A Methodology for Estimation of Software Architectural Complexity in Publish-Subscribe Systems

Abstract

Software systems are growing in complexity, and the trend is accelerating due to several forces in the industry. One of them is using more complex components that perform higher-level functions and use AI algorithms. The other is that architectures are expected to be universal and versatile and to support different types of hardware platforms. Reusing third-party components makes some of the tasks more manageable. Still, there is not always a complete understanding of how to estimate the complexity of a specific architecture, especially one based on diverse components from different sources. It is not a straightforward route to come up with an objective and quantitative measure that can estimate the complexity of particular software architecture, but it can be instrumental. Such a methodology is developed in this paper to be applied to a class of architectures for comparison, planning, and maintenance. The trend to be able to modify some architectures dynamically at runtime requires the user to have an idea of how the complexity can change when architectures are modified. Our study focuses on publish- subscribe style architectures used mainly in the robotics industry. The approach can be applied to any architecture with similar message delivery mechanisms and distributed components or subsystems.