Self-adaptive mechanism for disaggregated computing

This paper proposes an approach to adapting software in a pervasive computing environment whose computational resources are heterogenous, limited, and dynamic. It is unique to other existing adaptive approaches in introducing the notions of differentiation and dedifferentiation in cellular slime molds. When a software component delegates a function to another component coordinating with it, if the former has the function, this function becomes less-developed and the latter´s function becomes well-developed like that in cellular differentiation. When some differentiated components are not available due to system errors or network partitioning, this enables remaining components to automatically activate their degenerated functions corresponding to functions provided by the unavailable components like those in cellular-dedifferentiation. The approach enables applications to be naturally adapted to changes in their computational environments, e.g., other applications, their available resources, and network connectivity, without any special adaptation mechanisms inside the applications themselves in a self-organizing manner. It is constructed as a middleware system to execute general purpose applications on real pervasive computing systems, instead of any simulation-based systems. We present several evaluations of the framework in a distributed system and describe two applications.