Validation of Complexity Metrics of Agent-Based Systems Using Weyuker's Axioms

Abstract

Agent-oriented software, in spite of its increased complexity levels, emphasize on social ability, autonomy, reactivity, adaptability, intelligence, learning, proactivity, goal-oriented and mobility. In general the complexity of the software agent is divided into two aspects; computational complexity and psychological complexity and they should be measured on both concrete aspects. The fact of high computational complexity leads to a weak performance indicates the importance of correctness in complexity measurements. Hence the concept of complexity metrics validation plays a significant role in the complexity measurement theory. This work describes a standard procedure for validating complexity metrics of agent-based systems. The validation procedure narrated in this paper uses the method proposed by Weyuker. The nine axioms of Weyuker are used to determine the effectiveness of various complexity metrics of agent-based systems. The result defines the extent of satisfaction of validation axioms in case of agent complexity metrics.