability standards are typically met via hardware redundancy. Multiple, dissimilar components of hardware designated for a single purpose increases the physical redundancy and offers a viable way to detect and correct for hardware malfunctions. Unmanned Aircraft Systems (UAS) are often subject to weight/power constraints and therefore unable to accommodate similar redundancies. A common approach to solving this problem has been to employ analytical redundancy in the form of fault detection and isolation (FDI) algorithms. Although significant breakthroughs have been achieved in increasing UAS redundancies analytically, certification has been challenging. The primary objective of the proposed work is to develop a software architecture for UAS that provides certifiable analytical redundancy. The proposed approach is called algorithmic redundancy, owing to its underlying philosophy of increasing UAS reliability via multiple, dissimilar FDI algorithms. Certification of an algorithmically redundant system relies on well-established methods similar to those for hardware redundant systems. We propose to develop a framework for evaluating and selecting FDI algorithms, combining them to build algorithmically redundant software and finally, assess the overall reliability for certification