BACKGROUND Software technology and software systems greatly impact technological products, economic activities, defense, scientific research, and social life. The complexity of software continues to grow. Failures of software projects and software systems may incur high financial costs and even human life. There is no doubt that various software development processes and the complicated behavior of software systems must be kept functional and even evolved in the context of a changing environment. Conceptually, this is precisely the purpose of control theory, and hence the marriage of software and control engineering can be seen as the first stage in the development of software cybernetics. Presently, most software development follows ad hoc approaches and depends heavily on software development personnel and company resources. Feedback mechanisms, ubiquitous in software processes and systems, have not been formalized, quantified, or optimized. Since feedback and optimization are two central themes in control and decision theories, a natural question to further ask is: what roles can feedback control based approaches play in the control of various software processes and systems and, more fundamentally, in their development? Further, the widespread deployment of computers and embedded software in control systems poses a challenge to existing control theories that do not account for the special characteristics of software. In order to achieve satisfactory control of processes or, for example, the future intelligent home, the evolutional nature of software should be considered in synthesizing control policies. An important example of such a synthesis is the improvement of the reliability of fly-by-wire systems in modern aircraft whose underlying control laws should be robust to certain classes of software faults. It seems reasonable to consider software problems in the light of control theoretic formulation. Software cybernetics unifies and expands various seemingly unrelated research topics under different umbrellas, such as adaptive software, adaptive rejuvenation, active security enhancement, supervisory control approaches applied to software synthesis, etc. It also gives birth to new and challenging research topics, such as feedback control of the software test process and adaptive testing. |
||||
