Decentralized collaborative video caching in edge-caching cellular networks

Mahboob, Shadab
Thumbnail Image
Other Contributors
Abouzeid, Alhussein A.
Chen, Tianyi
Chakareski, Jacob
Kar, Koushik
Issue Date
Electrical engineering
Terms of Use
This electronic version is a licensed copy owned by Rensselaer Polytechnic Institute (RPI), Troy, NY. Copyright of original work retained by author.
Full Citation
In this thesis, the problem of video caching across a set of small-cell base stations (SBS) in edge-caching cellular networks is considered. The SBSs are connected to each other over a high-capacity short-delay back-haul link and linked to a remote server over a long-delay connection. The problem of minimizing the overall video delivery delay involves concave minimization along with packing type integrality constraints and is in general NP-hard. Nevertheless, our proposed Collaborative Caching Algorithm (CCA) can efficiently compute a close to the optimal solution, where the degree of sub-optimality depends on the worst case video-to-cache size ratio. The algorithm runs in $O(NK + K \log K)$ time, where $N$ is the number of SBSs (caches) and $K$ is the maximum number of videos. The algorithm is naturally amenable to a distributed implementation that requires minimal information exchange between the SBSs regarding the video requests from the users and runs in $O(N + K \log K)$ time. Simulations on real video access traces demonstrate the fact that CCA closely approaches the optimal solution as the cache size with respect to video sizes becomes large, which agrees with our theoretical results. We also do an online implementation of a practical environment where the video popularities are not known a priori but are estimated over time through a limited amount of periodic information exchange between the SBSs. It shows that CCA effectively uses the SBS caches to reduce the video delivery delay and conserve the remote server's bandwidth at the expense of low-cost local bandwidth. The effect of change in system parameters (weighting parameter of popularity estimation and window length of requests) on the performance of CCA is also analyzed, which helps to find the optimal combinations of these parameters that minimize bandwidth consumption and average delay. Finally, we show that CCA significantly outperforms two reference caching methods, LRU and LFU, in terms of average delay and remote bandwidth. This thesis concludes by mentioning possible future extensions to user-SBS association problems, systems with local and remote delay as functions of video properties, and adaptive window adjustments in dynamic implementation of the algorithm.
December 2021
School of Engineering
Dept. of Electrical, Computer, and Systems Engineering
Rensselaer Polytechnic Institute, Troy, NY
Rensselaer Theses and Dissertations Online Collection
Restricted to current Rensselaer faculty, staff and students in accordance with the Rensselaer Standard license. Access inquiries may be directed to the Rensselaer Libraries.