Source-linked AI summary
Fog Computing: Focusing on Mobile Users at the Edge
Tom H. Luan, Longxiang Gao, Zhi Li, Yang Xiang, Guiyi Wei, Limin Sun
TL;DR
Mobile networks must support growing, diverse, location-dependent applications despite mobile devices’ limited resources and costly long-distance cloud connections. The article presents Fog computing as a networking-oriented extension of cloud computing that places virtualized resources near users, and describes its architecture, benefits, applications, and research issues.
Problem
Mobile users need diverse, location-aware, real-time services, while resource limits and long-thin connections make remote cloud access costly and difficult to manage.
Method
The article defines and analyzes Fog computing from a networking perspective, covering its Mobile-Fog-Cloud architecture, localized resources, applications, design goals, and open issues.
Results
Fog computing provides local, cloud-like storage, computing, and communication resources that improve mobile service delivery and network efficiency.
Takeaways & Limitations
Fog computing is presented as a scalable approach for localized mobile and IoT services, with three-dimensional resource allocation as a key research focus.
Abstract
from arXiv · showhide
With smart devices, particular smartphones, becoming our everyday companions, the ubiquitous mobile Internet and computing applications pervade people daily lives. With the surge demand on high-quality mobile services at anywhere, how to address the ubiquitous user demand and accommodate the explosive growth of mobile traffics is the key issue of the next generation mobile networks. The Fog computing is a promising solution towards this goal. Fog computing extends cloud computing by providing virtualized resources and engaged location-based services to the edge of the mobile networks so as to better serve mobile traffics. Therefore, Fog computing is a lubricant of the combination of cloud computing and mobile applications. In this article, we outline the main features of Fog computing and describe its concept, architecture and design goals. Lastly, we discuss some of the future research issues from the networking perspective.
I. INTRODUCTION
Mobile networking must integrate cloud infrastructure with rapidly expanding, diverse, and location-dependent mobile services while meeting real-time demands and constrained connectivity. Fog computing is introduced as a networking-oriented approach that places cloud-like resources near users and identifies architecture, goals, and open issues.
- Motivation: Cloud computing underpins Internet services, while mobile devices have become major outlets for diverse applications and growing traffic.The passage reports that around 90% of global Internet users rely on cloud services and that smartphones have surpassed PCs in worldwide shipments since 2011.
- Motivation: Mobile users request services adapted to their locations and environments, creating demands that centralized cloud systems can hardly manage at scale.Examples include ehealthcare and Internet-of-Things applications requested by nomadic users.
- Motivation: Resource-constrained mobile applications outsource computation to the cloud but expect real-time responses, especially for emerging wearable devices.Google Glasses and Microsoft HoloLens are cited as devices relying intensively on cloud-based sensing and processing.
- Motivation: Long-thin connections between mobile users and remote clouds impose bandwidth and cost barriers to high-rate data exchange.Fog computing is presented as a more practical solution for converging cloud and mobile applications in content delivery and real-time processing.
- Article scope: The article presents Fog computing from a networking perspective, outlining its features, design goals, architecture, application scenarios, comparisons, and open research issues.The paper notes that Fog computing is new and lacks a standardized definition.
II. SYSTEM ARCHITECTURE
Fog computing extends cloud computing with a geo-distributed intermediate layer of virtualized servers near mobile users. These servers provide local services while bridging mobile devices, peer Fog servers, and the cloud.
- Architecture: The Mobile-Fog-Cloud hierarchy inserts an intermediate Fog layer between mobile devices and the cloud.The architecture is described as a three-layer hierarchy.
- Architecture: Fog servers are geo-distributed at local premises and provide storage, computing, and communication resources through short-distance high-rate wireless connections.They may be deployed in parks, bus terminals, shopping centers, or on moving vehicles.
- Deployment: Fog servers can be adapted from cellular base stations, WiFi access points, or femtocell routers by upgrading computing and storage resources.The wireless interfaces can be reused in these deployments.
- Connectivity: A Fog server bridges users and cloud through direct wireless access, peer-Fog connections, and cloud connectivity.It can independently provide predefined services within its wireless coverage.
