This paper aims to accurately evaluate the aging state of nodes in largescale networks and identify the optimal rejuvenation plan for these nodes. To this end, the aging phenomenon in distributed systems was described as a random low-rank matrix. The CMAC network was introduced to collect the data of network nodes and evaluate their aging state and rejuvenation plan. Based on the aging state and plan applicability, the node relationship was integrated with matrix completion, aiming to improve the efficiency of aging evaluation. Compared to the traditional methods, our method significantly improved aging evaluation and reduced hardware cost, and offered suitable rejuvenation plans for aging nodes. The improvement is partially attributable to the incorporation of node relationship. The research findings shed new light on software aging and group rejuvenation.

Software Aging, Cerebellar Model Articulation Controller (CMAC), Group Rejuvenation, Matrix Completion