Prominence-guided link prediction in fuzzy robotic networks

As the complexity and unpredictability of cyber-physical systems (CPSs) such as multi-agent robotic networks increase, having robust predictive models is crucial for ensuring dependable operations. This paper presents a modification of the Strength Prominence (SP) index, which was initially designed for fuzzy social networks, adapted for use in robotic and intelligent automation systems. The SP index has been reformulated for fuzzy interaction graphs, where nodes signify robotic components and edges represent uncertain communications or dependencies. The modified index assesses link probability by considering the strength of connectedness and prominence levels, even in the absence of common neighbors. Theoretical aspects such as symmetry, boundedness, and monotonicity are thoroughly demonstrated. Empirical validation utilizing real-world datasets and ROS-based robotic data shows that the SP index achieves superior predictive accuracy, surpassing traditional fuzzy indices like CN, RSM, and CAR in terms of precision, AUC, and AUP measurements. This method allows for the early identification of interaction failures, improves the prediction of collaboration, and aids in the development of fault-tolerant designs. This proposed approach provides a new interdisciplinary tool for fuzzy link prediction in CPS, with important implications for the design of autonomous systems, real-time robotic collaboration, and resilient network structures.