Next Generation Software Process and Their Support Overview

In his initial address, Barry Boehm overviewed the objectives of the workshop and its relation to the USC-CSE research program. This research program concurrently addresses three major software front-end issues:

This workshop is focused on NGPM/SS, with the objectives of assessing new directions in software life-cycle process, providing guidelines for affiliate activities, and stimulating USC-CSE/Affiliate collaboration in the software process area. Dr. Boehm's presentation illustrated the inter-relationship among CSE's three research directions. The software architecture research is expected to provide better understanding of candidate application areas for NGPM/SS, such as Satellite Ground Station (SGS) and, more generally, OCC. The cost modeling research is expected to help NGPM/SS evaluate objectives and offer alternatives subject to economical constraints.

In the next presentation, Prasanta bose presented experience with USC/CSE's NGPSS-0 prototype and results from the corresponding bootstrap experiment. THe NGPSS-0 prototype was based on Perceptronics' CACE/PM. Bose demonstrated hypothesis of the experiment: the adequacy of NGPSS-0 for supporting the NGPM; the generality of win conditions on NGPSS for applying to other software engineering environments (SEE's); and the generation of product and proces specific objectives, constraints and alternatives by adding Theory W to Spiral Model. In this experiment, four USC/CSE researchers played roles as user, developer, customer, and system engineer to specify win conditions on NGPM/NGPSS, identify conflicts, risks, and uncertainties (CRU's), resolve conflicts, and generate points of agreement (POA's) for the NGPSS-1. The experiment results show that NGPSS-0 language was inadequate for supporting win conditions and CRUs; however, the experiment itself was adequate in providing insights for the NGPM and valuable enhancement for the NGPSS. On the hypothesis of applying NGPSS-0 win conditions to other SEE's, the results demonstrated the commonality of NGPSS-0 and TRW SEE win conditions. The results also showed that product and process oriented objectives, constraints and alternatives are generated from win conditions. Bose concluded that future work on NGPM/SS would be to develop a SEE for domain specific information architecture.


Back to the Center for Software Engineering home page