Peripheral cellular control: an eigen-frequency model and a case-study in self-organization

Algorithmic behavior-based control architectures usually suffer from rigidity in the network structure which makes fault-tolerance an important issue when it comes to distributed hardware implementations. We are extending here the basic cell model for the example case of a peripheral navigation system that needs to demonstrate such plasticity in the sense of tolerance to hardware failures. This biologically-inspired extension consists in equipping each one of the cells with multiple motivations to process/consume different kinds of messages and in cancelling the rigid connections between levels so that cells of one level will compete for the consumption of the messages coming in from the previous level. Individual cell “motivations” are expressed as continuous variables that are excited by certain identification values of the incoming messages, thus as eigenfrequencies. Simulation results indicate that the cellular network self-organizes in case of failures by “discovering” alternative message flow pathways and that multiple failures slow down the system´s responsiveness to external events, since intermediate level cells are not specialized to one message type but distribute their processing time among different motivations. The presence of an additional developmental/ontogenetical factor allows for self-organization even in the case of major failures in the network, i.e. when a particular eigenfrequency disappears completely from the system