Novel Validation Algorithms for Safety Critical Embedded Software

This paper presents implementation of computationally efficient input data health-monitoring algorithms for a safety critical embedded system. These algorithms are used in the 14-seater indigenously developed passenger aircraft. The embedded system provides warning to the aircrew about the takeoff, landing, overspeed, stall, pitch trim and hydraulic low pressure. There are three algorithms that are developed for three different types of signals. These signals are analog, discrete and ARINC. Each of the signal characteristic is different from other so a modified moving window algorithm is developed for analog signals, modified debouncing validation algorithm for discrete signals and a modified persistence time algorithm for ARINC signals. The algorithms are verified using rigorous software and hardware level testing methodologies. They are finally validated during the test flights. The positive flight trial results demonstrate the functionality and reliability of these algorithms.