- Concept: The paper summarizes Fog computing as a virtualized cloud-like device deployed close to mobile users: “the cloud close to the ground.”The architecture is intended to support fast-rate local services and later discussion of application scenarios.
III. WHY FOG? PUT SERVICE CLOSE TO CONSUMER
Fog computing places contents and application services near consumers to address cloud computing’s lack of location awareness for mobile users. Localized delivery improves user service quality while reducing network traffic and resource costs.
- Location-aware services: Mobile users’ service demands depend on their locations, making globally pooled cloud information less suitable for localized needs.Shopping-center flyers and city-specific information are examples of location-dependent content.
- Location-aware services: Fog servers can pre-cache site-specific content, allowing mobile users to retrieve localized services through high-rate local connections without searching the cloud.The shopping-center example contrasts cloud retrieval with Fog retrieval.
- User benefits: Fog computing increases data rates and reduces service latency, response time, and bandwidth cost compared with cloud delivery.The benefits arise from local connections that avoid traversing the backbone network.
- Network benefits: Fog computing avoids back-and-forth traffic between cloud and mobile users, saving backbone bandwidth and reducing core-network energy consumption and carbon footprint.The article frames this reduction as supporting sustainable networking.
IV. EXEMPLARY APPLICATIONS
The paper illustrates Fog computing through localized deployments that pre-cache content, integrate sensors, and provide services tailored to shopping centers, parks, and inter-state buses.
- Shopping center: Shopping-center Fog servers can pre-cache floor-specific layouts and advertisements while supporting indoor navigation, ad distribution, and feedback collection.Servers may be deployed at different levels and collectively form an integrated localized information system.
- Scenic park: Park deployments can provide maps, tourist guides, navigation, alerts, and environmental monitoring through Fog servers connected with sensor networks.Servers may be placed at entrances and important locations inside the park.
- Inter-state bus: Greyhound’s on-board “BLUE” system uses a bus Fog server for video streaming, gaming, and social networking over WiFi.The server connects to the cloud through cellular networks to update cached content and can process passenger utility data.
V. COMPARISON TO CLOUD COMPUTING
Fog computing differs from cloud computing by extending resources to users’ premises and emphasizing localized, location-aware services. Its distributed deployment can also support collaborative content distribution across independently owned Fog servers.
- Fog computing extends cloud services to users’ premises, emphasizing localized applications and location awareness rather than a centralized global information portal.
- Fog servers in different locations may be deployed by separate operators and owners, forming a collaborative Fog computing system across a wider region.
- Independently owned Fog servers can operate as an integrated content distribution network for disseminating media contents across a city.
VI. DESIGN FROM STORAGE, COMPUTING AND COMMUNICATION
Fog computing combines storage, computing, and communication capabilities at localized Fog servers. Its design exploits predictable mobile demand and location-specific interests to provide proactive, intelligent services beyond caching alone.
- A Fog server is generic virtualized equipment with onboard storage, computing, and communication capabilities, enabling intelligent and adaptive services.
- Fog servers can predict local users’ information demands and proactively pre-cache desirable contents, such as airport flight and transportation information.
- Unlike CDN servers serving broad and less predictable interests, Fog servers target precise service regions with clearer users and demands.
- Unlike ICN cache servers, Fog servers also provide computing infrastructure for real-time processing and applications including IoT, vehicular communications, and smart grids.
- Mobile information-demand patterns can be predictable enough for proactive caching, and this framework can be applied within the broader Fog computing paradigm.
B. Computing
Fog computing provides a generic edge platform that serves localized information applications and computation requests. Fog servers can process local requests while bridging mobile data collection with cloud-based analysis.
- Fog servers act as intelligent computing systems that autonomously serve local computation and data-processing requests from mobile users.
- Fog servers can collect environmental or utility data from users, transmit it to the cloud for analysis, and provide results to third parties.
- Cloudlets similarly process local mobile-device requests in real time, while their primary focus is providing computing services.
- Fog computing is a generic edge-computing platform focused on localized information applications and computation requests.
