Saved in:
Bibliographic Details
Main Authors: Brembilla, Niccolò, Ma, Yinbin, Belotti, Pietro, Malucelli, Federico, Tuninetti, Daniela
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2511.19639
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912727437410304
author Brembilla, Niccolò
Ma, Yinbin
Belotti, Pietro
Malucelli, Federico
Tuninetti, Daniela
author_facet Brembilla, Niccolò
Ma, Yinbin
Belotti, Pietro
Malucelli, Federico
Tuninetti, Daniela
contents Inspired by prior work by Tian and by Cao and Xu, this paper presents an efficient computer-aided framework to characterize the fundamental limits of coded caching systems under the constraint of linear coding. The proposed framework considers non-Shannon-type inequalities which are valid for representable polymatroids (and hence for linear codes), and leverages symmetric structure and problem-specific constraints of coded caching to reduce the complexity of the linear program. The derived converse bounds are tighter compared to previous known analytic methods, and prove the optimality of some achievable memory-load tradeoff points under the constraint of linear coding placement and delivery. These results seem to indicate that small, structured demand subsets combined with minimal common information constructions may be sufficient to characterize optimal tradeoffs under linear coding.
format Preprint
id arxiv_https___arxiv_org_abs_2511_19639
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Computer-aided Characterization of Fundamental Limits of Coded Caching with Linear Coding
Brembilla, Niccolò
Ma, Yinbin
Belotti, Pietro
Malucelli, Federico
Tuninetti, Daniela
Information Theory
Inspired by prior work by Tian and by Cao and Xu, this paper presents an efficient computer-aided framework to characterize the fundamental limits of coded caching systems under the constraint of linear coding. The proposed framework considers non-Shannon-type inequalities which are valid for representable polymatroids (and hence for linear codes), and leverages symmetric structure and problem-specific constraints of coded caching to reduce the complexity of the linear program. The derived converse bounds are tighter compared to previous known analytic methods, and prove the optimality of some achievable memory-load tradeoff points under the constraint of linear coding placement and delivery. These results seem to indicate that small, structured demand subsets combined with minimal common information constructions may be sufficient to characterize optimal tradeoffs under linear coding.
title Computer-aided Characterization of Fundamental Limits of Coded Caching with Linear Coding
topic Information Theory
url https://arxiv.org/abs/2511.19639