WE1.R3.2

On Demand-Private Hotplug Caching Systems

Yinbin Ma, Daniela Tuninetti, University of Illinois Chicago, United States

Session:
Coded Caching: Privacy and Security

Track:
14: Secure Communication and Computation

Location:
Ypsilon IV-V-VI

Presentation Time:
Wed, 10 Jul, 10:10 - 10:30

Session Chair:
Daniela Tuninetti, University of Illinois Chicago
Abstract
Maddah-Ali and Niesen (MAN) showed that coded caching can effectively mitigate peak-time network load by leveraging local caches at the end users. In the MAN model, a server stores multiple files, populates the local caches, and serves single-file requests from cache-aided users via an error-free shared link. The delivered multicast messages in the MAN scheme to serve the users' requests require all users to stay active and expose demand information. To preserve demand privacy and accommodate for offline users, the hotplug caching model with demand privacy against colluding users was introduced and an achievable schemes based on the "privacy keys" idea was proposed at ISIT 2023. This ISIT 2024 paper proposes three new achievable schemes for this model which use Maximum Distance Separable codes coupled with the "virtual users" idea. These new schemes turn out to be exactly optimal under certain conditions, and otherwise optimal to within a constant factor.
Resources