A Study on Migration Technique Using Program Slicing for Large-Scale Information and Control Systems

The software structure of large-scale information and control systems, such as that for railway and electronic power systems, becomes more and more complicated with all the modifications over long time use. Extension development is made without fixing the existing source code in order to maintain the existing functions. As a result, the source code for the data and data access are scattered throughout the program, and thus, it is difficult to modify them. We present a technique that can be easily modified in this paper to migrate the software structure. For this purpose, we focus on the program composition of information and control systems. These systems monitor facilities, check for differences between the planning and actual situations, and decide on the output to the facilities. This means the data access is composed of planning and facility data. The migration technique extracts the access codes for the planning and facility data from the existing source code using program slicing, and identify them by analyzing the relation of the data access codes. This enables us to manage them so that they do not re-scatter. We applied this technique to actual system software and confirmed its effectiveness.

I. Introduction

Extension development is made without fixing the existing source code of large-scale information and control systems, such as that for railway systems [1] and electronic power systems, in order to maintain the existing functions. As a result, their software structures become more and more complicated in longtime use. For example, some new data tables are added without changing the existing data tables even if they represent the same facilities defined by the existing ones. Thus, the source code for the data and data access are scattered throughout the program, and thus it is difficult to modify them. Therefore, a drop in the productivity becomes a problem. We have been developing legacy migration techniques to convert to a software structure with an improved level of productivity to try to solve this problem.