🤖 AI Summary
In partially cooperative device-to-device (D2D) networks, existing coded caching schemes assume either known selfish user identities or operate only under high memory regimes, limiting practical applicability.
Method: This paper proposes the first universal coded caching mechanism applicable across the entire feasible memory regime. It requires no prior knowledge of the number or locations of selfish users and derives an information-theoretic lower bound on the transmission load; based on this bound, it constructs an optimal content delivery strategy.
Contribution/Results: The scheme achieves the theoretical minimum transmission load in the high-memory regime and significantly outperforms state-of-the-art schemes in medium- and low-memory regimes. Its core innovation lies in the first joint design that simultaneously handles *unknown selfishness* and *full-memory-range adaptability*, ensuring both robustness against user heterogeneity and asymptotic optimality. This work establishes a general theoretical framework and a practical solution for partially cooperative edge caching.
📝 Abstract
Device-to-device (D2D) communication is one of the most promising techniques for future wireless cellular communication systems. This paper considers coded caching in a partially cooperative wireless D2D network, where only a subset of users transmit during delivery, while all users request files. The non-transmitting users are referred to as selfish users. All existing schemes that do not require knowledge of the identity of selfish users before content placement are limited to the high-memory regime, particularly when the number of selfish users is large. We propose a novel coded caching scheme for a partially cooperative D2D network that operates in all feasible memory regimes, regardless of the number of selfish users. We also derive a lower bound on the transmission load of a partially cooperative D2D coded caching scheme. Using this bound, the proposed scheme is shown to be optimal in the high-memory regime.