A macro-micro system architecture analysis framework applied to Smart Grid meter data management systems by Sooraj Prasannan.

摘要

This thesis proposes a framework for architectural analysis of a system at the Macro and Micro levels. The framework consists of two phases -- Formulation and Analysis. Formulation is made up of three steps -- Identifying the System Boundary, Identifying the Object-Process System levels using the Object-Process Methodology (OPM) and then creating the Dependency Matrix using a Design Structure Matrix (DSM). Analysis is composed of two steps -- Macro-Level and Micro-Level Analysis. Macro-Level analysis identifies the system modules and their interdependencies based on the OPM and DSM clustering analysis and Visibility-Dependency Signature Analysis. The Micro-Level analysis identifies the central components in the system based on the connectivity metrics of Indegree centrality, Outdegeree centrality, Visibility and Dependency. The conclusions are drawn based on simultaneously interpreting the results derived from the Macro-Level and Micro-Level Analysis. Macro-Analysis is vital in terms of comprehending system scalability and functionality. The modules and their interactions influence the scalability of the system while the absence of certain modules within a system might indicate missing system functionality. Micro-Analysis classifies the components in the system based on connectivity and can be used to guide redesign/design efforts. Understanding how the redesign of a particular node will affect the entire system helps in planning and implementation. On the other hand, design Modification/enhancement of nodes with low connectivity can be achieved without affecting the performance or architecture of the entire system. Identifying the highly central nodes also helps the system architect understand whether the system has enough redundancy built in to withstand the failure of the central nodes. Potential system bottlenecks can also be identified by using the micro-level analysis. The proposed framework is applied to two industry leading Smart Grid Meter Data Management Systems. Meter Data Management Systems are the central repository of meter data in the Smart Grid Information Technology Layer. Exponential growth is expected in managing electrical meter data and technology firms are very interested in finding ways to leverage the Smart Information Technology market. The thesis compares the two Meter Data Management System architectures, and proposes a generic Meter Data Management System by combining the strengths of the two architectures while identifying areas of collaboration between firms to leverage this generic architecture.