Safety Comparison of Centralized and Distributed Aircraft Separation Assurance Concepts

This paper presents several models to compare centralized and distributed automated separation assurance concepts in aviation. In a centralized system, safety-related functions are implemented by common equipment on the ground. In a distributed system, safety-related functions are implemented by equipment on each aircraft. Failures of the safety-related functions can increase the risk of near mid-air collisions. Intuitively, failures on the ground are worse than failures in the air because the ground failures simultaneously affect multiple aircraft. This paper evaluates the degree to which this belief is true. Using region-wide models to account for dependencies between aircraft pairs, we derive the region-wide expectation and variance of the number of separation losses for both centralized and distributed concepts. This derivation is done first for a basic scenario involving a single component and function. We show that the variance of the number of separation losses is always higher for the centralized system, holding the expectations equal. However, numerical examples show that the difference is negligible when the events of interest are rare. Results are extended to a hybrid centralized-distributed scenario involving multiple components and functions on the ground and in the air. In this case, the variance of the centralized system may actually be less than that of the distributed system. The overall implication is that the common-cause failure of the ground function does not seriously weaken the overall case for using a centralized concept versus a distributed concept.