.png){kind=icon}
.png){kind=icon}
Return
A Low-Complexity User-Preference-Aware Decentralized Coded Caching Based on User Pairing
.png){kind=icon}
PRE
AI
DOI：10.1109/JIOT.2024.3425637.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Decentralized coded caching (DCC) is promising to relieve the load pressure of the networks (i.e., reducing the delivery rate) by creating multicast opportunities for a group of users. However, DCC suffers from high complexity (i.e., exponential of user number) because multicast opportunities are obtained by traversing all the possible user subsets. Considering individual user preferences, this article proposes a low-complexity user-pair-based modified DCC scheme (UP-MDCC). It only traverses user subsets with two users (i.e., user-pair) to generate coded subpackages, while obtaining similar delivery rate performance as DCC. To achieve this, we first update the traditional DCC to modified DCC (MDCC). Different to DCC which caches all N contents uniformly, MDCC can choose the favorite $N_{s}$ contents to cache, such that more caching capacity can be allocated to the contents with higher requesting probability. Then, given MDCC, the relationship between the delivery rate and the number of cached contents is derived. It is revealed that when the favorite contents are entirely cached for each user, more than 99% of the delivery rate gain is generated by the user subsets containing only two users. Therefore, a low-complexity UP-MDCC can be proposed by limiting MDCC to traverse only the user subsets containing two users. Moreover, to find the optimal user-pairs to provide the minimum delivery rate for UP-MDCC, a graph-partitioning-based user-pairing strategy (GP-UPS) is proposed. Simulations verify that with GP-UPS, UP-MDCC can achieve similar delivery rate as that of uncoded placement absolutely fair (UPAF) caching and MDCC with a gap of less than 1%. Moreover, compared to DCC, the number of user subsets traversed is significantly reduced in UP-MDCC and the complexity is reduced from exponential to square of user number.
Keywords:
Unicast
Internet of Things
Servers
Probability distribution
Computational complexity
Tin
Simulation
Coded caching
computational complexity
delivery rate
individual user preferences
multicast
Journal
IF:
8.9
Papers:
1.4W
Citations:
7.8W