Multi-Access Edge Computing Empowered Heterogeneous Networks: A Novel Architecture and Potential Works

June-Woo Ryu, Quoc-Viet Pham, Luan N. T. Huynh, Won-Joo Hwang, Jong-Deok Kim, Jung-Tae LeeSymmetry 2019 · DOI

TL;DR

This paper proposes a novel architecture for Multi-Access Edge Computing (MEC) integrated with Heterogeneous Networks (HetNets), specifically addressing the challenges of computation-intensive applications on mobile devices. The proposed architecture incorporates both wired and wireless backhaul, allows for flying base stations (UAVs) equipped with MEC servers, and aims to optimize resource allocation and task offloading in complex network environments.

What Problem Does This Paper Address?

Modern applications like online gaming, augmented reality, and intelligent video acceleration demand significant computational power and low latency, which current mobile devices often cannot provide due to limitations in battery life and processing capabilities. While Multi-Access Edge Computing (MEC) offers a solution by moving computation closer to users, and Heterogeneous Networks (HetNets) improve network capacity and coverage, their integration, especially in dynamic and complex scenarios, presents challenges. Specifically, existing research often simplifies network models (e.g., single MEC server, fixed user association) and overlooks the constraints of limited backhaul capacity, which is crucial for dense HetNets. This paper aims to bridge this gap by proposing a holistic architecture that addresses these complexities, including the use of UAVs for enhanced connectivity and computation.

Methods

The paper's primary contribution is the proposal of a novel architecture for MEC-empowered HetNets. This architecture is characterized by:

  • Integration of Wired and Wireless Backhaul: It supports both traditional wired connections and flexible wireless backhaul solutions for small cells, acknowledging the practical difficulties of deploying wired infrastructure everywhere.
  • UAV-Assisted MEC: Unmanned Aerial Vehicles (UAVs) are envisioned as flying MEC servers, capable of providing computation resources and enhancing connectivity, particularly in disaster-stricken areas or for temporary high-demand events.
  • Heterogeneous MEC Servers: The architecture considers MEC servers with varying sizes and computational capabilities, reflecting real-world deployments.
  • Multi-User and Multi-Server Considerations: It moves beyond single-user/single-server models to address scenarios with multiple users and multiple MEC servers, including those on UAVs.

The paper also provides an overview of computation offloading and resource allocation strategies in existing MEC systems, laying the groundwork for the proposed architecture. It discusses various techniques like fractional frequency reuse (FFR), power control, and 3D beamforming for interference management, and highlights the importance of joint optimization of offloading decisions and resource allocation.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • The Necessity of Integrated Architectures: The paper strongly argues that simply deploying MEC or HetNets individually is insufficient for future 5G networks. A holistic integration is required to meet the demands of advanced applications.
  • UAVs as a Game-Changer for Edge Computing: The inclusion of UAVs as mobile MEC servers offers a flexible and dynamic solution for extending coverage, providing computation in hard-to-reach areas, and responding to emergent needs.
  • Backhaul Constraints are Critical: The paper emphasizes that assuming infinite backhaul capacity is unrealistic and that efficient resource management must account for these limitations, especially in dense HetNets.
  • Joint Optimization is Key: The effectiveness of MEC in HetNets hinges on jointly optimizing multiple aspects, including interference management, user association, offloading decisions, and resource allocation, rather than treating them in isolation.
  • Heterogeneity is the Norm: Future networks will feature diverse MEC servers and user devices, requiring architectures that can adapt to these variations.

Key Results

The paper does not present empirical results from a deployed system or simulation. Instead, it focuses on:

  • Proposing a Novel Architecture: The core contribution is the conceptualization and detailed description of the MEC-empowered UAV-assisted HetNet architecture.
  • Identifying Key Challenges: The paper thoroughly outlines five major challenges that arise from this proposed architecture:
  • Joint Interference Management and Offloading Decision
    1. Joint User Association and Offloading Decision
    2. Hierarchical and Collaborative Computation Offloading
    3. Joint User Association and MEC Server Selection
    4. NOMA-Assisted MEC Networks
  • Highlighting Future Research Directions: By detailing these challenges, the paper implicitly sets the agenda for future research in this domain.

Limitations and Future Work

The paper acknowledges several limitations and points towards future research directions:

  • Complexity of Joint Optimization: The proposed architecture introduces significant complexity in jointly optimizing various parameters like interference management, offloading decisions, and resource allocation, especially in a distributed and dynamic environment.
  • UAV Limitations: UAVs have finite on-board energy, limiting their ability to process extremely energy-hungry tasks for extended periods.
  • Need for Advanced Algorithms: The paper suggests that conventional optimization techniques may be unsuitable for the dynamic nature of these networks, advocating for approaches like Markov decision processes and machine learning.
  • Specific Future Work Areas: The authors explicitly mention the need for further research in:
  • Content caching and delivery.
  • Implementation of real-world applications (e.g., AR/VR).
  • Optimal deployment strategies for MEC servers.
  • Network planning for UAV-mounted MEC servers (e.g., determining the number of UAVs and their placement).
  • Efficient resource allocation in NOMA-enabled MEC networks.

Summary updated

BibTeX
@article{ryu2019multi,
  author = {June-Woo Ryu and Quoc-Viet Pham and Luan N. T. Huynh and Won-Joo Hwang and Jong-Deok Kim and Jung-Tae Lee},
  title = {{Multi-Access Edge Computing Empowered Heterogeneous Networks: A Novel Architecture and Potential Works}},
  journal = {Symmetry},
  year = {2019},
  doi = {10.3390/sym11070842},
}
Figure