TY - GEN
T1 - Content-aware user clustering and caching in wireless small cell networks
AU - Elbamby, Mohammed S.
AU - Bennis, Mehdi
AU - Saad, Walid
AU - Latva-Aho, Matti
N1 - Publisher Copyright:
© 2014 IEEE.
Copyright:
Copyright 2014 Elsevier B.V., All rights reserved.
PY - 2014/10/21
Y1 - 2014/10/21
N2 - In this paper, the problem of content-aware user clustering and content caching in wireless small cell networks is studied. In particular, a service delay minimization problem is formulated, aiming at optimally caching contents at the small cell base stations (SCBSs). To solve the optimization problem, we decouple it into two interrelated subproblems. First, a clustering algorithm is proposed grouping users with similar content popularity to associate similar users to the same SCBS, when possible. Second, a reinforcement learning algorithm is proposed to enable each SCBS to learn the popularity distribution of contents requested by its group of users and optimize its caching strategy accordingly. Simulation results show that by correlating the different popularity patterns of different users, the proposed scheme is able to minimize the service delay by %42 and 27%, while achieving a higher offloading gain of up to 280% and 90%, respectively, compared to random caching and unclustered learning schemes.
AB - In this paper, the problem of content-aware user clustering and content caching in wireless small cell networks is studied. In particular, a service delay minimization problem is formulated, aiming at optimally caching contents at the small cell base stations (SCBSs). To solve the optimization problem, we decouple it into two interrelated subproblems. First, a clustering algorithm is proposed grouping users with similar content popularity to associate similar users to the same SCBS, when possible. Second, a reinforcement learning algorithm is proposed to enable each SCBS to learn the popularity distribution of contents requested by its group of users and optimize its caching strategy accordingly. Simulation results show that by correlating the different popularity patterns of different users, the proposed scheme is able to minimize the service delay by %42 and 27%, while achieving a higher offloading gain of up to 280% and 90%, respectively, compared to random caching and unclustered learning schemes.
KW - caching
KW - clustering
KW - offloading
KW - reinforcement learning
KW - small cell networks
UR - http://www.scopus.com/inward/record.url?scp=84911959019&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84911959019&partnerID=8YFLogxK
U2 - 10.1109/ISWCS.2014.6933489
DO - 10.1109/ISWCS.2014.6933489
M3 - Conference contribution
AN - SCOPUS:84911959019
T3 - 2014 11th International Symposium on Wireless Communications Systems, ISWCS 2014 - Proceedings
SP - 945
EP - 949
BT - 2014 11th International Symposium on Wireless Communications Systems, ISWCS 2014 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2014 11th International Symposium on Wireless Communications Systems, ISWCS 2014
Y2 - 26 August 2014 through 29 August 2014
ER -