Designing Cooperating Self-Improving Systems

The issue we consider is how to gain the capability of useful self-improvement in systems that are expected to participate in various systems of systems during their lifetime. This is not the same as improvement in cyber physical systems (CPSs), even though large CPSs are systems of embedded systems. The difference is that we design and define the CPSs, so the integration onus is on us, not the systems and / or their components. In a System of Systems, we cannot define everything in advance, so the onus is on the participating systems. This paper is about designing systems so that they can help integrate, improve, and cooperatively improve themselves in a System of Systems that is unknown and unspecified at the time of their construction, and not completely known even during deployment.