Smart Fog : Advanced Fog Server-centric Things Abstraction Framework for Multi-service IoT System

Gyeonghwan Hong, Eunsoo Park, Sihoon Choi, Dongkun ShinJournal of KIISE 2016 · DOI

TL;DR

This paper introduces "Smart Fog," a novel fog server-centric framework for multi-service IoT systems. It addresses limitations of existing approaches by providing a unified framework that enables real-time interaction, reduces network traffic, and conserves power on mobile devices.

What Problem Does This Paper Address?

The proliferation of multi-service IoT systems, where multiple services share the same physical devices, has led to the development of various "things abstraction frameworks." However, existing frameworks face significant challenges:

  • Distributed frameworks: Suffer from IoT service duplication, leading to increased power consumption and network traffic on mobile devices.
  • Cloud server-centric frameworks: While avoiding duplication, they struggle with real-time interactions due to high network latency.
  • Existing fog server-centric frameworks: While offering a balance, they often use incomplete interfaces, leading to a strong dependency between IoT services and physical devices.

This paper aims to bridge this gap by proposing a fog server-centric framework that overcomes these limitations, enabling efficient and responsive multi-service IoT deployments.

Methods

The core of the Smart Fog framework is a Smart Gateway that acts as a fog server. This gateway is equipped with a software stack built upon the IoTivity framework and adheres to the OIC standard. The novelty lies in the definition and implementation of three key IoT interfaces:

  1. Things Abstraction API: This API allows mobile applications to interact with the Smart Gateway, enabling them to access sensor data, control actuators, and install/manage IoT services.
  2. Physical Thing Interface: This interface defines a minimal set of common functions for device firmware to communicate with the Smart Gateway. This standardization allows the gateway to manage sensor data sampling and transmission methods for various devices.
  3. Logical Thing Interface: This interface allows IoT services running on the Smart Gateway to represent themselves as logical sensors or actuators. This enables services to process and synthesize data from multiple physical sensors or combine actions from multiple actuators, fostering more complex and intelligent IoT applications.

The Smart Gateway itself comprises several software modules: a Thing Arbiter for device discovery and data routing, a Thing Directory for managing device information, a Sensor Database for storing sensor data, and a Sensor Notifier for delivering data to mobile devices.

Datasets

The paper does not explicitly mention the use of pre-existing datasets for training. Instead, it focuses on a prototype implementation and experimental evaluation using a simulated environment. For their experiments, they created a simulated smart home environment with:

  • 32 simulated physical sensor devices connected to the local network.
  • 9 IoT services running on the Smart Gateway.

They also implemented a distributed things abstraction framework prototype for comparative analysis.

Key Insights and Ideas

  • Fog as the sweet spot: The research strongly suggests that a fog server-centric approach offers the optimal balance for multi-service IoT, mitigating the drawbacks of purely distributed or cloud-centric models.
  • Unified interfaces are crucial: The paper highlights that defining comprehensive interfaces (abstraction API, physical, and logical) is key to reducing dependencies between different components in an IoT system, leading to greater flexibility and scalability.
  • Smart Gateway's central role: The Smart Gateway acts as more than just a router; it's an intelligent hub that hosts services, manages devices, and orchestrates data flow, significantly offloading processing from mobile devices.
  • Real-time interaction is achievable with fog: By bringing computation closer to the edge, Smart Fog demonstrates that low-latency, real-time interactions are feasible, opening doors for more responsive IoT applications.

Key Results

The experimental results demonstrate the effectiveness of Smart Fog:

  • Reduced Network Latency: Smart Fog exhibits significantly lower network latency compared to cloud server-centric approaches. A simple HTTP request took an average of 15ms on Smart Fog versus 378ms on Amazon EC2.
  • Power Consumption Reduction: Compared to a distributed framework, Smart Fog reduced power consumption on mobile devices by 21%.
  • Network Traffic Reduction: Smart Fog achieved a substantial reduction in network traffic, with received data decreasing by approximately 85% and transmitted data by 47%, leading to an overall network traffic reduction of about 74%.
  • CPU Load Reduction: The CPU core load on mobile devices was reduced from 20.69% in the distributed approach to 6.76% with Smart Fog.

Limitations and Future Work

The authors acknowledge a potential limitation:

  • Increased device power consumption: While mobile devices benefit from reduced load, increased sensor data sampling and frequent data transmission for real-time interactions could lead to higher power consumption on the IoT devices themselves.

Future work suggested by the authors includes:

  • Optimizing sensor data sampling and transmission: Research into more efficient methods for data acquisition and transfer from IoT devices to further mitigate power consumption and network traffic.

Summary updated

Abstract

최근 여러 사물 인터넷 서비스가 사물 장치를 공유하는 다중 서비스 시스템을 구현하기 위해, 다양한 구조의 사물 추상화 프레임워크들이 제시되었다. 분산형 구조는 사물 인터넷 서비스 중복 문제가 있으며, 클라우드 서버 중심 구조는 실시간 인터랙션을 할 수 없다. 또한, 기존의 포그 서버 중심 구조에서는 불완전한 인터페이스가 사용되었다. 본 논문에서는 기존 구조의 문제를 해결한 사물 추상화 프레임워크인 Smart Fog를 제안하였다. Smart Fog는 스마트 게이트웨이와 3개의 IoT 인터페이스들로 구성된다. Smart Fog는 IoTivity와 OIC 표준을 기반으로 구현되었고, 이를 이용하여 실제 임베디드 장치인 Odroid-XU3에서 프로토타입을 구현하였다. 프로토타입 상에서 실험한 결과, Smart Fog가 실시간 인터랙션이 가능할 정도로 네트워크 지연 시간이 짧고, 분산형 구조에 비해 모바일 장치에서 발생하는 네트워크 트래픽이 74%, 전력 소모가 21% 감소함을 확인하였다.
BibTeX
@article{hong2016smart,
  author = {Gyeonghwan Hong and Eunsoo Park and Sihoon Choi and Dongkun Shin},
  title = {{Smart Fog : Advanced Fog Server-centric Things Abstraction Framework for Multi-service IoT System}},
  journal = {Journal of KIISE},
  year = {2016},
  doi = {10.5626/jok.2016.43.6.710},
}
Figure