A Novel Strong Cache Consistency Mechanism in ICN Based on Role Division and Lease Model
Shuokang Huang,Shuai Zhu,Kai Lei,Haiyi Lian,Shouyang Wei
DOI: https://doi.org/10.1109/ispa-bdcloud-socialcom-sustaincom51426.2020.00184
2020-01-01
Abstract:Information-Centric Networking (ICN) takes advances in content distribution, thanks to the in-network caches which naturally reduce the delay of upper-layer applications. Though attaining high dissemination efficiency, the in-network caching amplifies the challenges in the consistency of cached copies. Since ICN decouples data from the locations, it's hard for ICN to synchronize the caches efficiently once an update occurs, which could lead to potential issues of correctness and security. Herein, we propose a novel cache consistency mechanism based on role division, namely RD-CCM, to achieve strong consistency in ICN. RD-CCM divides network nodes into several roles and applies the Lease model among specific routers to maintain synchronization with high efficiency. We set up an adaptive parametric model to determine the Lease duration of RD-CCM, which balances the cost between storage and communication. Further, we optimize RD-CCM based on content popularity and Invertible Bloom Filter (IBF), diminishing the overhead of synchronization. We simulate RD-CCM on ndnSim to evaluate its performance, compared against Time- To-Live (TTL) and Polling-Every- Time (PET), which are employed in Named Data Networking (NDN). RD-CCM reduces the average hop counts by 32.5% compared against NDN-PET and reduces the cache staleness rate by 59.2% compared against NDN-TTL.