FOG Computing

Dr. Risala Tasin Khan

Professor
IIT, JU
Introduction
• Fog computing is an emerging technology utilized for the Internet of Things (IoT), a
paradigm that extends cloud computing and services to the edge of the network .
• Fog computing provides data, computation, storage and applications to the end-users
similar to cloud computing.
• What distinguishes Fog computing from Cloud computing is close proximity to
end-users, support for mobility and dense geographical distribution
• Cloud computing model makes it efficient for customers and could help to meet and scale
the IoT requirements by initialising on-demand scalability of computation power, storage
and resources along with “pay-as-you-go”
• However, the delay in transferring data between cloud and end-devices, especially those
in health-monitoring, emergency services and latency sensitive applications will be
unacceptable.
• In addition to low latency end-devices requires geographical distribution and mobility
support

Fog Computing: A Short Review of Concept and Applications 2

IoT
• IoT refers to the increasing number of physical
objects which are being linked to the Internet at an
extraordinary rate.
• The IoT allows physical objects to hear, think, and
see, and to create tasks that allow them to
communicate with each other in order to coordinate
decisions and information sharing.
• The IoT converts these objects from traditional
forms to smart entities by utilizing their underlying
technologies.
• IoT applications have spread from the home area
network, smart buildings, industrial automation, and
smart transportation to pervasive healthcare .
• For example, smart homes will deeply count on IoT
devices to identify possible gas leakages, ambient
temperature, air quality etc.

Fog Computing: A Short Review of Concept and Applications 3

• The number of connected devices or IoT devices has already exceeded the population of earth
• These end-devices will produce large number of data, send to cloud for storage and processing and receive the
processed data.
• As the IoT is distinguished by limited storage and processing power, it also has relatively weak performance,
privacy, reliability, and security.
• Combining IoT and cloud networks in the Cloud of Things (CoT) is one of the best ways to overcome almost all
the barriers to the IoT.
• Moreover, the CoT can streamline the processes of IoT data collection, processing, integration, and deployment.
• Data collected by sensors must be analyzed in the cloud, which requires a high bandwidth for the network used.
• This will result in heavy traffic between the cloud and end-devices affecting the quality of service, saturating the
network bandwidth.
• Thus, these issues can be solved by using fog computing
• Because Fog is a cloud close to the ground or to the edge of network, it is able to meet the requirements of
solving these issues

Fog Computing: A Short Review of Concept and

4
Applications
Fog Computing Platform
• Fog computing can be viewed as a development of the cloud computing paradigm from the core of
the network to the edge of the network.
• Several fields, especially the IoT, can benefit from fog computing, a new technology that offers many
advantages.
• It was invented by Cisco and provides a way to process data near the information source, making it
more rapidly accessible and efficient
• Fog computing, often referred to as edge computing, is a decentralized computing
paradigm that extends cloud computing capabilities to the edge of the network, closer
to the data source and end-users
• Fog computing is a greatly virtualised platform which supply computing facilities, storage services,
and networking between users and cloud server or data centres .
• Fog computing combines the mobile cloud computing (MCC) and mobile-edge computing (MEC),
which are of similar concepts .
• MEC is viewed as a cloud server located at the edge of a mobile network performing dedicated tasks
that cannot be achieved by using the traditional network, while MCC is viewed as an infrastructure
where both computation of data and storage is done outside the mobile devices

Fog Computing: A Short Review of Concept and Applications 5

Fog Computing Platform
• Figure 1 represents and
illustrates the role of Fog
computing and its support in IoT
applications and devices .
• As shown in the figure, Fog is
deployed in between the Cloud
and end-devices.
• Fog servers are geographically
distributed and deployed at the
local premises such as parks,
shopping malls, terminals, etc.

