Understanding information centric layer of adaptive collaborative caching framework in mobile disconnection-prone networks

Smart networks and services leverage in-network caching to improve transmission efficiency and support large amount of content sharing, decrease high operating costs and handle disconnections. In this paper, we investigate the complex challenges related to content popularity weighting process in collaborative caching algorithm in heterogeneous mobile disconnection prone environments. We describe a reputation-based popularity weighting mechanism built in information-centric layer of our adaptive collaborative caching framework CafRepCache which considers a realistic case where caching points gathering content popularity observed by nodes differentiates between them according to node's reputation and network's connectivity. We extensively evaluate CafRepCache with competitive protocols across three heterogeneous real-world mobility, connectivity traces and use YouTube dataset for different workload and content popularity patterns. We show that our collaborative caching mechanism CafRepCache balances the trade-off that achieves higher cache hit ratio, efficiency and success ratios while keeping lower delays, packet loss and caching footprint compared to competing protocols across three traces in the face of dynamic mobility of publishers and subscribers