A comparison of discrete and hybrid models of an aircraft conflict detection algorithm

Aerospace systems are becoming more dependent on software for functions such as command and control. Software introduces many new interactions with other system components that are difficult to predict in their entirety. Traditional discrete modeling techniques for verifying safety properties of software systems do not always detect all undesirable behavior in hybrid systems. Modeling languages that are both continuous and discrete are better at encapsulating previously undetected interactions. An aircraft conflict detection scheme is used to illustrate the advantages of including continuous components in the system model. The system is first modeled in SpecTRM-RL, a discrete modeling language. Next, the example is modeled in an extended version of the SpecTRM-RL that allows discrete and continuous modeling. A comparison of the hazard analyses based on the two models shows that modeling both discrete and continuous behavior of the system aids in conflict detection by reducing the false alarm rate.