[Paper Review] A Robust Dynamic Edge Network Architecture for the Internet-of-Things

Beatriz Lorenzo, Juan García‐Rois+3 more|arXiv (Cornell University)|Oct 13, 2017

Advanced MIMO Systems Optimization15 references3 citations

TL;DR

This paper proposes a Robust Dynamic Edge Network Architecture (RDNA) for IoT that leverages mobile devices' computing, storage, and communication capabilities to create decentralized, dynamic edge networks, reducing backhaul congestion and improving latency, reliability, and energy efficiency. The architecture integrates multi-layer solutions across physical, access, networking, application, and business layers to enhance robustness in highly dynamic IoT environments with high device density.

ABSTRACT

A massive number of devices are expected to fulfill the missions of sensing, processing and control in cyber-physical Internet-of-Things (IoT) systems with new applications and connectivity requirements. In this context, scarce spectrum resources must accommodate a high traffic volume with stringent requirements of low latency, high reliability and energy efficiency. Conventional centralized network architectures may not be able to fulfill these requirements due to congestion in backhaul links. This article presents a novel design of a robust dynamic edge network architecture (RDNA) for IoT which leverages the latest advances of mobile devices (e.g., their capability to act as access points, storing and computing capabilities) to dynamically harvest unused resources and mitigate network congestion. However, traffic dynamics may compromise the availability of terminal access points and channels and, thus, network connectivity. The proposed design embraces solutions at physical, access, networking, application, and business layers to improve network robustness. The high density of mobile devices provides alternatives for close connectivity which reduces interference and latency, and thus, increases reliability and energy efficiency. Moreover, the computing capabilities of mobile devices project smartness onto the edge which is desirable for autonomous and intelligent decision making. A case study is included to illustrate the performance of RDNA. Potential applications of this architecture in the context of IoT are outlined. Finally, some challenges for future research are presented.

Motivation & Objective

• To address the limitations of centralized IoT network architectures in handling high traffic volume, low latency, and energy efficiency requirements.
• To mitigate backhaul congestion caused by massive IoT device connectivity through decentralized, device-centric networking.
• To enhance network robustness in dynamic IoT environments where terminal access points and channels may become unavailable.
• To exploit the computing, storage, and communication capabilities of mobile devices to enable smart, localized decision-making at the network edge.
• To design a holistic, multi-layered architecture spanning physical, access, networking, application, and business layers for improved resilience and performance.

Proposed method

• Utilizes mobile devices as dynamic edge nodes capable of acting as access points, storing data, and performing local computation.
• Employs dynamic resource harvesting to exploit unused spectrum, processing, and storage capacity in mobile devices.
• Integrates cross-layer design across physical, MAC, network, application, and business layers to ensure end-to-end robustness.
• Enables close-proximity connectivity among devices to reduce interference and propagation delay.
• Leverages device intelligence for autonomous, real-time decision-making and adaptive network reconfiguration.
• Employs a case study to evaluate performance under realistic IoT traffic and mobility dynamics.

Experimental results

Research questions

• RQ1 How can a decentralized edge network architecture dynamically utilize underutilized resources in mobile devices to reduce backhaul congestion in IoT systems?
• RQ2 What multi-layer architectural design is required to ensure robustness in highly dynamic IoT environments with frequent topology changes?
• RQ3 To what extent can device-centric edge computing improve latency, reliability, and energy efficiency in dense IoT deployments?
• RQ4 How do cross-layer interactions across physical, access, network, application, and business layers enhance overall network resilience?
• RQ5 What are the key performance trade-offs and limitations of relying on mobile devices as edge infrastructure in IoT?

Key findings

• The RDNA architecture significantly reduces backhaul load by offloading traffic to mobile devices acting as edge nodes.
• Proximity-based connectivity among devices leads to lower latency and reduced interference compared to traditional centralized backhaul models.
• The integration of multi-layer solutions enhances network robustness, particularly in scenarios with fluctuating access point availability.
• Mobile devices’ computing capabilities enable localized, intelligent decision-making, supporting real-time control and processing in IoT applications.
• The case study demonstrates improved reliability and energy efficiency under dynamic traffic and mobility conditions.
• The architecture shows strong potential for scalable deployment in smart cities, industrial IoT, and other high-density IoT environments.