Fog Computing: A Short Review of Concept and Applications 6

Fog Computing Platform
• This system architecture brings computation, storage and communication
resources to a close proximity with the end-devices, thus allowing end-devices to
achieve low latency, better and improved quality-of-service (QoS) for streaming
real-time applications.
• The geographical distribution of Fog servers makes it possible for location
awareness and localisation, unlike Cloud servers located in a central data centre.
• In Fog architecture, as shown in the Figure 1, each layer will have different
properties in terms of network bandwidth, storage, etc.
• Because fog nodes or end-devices exist at various layers with interconnected
topologies between them, it has the capability to form a connected graph .
• Fog necessarily does not replace the Cloud, it supplements or act as a
middle-man between the cloud and end-devices or fog nodes to provide critical
services to the end-devices.
•
Fog Computing: A Short Review of Concept and Applications 7
Fog Computing Platform
• Figure 2 (a) shows downloading of
the file from cloud server without a
Fog server in between.
• In this scenario, the localised flyer
or file will have to be uploaded to
the cloud server first, by the store
owner.
• The user then connects to the
Internet and directly download the
flyer or file from the cloud server
through a long distance link, even
though the user is physically near
to the store.

Fog Computing: A Short Review of Concept and Applications 8

Fog Computing Platform
• Unlike that scenario, in Figure 2
(b), the user retrieves the localised
file directly from the Fog server
instead of connecting to the cloud
server.
• Because the file is downloaded
from Fog server, the back-and-forth
traffic between the cloud and the
user or end-device helps to save
backbone bandwidth and increase
data transfer rate, and reduce
service latency and response time.

Fog Computing: A Short Review of Concept and Applications 9

Fog Computing Platform
• The same scenario is applied with IoT end-devices deployed, instead
of users, it will be various sensors and IoT compatible end devices,
communicating between the Fog servers.
• The end-devices can communicate through single-hop wireless
connections.

Fog Computing: A Short Review of Concept and Applications 10

Why Fog
• Fog Computing has the capability to handle the unprecedented amount of
data generated by IoT devices.
• Cloud computing can support IoT by providing on demand growing services
to meet the requirements.
• But applications that require low latency and high quality of service (QoS)
will not be able to perform well, due to congestion in the network.
• Fog computing develops cloud computing to the edge of the networking to
overcome the problems raised, in addition to the other characteristics.
• The characteristics reduces the burden on the networks, improves latency
and performance.

Fog Computing: A Short Review of Concept and Applications 11

Characteristics of Fog Computing
• The building blocks of both Fog and Cloud are computation power, storage,
and networking components.
• Cloud computing is located in a centralized place and have scalable
storage space and computing power.
• But the Fog server has limited storage, computational power and a wireless
interface.
• IoT devices utilizes and relies on Fog server or layer to carry out
computation, storage, communication, control, configuration, and
management.
• IoT end-devices are located close to the source of information, the Fog
server will reduce the latency and jitter.

Fog Computing: A Short Review of Concept and Applications 12

Characteristics of Fog Computing
• The diversity in geographical distribution of Fog servers allows the
capability for IoT devices to be aware of their location based on the
deployed location of Fog servers.
• Geographical distribution of Fog servers also allows data to be stored
at an optimal depth in the network, allowing caching structures can
be optimized.
• According to CISCO , IoT devices will generate huge amount of data
and traffic which will overload interconnect links.
• With the Fog layer or servers deployed, the data associated with IoT
devices can be managed closed to the source, thus saving the
bandwidth on backbone links and reducing the network traffic
Fog Computing: A Short Review of Concept and Applications 13
Characteristics of Fog Computing
• In addition, Fog computing also includes the support for user
mobility, resource and interface heterogeneity .
• The Fog also enables real-time interactions and distributed data
analytics to address the requirements of widely distributed
applications that requires low latency and provides better security.

Fog Computing: A Short Review of Concept and Applications 14

Fog Computing Architecture
• The Fog computing architecture
consists of physical and logical
elements in the form of hardware
and software to implement IoT
(Internet of Things) network.
• As shown in figure, it is composed
of IoT devices, fog nodes, fog
aggregation nodes with the help of
fog data services, remote cloud
storage and local data storage
server/cloud.
•

Fog Computing: A Short Review of Concept and Applications 15

• IoT devices:
• These are devices connected on
IoT network using various wired
and wireless technologies.
• These devices produce data
regularly in huge amount.
• There are numerous wireless
technologies used in IoT which
include Zigbee, Zwave, RFID,
6LoWPAN, HART, NFC,
Bluetooth, BLE, NFC,
ISA-100.11A etc.
• IoT protocols used include IPv4,
IPv6, MQTT, CoAP, XMPP, AMQP
etc.
Fog Computing: A Short Review of Concept and Applications 16
• Fog Nodes:
• Any device with computing, storage and
network connectivity is known as fog node.
• Multiple fog nodes are spread across larger
region to provide support to end devices.
• Fog nodes are connected using different
topologies.
• The fog nodes are installed at various
locations as per different applications such as
on floor of a factory, on top of power pole,
along side of railway track, in vehicles, on oil
rig and so on.
• Examples of fog nodes are switches,
embedded servers, controllers, routers,
cameras etc.
• High sensitive data are processed at these
fog nodes.

