HALDIA INSTITUTE OF

TECHNOLOGY

A TECHNICAL REPORT ON INDUSTRIAL IOT AND

SMART MANUFACTURING

SUBMITTED BY:
1. ADITYA KUMAR (24/ECE/009)
2. ABHIJEET RATHORE (24/ECE/002)
3. ANKIT KUMAR (24/ECE/026)
4. DWIP MUZRAI (24/ECE/057)
5. ARUNASISH (24/ECE/037)
TO:
1. DR. AVISHEK DAS
2. DR. SHANTUNU KR MAITY
TABLE OF CONTENT
Sl. Page
Section Subsections
No. No.
Summary of Edge AI and
1 Abstract 3
Real-Time IoT Applications
Edge AI, IoT, Real-Time
2 Introduction 4-5
Processing, Edge Computing
Edge Devices, AI Models,
Core
3 Sensors, Communication 6
Technologies
Networks
Smart Healthcare,
4 Applications Autonomous Vehicles, Smart 7
Cities, Industrial Automation
Low Latency, Privacy,
5 Key Benefits Reliability, Reduced 8
Bandwidth Usage
Security, Power
6 Challenges Consumption, Scalability, 9
Hardware Limitations
Future Scope Future Developments and
7 10
and Conclusion Summary
11-
8 References Research Papers and Articles
12
ABSTRACT
Moreover, Edge Artificial Intelligence (Edge AI) is
revolutionizing the IoT industry by empowering smart
gadgets to do intelligent analysis of data without
necessarily involving centralized cloud servers. IoT real-
time applications demand quick decision making and
minimal data handling. However, centralized cloud
systems may not offer the expected performance levels
since network delays are common. Edge AI uses the
power of both machine learning algorithms and edge
computing capabilities to ensure speedy, reliable, and
secure processing of data.
Through the use of Edge AI, devices can be able to
analyse data, identify anomalies and make independent
decisions based on their findings. Some examples of
application areas for this type of technology include
smart healthcare, industry automation, autonomous
driving, surveillance, and smart cities among others.
Edge AI reduces the need for cloud communication,
thus resulting in lower latency rates and lower usage of
bandwidths, hence improved system security.
New inventions in the field of accelerators for AI,
embedded processors and lightweight neural networks
have made it possible to execute tasks like image
processing, speech processing, predictive maintenance,
and environment monitoring efficiently. The above
factors have contributed significantly to the
development of smarter and faster applications within
the Internet of Things industry.
INTRODUCTION
 Internet of Things (IoT) means a network that is
built with interconnected devices which
communicate and share data via internet. There
are a lot of data created by the IoT devices from
sensors, cameras, and smart machinery. The way
these data are normally handled is via transmission
to cloud servers for further processing and
analysis. But the processing done through cloud
servers poses problems of latency and reliance on
bandwidth and security.
 Edge AI technology is based on advanced
computer technology where artificial intelligence
technologies are implemented on the device's end
rather than sending all the data to the cloud. In
other words, edge AI works by processing the
collected data right at or near the place of origin
instead of transmitting them to cloud servers.
 Edge computing forms the basis of Edge AI. The
deployment of computing resources nearer the IoT
devices helps decrease the burden on centralized
cloud-based infrastructures. When coupled with
machine learning and deep learning algorithms,
edge computing helps create intelligent and
responsive systems that analyse data in real-time.
  With the development of Industry 4.0, intelligent
cities, and intelligent automation, the demand for
Edge AI is increasing rapidly. Advanced devices
used today incorporate robust processors, GPUs,
and AI accelerators, which allow them to execute
lightweight AI algorithms in an efficient manner.

CORE TECHNOLOGIES IN EDGE AI

AND REAL-TIME IOT
1 | Edge Devices and Embedded Systems
Edge devices are hardware-based systems that do
computations and data analysis locally at the point
close to where the data is generated. Examples of edge
devices include Raspberry Pi, NVIDIA Jetson Nano,
Arduino, smart cameras, and industrial gateways. Such
devices do computations and AI inference without
being completely dependent on cloud services.
2 | Artificial Intelligence and Machine Learning
AI and ML algorithms help IoT devices understand
patterns, classify objects, find anomalies and make
intelligent decisions. Light-weighted deep learning
models are specifically designed for use on edge
devices to minimize computation power.
3 | Sensors and Data Acquisition
Sensors are IoT devices used to measure variables like
temperature, humidity, motion, pressure, and images.
Sensors are key components in generating data for
Edge AI systems.
4 | Communication Networks
Communication technologies that enable IoT devices to
communicate with edge servers include 5G, Wi-Fi,
Bluetooth, Zigbee, and LoRa WAN technologies. Fast
communication networks facilitate efficient data
transfer.

APPLICATIONS OF EDGE AI IN REAL-

TIME IOT
1 | Smart Healthcare
Edge AI technology allows wearable gadgets and
clinical sensors to track patient vitals in real-time. The
smart healthcare system can diagnose abnormal
heartbeat rates, track glucose level, and even offer
instant alerts.
2 | Self-Driving Cars
Edge AI technology helps self-driving cars detect
objects, track lanes, and avoid collisions by processing
the information from various sensors and cameras
installed on the vehicle. Instant processing is crucial to
maintain road safety.
3 | Manufacturing Processes
The manufacturing industry applies Edge AI technology
in various processes such as predictive maintenance,
product quality inspection, robotics, and process
monitoring. Smart IoT systems analyse the data of
machines to predict failures.
4 | Smart Surveillance Systems
Edge AI technology helps cameras conduct facial
recognition, motion detection, and suspicious behaviour
detection without having to send the entire footage to
the cloud server.
5 | Smart City
Edge AI technology is applied in various applications in
smart cities including traffic management, street
lighting, waste management, and environmental
monitoring.

