Abstract
Multi-Access Mobile Edge Computing (MEC) is proclaimed as a key technology for reducing service processing delays in 5G networks. One of the use cases in MEC is content caching as a way of bringing resources closer to the end-users. Consequently, both latency and QoE are reduced. However, MEC has a limited storage space compared to the cloud. Therefore, there is a need to effectively manage the cache storage. This article proposes and evaluates a novel scheme (PCR) that combines proactive prediction, collaboration among MECs and replacement algorithm to manage content caching in MEC. Results show that the proposed replacement scheme outperforms conventional baseline content caching algorithms LFU, LRU, MQ, FBR, LFRU. This has been validated with experimental results using a real dataset (MovieLens20M dataset) and comparison with contemporary Long Short Term Memory (LSTM) based caching algorithm.
Highlights
Ultra-Reliable Low-Latency Communications (URLLC) [1] is likely the most talked-about 5G use case mainly because of the huge services it can support
4) SIMULATION EXPERIMENTS To evaluate the efficiency of the proposed content caching algorithm, SELECTIVE HISTORIC LEAST FREQUENTLY USED (SHLFRU) has been compared with existing algorithms
It can be deduced that SHLFRU performs better than the compared algorithms
Summary
Ultra-Reliable Low-Latency Communications (URLLC) [1] is likely the most talked-about 5G use case mainly because of the huge services it can support. The aforementioned concerns of the caching algorithms have been addressed by proposing a three-fold algorithm solution to improve the cache hit ratio and access delay within a MEC environment. This includes a novel delay aware replacement caching algorithm that can find a victim in O(1), a proactive online association-based caching strategy that prefetches cache objects based on anticipated user behaviour and a MEC collaborative caching algorithm. Experimental results show that the proposed scheme outperforms conventional algorithms with regards to hit ratio It outperforms an offline caching algorithm with a pre-trained model.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