Fog Computing: A Short Review of Concept and Applications 17

• Fog aggregate nodes (Fog
Services):
• Each fog nodes have their aggregate
fog node.
• It analyzes data in seconds to
minutes.
• IoT data storage at these nodes can
be of duration in hours or days.
• Its geographical coverage is wider.
• Fog data services are implemented
to implement such aggregate node
points.
• They are used to address average
sensitive data.

Fog Computing: A Short Review of Concept and Applications 18

• • Remote Cloud:
• All the aggregate fog nodes are
connected with the cloud.
• Time insensitive data or less
sensitive data are processed,
analyzed and stored at the cloud.
• • Local server and cloud:
• Often fog computing architecture
uses private server/cloud to store
the confidential data of the firm.
• These local storage is also useful
to provide data security and data
privacy.

Fog Computing: A Short Review of Concept and Applications 19

Fog Computing Working Operation
• As we know there are three types of data viz. most time sensitive data, less time
sensitive data and time-insensitive data.
• Fog computing architecture works based on type of data it receives.
• Nearest fog nodes takes data input from the devices.
• Let us understand working of fog computing architecture.
➨Most time sensitive data are handled by nearest fog node to end device which has
generated the data. After the received data is analyzed, decision or action is transmitted to
the device. After this, fog node sends and stores summary to the cloud for future analysis.
The data at fog node is analyzed in fraction of a second.
➨Less time sensitive data are sent to aggregate node for analysis. After analysis is
performed, aggregate node sends decision or action to the device through nearest node.
Aggregate fog node takes seconds or minutes to complete the analysis. The aggregate
node later sends the report to cloud for future analysis purpose.
• ➨The time insensitive data can wait for longer duration (in hours, days or weeks). The data
is sent to cloud for storage and future analysis.

Fog Computing: A Short Review of Concept and Applications 20

Working of Fog Computing

• Step One: Data is collected from

various devices and sensors
connected to the internet.
• Step Two: That data is then sent to
a fog node, which can be anything
from a smartphone to a smart
thermostat.
• Step Three: The fog node
processes the data and sends it
back to the devices or sensors.
• Step Four: The fog node sends the
data to a central cloud server.
• Step Five: The data is then
processed by the cloud server and
stored in a database.
Fog Computing: A Short Review of Concept and Applications 21
Challenges of Fog Computing
• Increased complexity: It can not be
easy to manage all the devices
connected to the internet. Fog nodes
can help, but they can also add more
complexity.
• Needs more power: It requires a lot of
power, which can be a problem for
battery-powered devices.
• Hard to track devices: It can be hard
to keep track of all the devices
connected to the internet.
• Difficult to maintain
consistency: Fog nodes must
maintain high consistency across data.
• Hard to scale: Fog nodes need to be
able to scale quickly and handle
variable loads.
Fog Computing: A Short Review of Concept and Applications 22
 Difference between Fog computing and Cloud
Computing
Feature Cloud Computing Fog Computing
Geographical
Centralized. Decentralized and distributed.
Distribution

Server Nodes Location Within the internet. Edge of the local network.

Wireless communication: WLAN,

Communication mode IP network WiFi, 3G, 4G, ZigBee, etc. or wired
communication

Dependence on the Can also work in Weak network

Requires strong network core.
quality of core network core.

Client and Server

Multiple hops One hop
Distance
Advantages of Fog Computing
1. Low Latency: Fog computing reduces latency by processing data closer to the source. For example, in
autonomous vehicles, real-time decision-making for collision avoidance requires low latency. Fog
computing can process sensor data locally, making faster decisions compared to sending data to a
remote cloud server.

2. Scalability: It can eiciently scale resources based on demand. For instance, in a smart
manufacturing facility, as more machines are added, fog nodes can be deployed to distribute computing
power, ensuring smooth operations without overloading a centralized cloud server.

