Cache Management for Cloud RAN and Multi-Access Edge Computing with Dynamic Input

Abstract

Caching is one of the most popular ways to efficiently reuse previously retrieved or computed data. It has been widely used in many physical as well as virtualized computing systems. Cloud Radio Access Networks (CRAN) and Multi-Access Edge Computing (MEC) are two vital technologies proposed for 5G mobile networks. CRAN utilizes the cloud model on top of the traditional RAN, thus providing scalability and flexibility, and improved resource utilization. MEC brings cloud computing services closer to the users to provide high bandwidth, low latency, and real-time access and to improve the overall user experiences. A cache may be included in both CRAN and MEC architectures to speed up time-critical communications as well as to provide preferential services. Addressing the challenge of dynamic, often unbalanced traffic in 5G networks, new dynamic cache management algorithms supporting dynamic input are proposed for CRAN and MEC models. Zipf distribution is applied to determine the correlation between the input request distribution and cache placement. First, three new dynamic cache management algorithms, D-H-EXD-AHP (Dynamic Hierarchical Exponential Decay and Analytical Hierarchy Process), D-H-PBPS (Dynamic Hierarchical Probability Based Popularity Scoring), DRRM-PBPS (Dynamic Reverse Random Marking with PBPS), are proposed to improve the existing H-EXD-AHP, H-PBPS and RRM-PBPS respectively. Next, three dynamic versions of adaptive hierarchy-based cache management algorithms, D-AH (Dynamic - Adaptive Hierarchical), D-IAH (Dynamic – Improved AH) and D-GGAH (Dynamic - Good Guess AH), are transformed from their existing static algorithms, with the goal to improve guaranteed-QoS. Performance evaluation has been carried out using iFogSim. The proposed dynamic algorithms have achieved improved cache hit rate, cache access time, and reduced the time to reach minimum guaranteed QoS. This work contributes significantly in realizing the support of 5G for IoT by enhancing CRAN and MEC performance and would have wide applications in other real-time systems that require efficient dynamic cache management with guaranteed QoS.