Content-Centric Mobile Edge Caching

Tiankui Zhang, Xinyuan Fang, Yuanwei Liu, Arumugam Nallanathan

Introduction

As global mobile data traffic approaches the zettabyte era, traditional "host-centric" IP-based networking is struggling to keep pace. In current architectures, content is often fetched repeatedly from distant servers, creating a massive "backhaul bottleneck" between base stations and the core internet. Edge caching—storing popular content at the network's periphery—offers a solution to boost spectral efficiency and reduce latency. However, implementing this effectively requires a shift toward Content-Centric Networking (CCN), where data is identified by name rather than location.

An Integrated CCN Framework for 5G

The proposed framework integrates CCN functionality directly into the 5G protocol stack. By embedding these protocols within Baseband Unit (BBU) pools and User Terminals (UTs), the network can locate the nearest copy of a requested file, whether it resides at a base station or on a neighboring device.

The proposed 5G network architecture integrating CCN for edge caching across BBU pools and user terminals.

This architecture distinguishes between "ideal" and "non-ideal" fronthaul scenarios. In ideal scenarios, the BBU pool manages the Radio Resource Head (RRH) centrally. In non-ideal scenarios, Distributed Units (DUs) are placed closer to the RRHs to handle time-sensitive processing, while caching remains coordinated at the BBU or UT level.

CCN-Based Functional Entities

Central to this framework is the redesign of the BBU pool to act as a content-aware hub. The BBU pool is equipped with two primary storage components:

  1. Local Cache Store (L-CS): A repository for frequently accessed content within the CCN Core.
  2. Extended Cache Store (E-CS): A directory that tracks content stored across all user devices within the cell, enabling efficient Device-to-Device (D2D) sharing.
Detailed functional architecture of the BBU pool, highlighting the CCN Core and the Extended Cache Store (E-CS).

The CCN Core includes standard components like the Forwarding Information Base (FIB) for request routing and the Pending Interest Table (PIT) for tracking unsatisfied requests. This setup allows the BBU to act as a "matchmaker," establishing low-latency D2D links between users without involving the core network.

Incentivizing the Edge: Auction-Based Caching (ABC)

A significant hurdle in edge caching is "user selfishness"—users may be reluctant to use their own device's battery and storage to benefit others. To address this, the authors propose an Auction-Based Caching (ABC) mechanism modeled as a second-price sealed-bid auction.

In this model, the Base Station auctions "caching tasks" to User Terminals. Users bid based on their available resources and preferences. This ensures "truthfulness," where users bid their actual costs, optimizing the overall social welfare of the cell. Because the resulting optimization problem is NP-hard, the framework employs Semidefinite Programming (SDP) relaxation to find near-optimal content placement efficiently.

Performance and Insights

Simulations based on 3GPP Technical Reports and Zipf-like content popularity distributions demonstrate the framework's efficacy. Key results include:

  • Reduced Latency: The ABC method significantly outperformed random caching and social-aware benchmarks, reducing average peak-period delays from 6.3 seconds to much lower levels.
  • Traffic Offloading: The strategy consistently achieved higher backhaul offloading ratios by satisfying more requests locally.
  • Content Diversity: By utilizing a "cache conflict" matrix, the system prevents nearby users from caching identical files, effectively turning a cluster of devices into a distributed neighborhood library.

Conclusion and Future Directions

Content-centric edge caching represents a vital step toward scalable 5G and 6G networks. While the current framework addresses protocol efficiency and user incentives, future work must tackle the challenges of high-speed mobility and the specific energy costs of D2D sharing. Furthermore, integrating big data for proactive caching—predicting user needs before they arise—could further eliminate traffic spikes and enhance the mobile user experience.

Summary updated

Abstract

With the explosion of data volume, it becomes challenging to deliver high quality service to mobile users. Therefore, edge caching has received significant attentions since it can bring contents near mobile users, to boost spectral efficiency, and reduce backhaul load of mobile networks. Due to limited storage resources within mobile networks, it is important to improve efficiency of content management in edge caching. In this article, we propose an integrated content-centric mobile network framework for edge caching in 5G networks, which can leverage content-centric networking (CCN), achieve content-oriented information management, and increase content delivery efficiency. We elaborately design the cache-enabled mobile network architecture, CCN based function entities, CCN embedded protocol stack, and content retrieval process, and develop several effective approaches for tackling practical implementation constraints of CCN based edge caching. We demonstrate that our content caching strategies can significantly enhance edge caching performance. To further improve the performance of content-centric mobile edge caching, we identify promising open research directions.
BibTeX
@inproceedings{zhang2019content,
  author = {Tiankui Zhang and Xinyuan Fang and Yuanwei Liu and Arumugam Nallanathan},
  title = {{Content-Centric Mobile Edge Caching}},
  booktitle = {IEEE Access},
  year = {2019},
  doi = {10.1109/access.2019.2962856},
}
Figure