A Study of Replica Selection Algorithm for Heterogeneous ICN Nodes
Information Centric Networking(ICN)is an innovative network architecture that breaks the constraints of traditional TCP/IP end-to-end transmission and improves content distribution efficiency.ICN establishes a network-wide caching system with multiple con-tent replica redundantly cached within the network for nearby access by users.Unlike traditional caching systems in the Internet,ICN's caching is characterized as ubiquitous,with networking devices serving as the infrastructure,resulting in widespread heterogeneity of service resources.In such an environment,selecting appropriate replica nodes becomes a significant research challenge.This paper first conducts abstract modeling and analysis of heterogeneous ICN nodes with M/M/l queueing model.The selection of heterogeneous replica is modeled as Multi-Armed Bandit(MAB)problem,and the UCB1 algorithm is introduced to explore and learn optimal deci-sions.Results of the simulation experiments demonstrate the algorithm's significant advantages in terms of improving cache service reli-ability and reducing content retrieval latency.The algorithm achieves a service reliability of 99.15％and reduces the average content retrieval latency by a maximum of 8.63％.
information-centric networkingin-network cachingreplica selectionM/M/1 queueing modelmulti-armed bandit problem
NETL
NSTL
万方数据
