An infrastructure for the life cycle management of multi product lines

Abstract

Large-scale industrial systems comprise multiple heterogeneous systems that need to be adapted and configured individually to meet the customers' needs. Product line techniques such as variability modeling and product derivation are frequently used to enable such customization. However, managing the life cycle of multiple product lines in systemof-systems architectures is challenging. For instance, support is lacking for defining the architecture of multi product lines. Further, system architects, modelers and configurers are missing a common infrastructure supporting domain and application engineering. This paper contributes a tool infrastructure providing essential structures and facilities for the development and operation of multi product lines. Based on an analysis of industrial scenarios we developed support for managing the life cycle of multi product lines. This comprises architecture definition and support for sharing and deploying product line models to the involved stakeholders. In this paper we show how the infrastructure supports system architects performing life cycle management of multi product lines. We demonstrate the feasibility of the approach based on a real-world industrial system.

Delta-oriented multi software product lines

“…Several commercial tools, such as pure::variants (www.pure-systems.com) or GEARS(www.biglever.com), have support for multi-product line configuration in a similar manner [13]. In the DOPLER tool suite [15,8], product line bundles (PLiBs) integrate several SPLs into one MSPL from a tool-oriented perspective. PLiBs go beyond pure problem space variability by including configuration, but do not explicitly cover the solution space artifacts.…”

“…MSPLs are a union of several SPLs with a common variability model [9]. MSPLs are prevalent in today's large-scale systems, such as in industrial automation [8], data bases systems [14]. MSPLs are beneficial in large-scale system development as they allow reuse of existing SPLs in a new context, reduce complexity by decomposition of a large SPL and enable the distribution of work over several development teams.…”

“…Most existing MSPL modeling approaches have considered only the problem space variability model explicitly and treated the problem space code artifacts and the configuration mechanism as a blackbox [7,15,8]. However, in order to provide reuse within MSPLs also on the code level, we need a MSPL modeling approach that spans problem space (variability model), solution space (code artifacts) and configuration space (product generation).…”