ABSTRACT Software process simulation (SPS) has been evolving over the past two decades after being introduced to the software engineering community in the 1980s. At that time the SPS technology attracted a great deal of interest from both academics and practitioners in the software process community—even to the extent of being one of the recommended techniques for achieving multiple Key Process Areas of Level 4 of the Capability Maturity Model Integration. However, in recent years, the growth of SPS seems to have slowed along with the number of reported applications in industry. This article summarizes the special panel that was held during ICSSP 2012 whose goals were to assess whether this technology remains applicable to today's software engineering projects and challenges and to point out the most beneficial opportunities for future research and industry application. Copyright © 2014 John Wiley & Sons, Ltd.

Simulation is an important method and tool in many fields of engineering. Compared to these, simulation plays only a minor role in the field of software processes and software engineering. Examining this discrepancy, four theses are formulated as suggestions for future directions of software process simulation: 1. Simulation requires efforts, but “not simulating” might cause considerable costs as well, e.g. by wrong assumptions or expectations. These costs must be addressed and understood as well. 2. A model is always a simplification with many uncertainties. However, this is not a counter-argument by itself but must be evaluated in the perspective of purpose and available alternatives. 3. Future process simulation models must and will be much more complex than today. The necessary complexity can only be handled by relying on a rich set of mature components. This requires a joined effort and appreciation of the respective groundwork. 4. There are areas of software process modelling, which have already achieved some maturity, e.g. the interrelationships of volume of work, productivity, resources, and defect injection and removal. However, there are other aspects, which need further research to develop adequate modeling concepts, e.g. influence of architectural quality on later process stages, influence of process area capabilities within a dynamic simulation, or combined effects of human factors like time pressure, motivation, or knowledge acquisition.

Managed software development organizations are de facto closed loop systems, in which management actions influence future behavior based on observations of previous behavior thus forming a closed feedback loop. It is well known from control theory that a closed loop system might exhibit dynamics very different from the dynamics of its open loop components. Therefore, it is the goal of this work to study the importance of closed loop effects in the context of software process dynamics. To this end, an exemplary model is presented with a work-test-rework cycle as operational part and a productivity estimation and personnel allocation model as management component. The behavior of this closed loop system is probed with respect to typical disturbances and nonlinearities of software processes. The results underscore the importance of such a closed loop analysis.

Software process improvement needs guidance in proposing, assessing, and selecting a right set of steps and measurements. Often, a process model or a specific methodology is used for this purpose, e.g. the V-Modell, RUP, or agile process methodologies. Each such model or methodology consists of certain activities, procedures, or methods, i.e. a set of process fragments. Debates usually focus on the choice among these models or methodologies as a whole, but not on the specific, quantitative contributions of the distinct process fragments with respect to the particular goals of the organization. Such a simplification misses important chances for adaptation and motivation in diversified system development environments. Therefore, a conceptual framework and tool set is presented in this paper to search for sets of process fragments, which are able to meet actual objectives in an optimal way.

The problem of reinforcing local evidence of linear structure while suppressing unwanted information in noisy images is considered, using a modified form of relaxation labeling. The methodology is based on parametrizing a continuous set of orientation labels via a single vector and using a sigmoidal thresholding function to bias neighborhood influence and ensure convergence to a meaningful stable state. Label strength and label/no-label decisions are incorporated into a single functional. Optimal points of the functional represent the cases where as many pixels (objects) as possible have achieved the desirable linear-structure-reinforced and noise-suppressed labelings. Three different linear structure reinforcement tasks are considered within the general framework: edge reinforcement, edge reinforcement with thinning, and bar (line segment) reinforcement. Results from several image data sets are presented. This approach can directly handle continuous feature information from low-level image analysis operators, and the computational complexity of labeling is reduced.>

The problem of reinforcing local evidence of edges while suppressing unwanted information in noisy images is considered using a form of relaxation labeling. The methodology is based on parameterizing a continuous set of edge orientation labels using a single vector. A sigmoidal thresholding function similar to that used in artificial neural networks to bias neighborhood-influence and insure convergence to meaningful stable states is also utilized. A global optimization function is defined, and a decentralized parallel algorithm is derived that uses a steepest-gradient-descent approach to arrive at the optimal point on the functional surface, corresponding to desirable edge-reinforced and noise-suppressed labelings. In addition, a modification to the functional is presented which incorporates a thinning operation to insure that each edge is marked by only a single-pixel-wide response. Results from several image data sets indicate that the algorithm performs as well as or better than other relaxation labeling methods, and with improved computational efficiency.>