A Caching Strategy Towards Maximal D2D Assisted Offloading Gain

TL;DR

This paper develops a caching strategy to maximize D2D offloading gain by considering user collaboration and transmission conditions, providing analytical solutions and algorithms with verified effectiveness through simulations.

Contribution

It introduces a novel caching strategy considering user trust and physical conditions, with analytical derivations and algorithms for optimal offloading gain.

Findings

Analytical expression for D2D success probability derived.

Optimal caching strategy obtained in closed form for special cases.

Proposed algorithms outperform existing schemes in simulations.

Abstract

Device-to-Device (D2D) communications incorporated with content caching have been regarded as a promising way to offload the cellular traffic data. In this paper, the caching strategy is investigated to maximize the D2D offloading gain with the comprehensive consideration of user collaborative characteristics as well as the physical transmission conditions. Specifically, for a given content, the number of interested users in different groups is different, and users always ask the most trustworthy user in proximity for D2D transmissions. An analytical expression of the D2D success probability is first derived, which represents the probability that the received signal to interference ratio is no less than a given threshold. As the formulated problem is non-convex, the optimal caching strategy for the special unbiased case is derived in a closed form, and a numerical searching algorithm is proposed to obtain the globally optimal solution for the general case. To reduce the computational complexity, an iterative algorithm based on the asymptotic approximation of the D2D success probability is proposed to obtain the solution that satisfies the Karush-Kuhn-Tucker conditions. The simulation results verify the effectiveness of the analytical results and show that the proposed algorithm outperforms the existing schemes in terms of offloading gain.