- Fog servers’ networking role supports their broader computing function within the Fog architecture.
VII. RESEARCH ISSUES FROM NETWORKING PERSPECTIVE
The networking design of Fog computing uses three connection types around geo-distributed Fog servers. This architecture enables local wireless service, Fog-to-Fog coordination, and cloud-based control and content management.
- Networking architecture: Each Fog server maintains wireless connections to local mobile users, connections with peer Fog servers, and connections to the cloud.
- A. Communications between Mobile and Fog: Fog servers may use WiFi and Bluetooth for mobile access while exploiting application and location awareness for wireless resource allocation.
- A. Communications between Mobile and Fog: Because Fog servers manage both applications and wireless communications, they can support cross-layer design for improved service quality.
- A. Communications between Mobile and Fog: Location-specific deployments can expose predictable user features and service demands, such as differing mobility and traffic patterns across sites.
- Networking architecture: Networking connections in Fog computing link mobile users, Fog servers, peer Fog servers, and the cloud.
- B. Communications between Fog and Cloud: The cloud coordinates geo-distributed Fog servers and serves as the central information depot supplying cached content.
- B. Communications between Fog and Cloud: Cloud applications and contents are synchronized with selectively provisioned localized applications at Fog servers.
- B. Communications between Fog and Cloud: An SDN approach can separate control and data planes so the cloud establishes routes for updating distributed Fog servers.
C. Communications between Fogs
Fog servers can collaborate across locations, but communication must accommodate heterogeneous ownership policies, varied topologies, and wired or opportunistic wireless connections.
- Inter-Fog Collaboration: Fog servers can pool local resources to support collaborative service provision and content delivery among peered servers.Routing may use centralized SDN-based control or fully distributed mechanisms such as OSPF.
- Communication Challenges: Different Fog-server owners may impose heterogeneous service policies that inter-Fog routing must address.
- Communication Challenges: Co-located Fog servers may use high-rate, low-cost connections through the same Internet service provider to improve collaboration and reduce cloud–Fog traffic.The passage associates this topology with saved bandwidth cost and enhanced data rate.
- Communication Challenges: Inter-Fog data routing must account for wired Internet links and wireless opportunistic connections, including vehicular contacts for content shipment.
D. Challenges of Fog Computing Deployment
Deploying Fog computing requires operators to manage localized applications, forecast and provision resources, and determine server placement while absorbing additional management and maintenance costs.
- Deployment Challenges: Fog deployment adds computing and storage resources at the edge to process localized requests quickly using local resources and connections.The distributed deployment also creates extra management and maintenance cost because services must adapt across locations.
- Application: Operators must customize the applications embedded in each Fog server according to demand at its specific location.
- Scaling: Operators must anticipate demand at each Fog server and deploy adequate Fog resources for sufficient provisioning.
- Placement: Optimal Fog-server placement is challenging because user demands differ across locations, although groups of servers can collaborate nearby.The paper gives multi-server deployment inside a shopping center as an example.
E. Incorporating with Emerging Technologies
Fog computing is presented as a virtualized, location-focused layer that can operate with 5G and complement NFV and SDN in connecting cloud resources with localized mobile services.
- Architecture Integration: Fog computing is described as a layered architecture incorporating emerging networking technologies.The section specifically discusses 5G, NFV, and SDN relationships.
- 5G Technologies: Existing access networks such as WiFi and emerging 5G wireless technologies can support virtualized Fog layers for customized location-based applications.
- Network Function Virtualization: Unlike NFV’s virtualized network functions inside nodes, Fog computing virtualizes location-based applications at edge devices for localized mobile users.
- Software-defined Networking: SDN can let the cloud manage the entire network using a global network view while Fog servers synchronize data updates with the cloud.
- Conclusion: Fog computing uses nearby reserved compute and communication resources to absorb intensive mobile traffic through local fast-rate connections.The article connects this design with improved perceived service quality and reduced backbone bandwidth cost and energy consumption.
- Conclusion: Three-dimensional service-oriented resource allocation is identified as a key issue, alongside research opportunities in network management, traffic engineering, big data, and service delivery.