Complexity Metrics for Hierarchical State Machines

Automatically generated state machines are constrained by their complexity, which can be reduced via hierarchy generation. A technique has been demonstrated for hierarchy generation, although evaluation of this technique has proved difficult. There are a variety of metrics that can be used to provide indicators of how complicated a state machine or statechart is, one such example is cyclomatic complexity (the number of edges - the number of states + 2). Despite this, the existing complexity metric for statecharts does not operate on the hierarchy, instead providing an equivalent cyclomatic complexity for statecharts by ignoring it. This paper defines two new metrics; Top Level Cyclomatic Complexity and Hierarchical Cyclomatic Complexity. These metrics assess the complexity of a hierarchical machine directly, as well as allowing for comparison between the original, flat state machine and its hierarchical counterpart.