3. Privacy and Data Security: Fog computing can enhance data privacy and security by keeping
sensitive data within the local network. In healthcare, patient data can be processed locally, ensuring
compliance with data privacy regulations like HIPAA.

4. Reduced Bandwidth Usage: By processing data locally, fog computing reduces the amount of data
that needs to be transmied to the cloud. In video surveillance systems, only relevant video footage may
be sent to the cloud, reducing bandwidth costs.

5. Reliability: Fog computing can ensure continuity of operations even in the absence of internet
connectivity. For example, in a remote agricultural seing, fog nodes can control irrigation systems
independently of cloud connectivity.
 Disadvantages of Fog Computing
1. Resource Limitations: Fog devices may have limited computational resources compared to cloud
servers. This can be a limitation for processing complex tasks. For instance, running deep learning
models for image recognition may be challenging on resource-constrained edge devices.

2. Management Complexity: Managing a distributed fog computing infrastructure can be more complex
than managing a centralized cloud. Ensuring all edge devices are updated, secure, and coordinated can
be challenging.

3. Cost: Deploying and maintaining fog nodes at the edge can be costly, especially when dealing with a
large number of devices. The cost of hardware, maintenance, and energy consumption can add up.

4. Compatibility Issues: Ensuring compatibility and interoperability among various fog devices and
software can be challenging. In a smart home ecosystem, integrating devices from dierent
manufacturers may require significant eort.

5. Security Concerns: While fog computing can enhance security in some cases, it also introduces new
security challenges. Edge devices may be more vulnerable to physical aacks or tampering, and
managing security across a distributed network can be complex.
Applications

1. Smart Grid

In this application, grid devices

like sensors, which are present at
the edges, are aached to the fog
devices. These grid devices
generate the data and fog
devices issues commands to the
actuators for managing the
generated data. Comparing the
energy demand, these devices
have the capability of switching to
the alternative energies like wind
and solar energy system
Applications

2. Intelligent Transportation

In this, all the vehicles are

having sensors that are
connected with the fog
devices. This also provides an
ease for vehicle-to-vehicle
communication. This system
involves many
communications including
satellite communications,
vehicle-to-vehicle
communications, safety
systems, traffic signs, fleet
Applications
3. Healthcare

The focus of the applications during analytics of data in biomedical area for correctly identifying smart health-care
decisions in the future.

In real time, this is possible by

three-level hierarchal structure
comprising of fog nodes rather than
using cloud computing technology
independently.
It has three main sections: front end,
back end and communication module. .
In this application, front end and back
end both take independent decisions,
and then, their collaborative results
form the actual solution.
Applications
4. Video Surveillance

This technology overcomes latency

issues in the packets transmied over
the network, thus making the video
quality rich to the user. It plays a vital
role in real-time data analysis. Video
surveillance is used in malls and other
large public areas and has also been
implemented in the streets of
numerous communities. Fog nodes can
detect anomalies in crowd paerns
and automatically alert authorities if
they notice violence in the footage.
Disadvantages of Fog Computing
1. Resource Limitations: Fog devices may have limited computational
resources compared to cloud servers. This can be a limitation for processing
complex tasks. For instance, running deep learning models for image
recognition may be challenging on resource-constrained edge devices.

2. Management Complexity: Managing a distributed fog computing

infrastructure can be more complex than managing a centralized cloud.
Ensuring all edge devices are updated, secure, and coordinated can be
challenging.

3. Cost: Deploying and maintaining fog nodes at the edge can be costly,
especially when dealing with a large number of devices. The cost of hardware,
maintenance, and energy consumption can add up.

4. Compatibility Issues: Ensuring compatibility and interoperability among

Application of Fog Computing
• Healthcare
• Healthcare services and application are delay sensitive and generates private
data of the patients .
• The data generated contains sensitive and personal information, and location
data might be sensitive in some scenarios.
• Higher jitter and latency can cause several problems in telehealth and
telemedicine applications, which makes Fog computing a suitable paradigm in
healthcare applications.
• Fog computing can provide better management for the flow of data from and
to the cloud.

Fog Computing: A Short Review of Concept and Applications 31

Babak Bashari Rad, Ali Aseel Shareef ,“Fog Computing: A Short Review of
Concept and Applications”, IJCSNS International Journal of Computer Science
and Network Security, VOL.17 No.11, November 2017

Fog Computing: A Short Review of Concept and Applications 32