KEY BENEFITS OF EDGE AI FOR

REAL-TIME IOT APPLICATIONS
1 | Low Latency
As Edge AI performs data analysis locally, latency
issues due to data transfer to cloud-based servers are
reduced, which is crucial for applications like medical
care and self-driving cars.
2 | Lower Bandwidth Consumption
Since data is processed locally, only necessary data is
transferred to the cloud, resulting in lower network
bandwidth usage.
3 | Higher Security
Personal data stays on local devices rather than being
constantly transferred to the internet, increasing data
security and lowering cybersecurity threats.
4 | Increased Resilience
Network disruptions do not affect Edge AI's
functionality as it operates locally and does not depend
on network connectivity.
5 | Energy Efficiency
Current edge devices run highly-efficient AI models
requiring minimal energy for operation.
6 | Scalability
Edge AI technology allows scaling IoT implementations
since computations can be distributed among
numerous edge devices.

CHALLENGES OF EDGE AI IN IOT

1 | Lack of Computational Resources
The resources available on edge devices are not as
much as those provided by cloud-based servers. It is
tough to run complex machine learning models on such
devices.
2 | Security Threats
Internet of Things (IoT) networks are exposed to
security vulnerabilities and can be subjected to various
cyber threats.
3 | High Energy Consumption
Battery-powered IoT devices necessitate low power
consumption AI models for their optimal functioning.
4 | Scalability and Device Management
Challenges
It becomes difficult to manage multiple devices at once
as their number increases.
5 | Optimization of Models
Machine learning models should be compacted to fit
into resource-restricted edge devices without affecting
accuracy.
6 | Connectivity Problems
Even though edge computing minimizes dependency
on the cloud, effective connectivity is still required for
the working of devices.

FUTURE SCOPE OF EDGE AI IN

REAL-TIME IOT

 Future prospects for Edge AI seem very bright,

given the fast pace at which developments in AI
hardware, communication technologies, and
embedded systems occur. The coming
implementation of 5G will enable faster
performance of IoT applications in real time.
 In the future, Edge AI will be increasingly self-
sufficient, energy-efficient, and intelligent. With the
help of advanced AI accelerators and neuromorphic
computing, sophisticated AI applications will be
performed directly on small IoT devices. TinyML
 and federated learning will be applied to enhance
privacy and distribution.
 Edge AI will have a critical impact on smart
transport, robotics in healthcare, intelligent
manufacturing processes, defence operations, and
environmental sustainability. Edge AI together with
blockchains and cybersecurity technologies will
significantly increase the security of IoT systems.
 In the course of the implementation of the industry
4.0 concept by different industries, Edge AI will
take a critical role.

CONCLUSION
Edge AI is changing the landscape of IoT real-time
applications by providing intelligent computing at the
edge of the network. In addition, Edge AI decreases
latency, ensures privacy, lowers bandwidth
consumption, and increases the efficiency of IoT
technology. Examples of the use of Edge AI include
healthcare applications, autonomous vehicles,
industrial automation, and smart surveillance.
Although there are some difficulties that exist in
implementing Edge AI, such as low processing power of
edge devices, security threats, and scalability
problems, improvements in artificial intelligence
algorithms and edge computing hardware can enhance
the performance of Edge AI systems. Moreover,
incorporating Edge AI into IoT is likely to increase smart
system applications.
Going forward, Edge AI is forecasted to become one of
the most significant technologies for digital
transformation in the next few years.

REFERENCES
[1] Y. LeCun, Y. Bengio, and G. Hinton, “Deep Learning,”
Nature, vol. 521, no. 7553, pp. 436–444, 2015.
[2] Cisco Systems, “Fog Computing and the Internet of
Things: Extend the Cloud to Where the Things Are,”
Cisco White Paper, 2015.
[3] M. Satyanarayana, “The Emergence of Edge
Computing,” IEEE Computer, vol. 50, no. 1, pp. 30–39,
2017.
[4] NVIDIA Corporation, “Edge AI and Intelligent IoT
Solutions.”
Available: [Link]
[5] TensorFlow Lite Documentation, “Machine Learning
for Mobile and Edge Devices.”
Available: [Link]
[6] Amazon Web Services, “AWS IoT Greengrass.”
Available: [Link]
[7] Microsoft Azure, “Azure IoT Edge.”
Available:
[Link]
[8] Qualcomm Technologies, “AI at the Edge: Enabling
Real-Time Intelligence.”
Available: [Link]
[9] IEEE Xplore Digital Library, “Edge AI for Smart IoT
Systems.”
Available: [Link]
[10] Elsevier Journal, “Applications of Edge Computing
in IoT.”
Available: [Link]
[11] ResearchGate, “Edge AI and Future IoT
Applications.”
Available: [Link]