https://www.mdu.se/

The idea with product integration is that separate components are combined into a working system. However, this process of assembling parts into bigger units, products and systems is not well performed in industry, especially not when a substantial part of the product functionality is implemented in software. Many faults that are introduced in early phases are found as late as in the product integration phase, or even worse, in the verification or validation of the final delivery, or after delivery of the product or system. This leads to high costs for error correction and additional efforts for re-testing. There is consequently a need to further investigate the area of product integration to understand how the performance can be improved. Different practices have been described in standards and models, but the area is still under development. No widely agreed upon body-of-knowledge has so far been defined for product integration. A large part of the development of products containing software for industrial use is conducted in small or medium sized teams. This requires that any data collection methods used to acquire reliable information regarding performance in a project or organization minimize the intrusion. A facilitating approach was needed to understand how units with distinct characteristics should be approached. Based on several years of interaction with different types of organization, the presented research includes an analysis of various methods for data collection. The result is a proposed method for selecting different sizes of investigations based on the openness and maturity of the organization. The main purpose of this research is to understand which factors influence the integration process and what can be done to improve the execution of it. It includes investigations to understand if the described best practices are appropriate, and if there are other means to achieve successful product integration. The research combines investigations of existing compilations of best practices with case studies in industry. Our conclusion is that the type of organization that we have investigated can reduce problems in the product integration process by following the basic practices described in standards and reference models. Problems found in product integration can in most cases be related to the fact that the organization does not follow the proposed practices. The investigations have revealed that the practices are not used in a sufficient way, that additional efforts must be put into fulfilling the requirements in standards and models, and that it is difficult to implement the practices. We have also found indications that specific technology, component based software, may assist in executing the practices. Finally, we conclude that not all standards and models include support to avoid all types of problems in product integration. This is an indication that the on-going development of the area is necessary and that an increased agreement on what can be considered to be best practices is needed.

The product integration is a particularly critical phase of the software product development process as many problems originating from earlier phases become visible in this phase. Problems in product integration result in delays and rework. One of the measures to decrease the late discovery of problems is the use of development standards and guidelines that define practices to ensure correctness of the product integration. However, even if such standards and reference models exist, they are in not used consistently. One of the reasons is a lack of a proof that they indeed improve the integration process, and even more important, that they are sufficient for performing efficient and correct product integration.

The conclusion of the presented research is that the available descriptions in standards and reference models taken one by one are insufficient and must be consolidated to help development organizations improve the product integration process. The research has resulted in a proposed combination of the activities included in the different reference models. This combination has been based on a number of case studies. Through the case studies performed in seven different product development organizations, a relationship between problems that are observed and the failure to follow the recommendations in reference models is identified. The analysis has indicated which practices are necessary, and how other practices support these. The goal with the research is to provide product development organizations with guidelines for how to perform software product integration.

One additional finding of the research is the existence of relation between software architecture and the development process. A method for identifying dependencies between evolvement of software architectures and adaptation of integration practices has been demonstrated.