MODULE 3

IoT Processing Topologies and Types

DATAFORMAT: The Internet is a huge place where a massive
amount of data is created and shared every day. As of 2018, there
were over 4 billion people using the Internet, and each person
uses multiple devices like smartphones, laptops, and tablets.
Apart from human users, machines like sensors and monitoring
systems also generate data.
This data comes in many forms, such as:
• Emails
• Text documents (Word, PDFs, etc.)
• Social media posts
• Videos, audio files, and images
 However, all this data can be broadly classified into two types
based on how it is stored and accessed:
1. Structured Data

This type of data is organized in a fixed format, like tables in a database.

Think of a spreadsheet where each column has a specific type of data
(names, phone numbers, account balances, etc.).
Structured data is mainly used in:
Banking systems
Flight and train reservations
Inventory management
 Because it is well-organized, searching through structured data is easy
using tools like SQL (Structured Query Language). However, in IoT,
structured data makes up only a small portion of the total data
generated.
2. Unstructured Data
This is all the data that does not follow a fixed format. It can come from various
sources and can be very different from one another.
Examples of human-generated unstructured data:

Emails
Videos and images
Audio recordings
Chats and messages
Examples of machine-generated unstructured data:

Traffic sensor data

Data from industries and buildings
Satellite images
Surveillance videos
 Since unstructured data has no specific format, searching through it is
difficult. Instead of SQL, NoSQL (Not Only SQL) is used for handling
unstructured data.
•Structured data is like a neat and organized bookshelf where everything is
labeled.
•Unstructured data is like a messy room with things scattered everywhere,
making it harder to find what you need.
Key Components:
1. Central Node – Internet:
1. The Internet acts as the hub, connecting multiple data sources.
2. Data Storage Systems:
1. Databases: Represent structured storage of data.
2. Cloud: Remote storage and computing services.
3. Data Sources:
1.Industry: Represents industrial data generation (e.g., manufacturing,
production).
2.Transport: Includes vehicles and public transportation systems that
generate data.
3.Buildings: Smart buildings and infrastructure contribute data.
4.Traffic: Vehicles and road systems provide real-time traffic data.
5Transaction: Digital transactions, banking, and financial systems produce
data.
6.Human: Individuals interact with the digital world, contributing data
through devices, applications, and online activity.
Explanation:
• The arrows indicate the flow of data between sources
and the Internet.
• Data from different domains (industry, transport,
buildings, traffic, transactions, and human activities) is
collected, processed, and stored in databases or cloud
systems.
• This interconnected system represents the backbone of
big data, IoT (Internet of Things), and digital
transformation.
Importance of Processing in IoT
 Since the Internet generates huge amounts of data, we need smart
and efficient ways to process it. With the rise of IoT (Internet of
Things), more and more devices are connected, putting pressure on
the global network. To manage this, we must decide what to process
and when to process it.
Types of Data Processing Based on Urgency
Data is divided into three categories depending on how fast it needs to
be processed:
Very Time-Critical Data (Needs Instant Processing)
• Data that requires immediate action within milliseconds.
• Examples:
Flight control systems – Planes need real-time adjustments.
Healthcare systems – A patient’s life may depend on immediate
data.
2.Time-Critical Data (Can Wait a Few Seconds)
• Data that needs to be processed quickly but can tolerate a short
delay of a few seconds.
• Examples:
Vehicle and traffic data – Smart traffic lights adjusting signals.
Smart home systems – Adjusting room temperature based on
occupancy.
Surveillance cameras – Detecting suspicious activity.
3. Normal Data (Can Wait Minutes or Hours)
• Data that does not require immediate processing and can be
handled slowly.
• Examples:
Agriculture data – Weather patterns, soil moisture updates.
• Environmental monitoring – Air quality analysis, pollution
tracking.
Where Should Data Be Processed?
• Very time-critical data → Needs to be processed immediately and
near the source (e.g., on the device itself).
• Time-critical data → Can be processed remotely (e.g., in cloud
servers or through shared processing systems).
• Normal data → Can be processed later, without urgency (e.g.,
stored and analyzed over time).
Processing Topologies:

In an IoT system, deciding how and where to process data is very

important because it affects speed, cost, and energy usage. A well-
planned system can save internet bandwidth, reduce power
consumption, and ensure that everything runs smoothly without
delays.
We can broadly classify IoT data processing into two main
categories:
1. On-site Processing (Local Processing) – This means processing
the data near where it is collected, like within the IoT device
itself or a nearby computer. This is useful when quick responses
are needed, such as in smart homes or self-driving cars.
2. Off-site Processing (External Processing) – Here, the data is
sent somewhere else for processing. This can be further divided
into:
1. Remote Processing – The data is sent to a distant server or
cloud for processing. This is useful when large amounts of
data need to be analyzed, such as in weather forecasting or big
data applications.
2. Collaborative Processing – Instead of one single remote
location, multiple servers or systems work together to process
the data. This is helpful in applications where data is collected
from different locations and needs to be analyzed together,
such as in traffic management or smart cities.
For example, in a fire detection system, a temperature sensor can
quickly detect a rise in temperature and process the data
immediately to trigger an alert. This avoids delays caused by
sending the data to another location for processing. However, if
needed, the data can still be sent elsewhere for further analysis or
storage.
 This method ensures fast and reliable responses, which is critical
in situations where even a small delay can cause big problems.
Off-site processing means that data is not processed where it is
collected but is sent to another location for processing. This
method is slower than on-site processing because of network
delays, but it is much cheaper and more practical for large-scale
IoT systems.
 In off-site processing, sensor nodes only collect and send data;
they do not need powerful processors. Instead, a few high-
powered servers or cloud systems handle the processing for many
sensors. This reduces costs and makes it easier to manage large
networks.
There are two main ways this can work:
1.Remote Processing – Data is sent to a far-away server or
cloud for analysis.
2.Collaborative Processing – If a direct connection to a remote
server is not possible, multiple nearby devices work together to
process the data.
This approach is useful when real-time processing is not
necessary, such as in weather monitoring or smart city
applications, where data is analyzed later rather than instantly.
Remote Processing
Remote processing is a common method used in IoT (Internet of Things) systems.
Instead of processing data where it is collected, the data is sent to a powerful
remote computer or cloud server for analysis.
How It Works:
1. Sensors collect data (e.g., temperature, motion, air quality).
2. The data is sent to a remote server or cloud through the internet.
3. The server processes and analyzes the data to make decisions (e.g.,
detecting a fire and triggering an alarm).
Advantages:
 Cost-effective & energy-saving – Many sensors can share a single, powerful
processing system instead of needing expensive hardware at every location.
Scalable – The system can handle data from thousands of sensors without
much extra cost.
Example:
A smart city system has thousands of traffic cameras.
Instead of processing video at each camera, they send the
footage to a central cloud server, which analyzes traffic
patterns and adjusts signals accordingly.
Collaborative Processing
Collaborative processing is used when there is little or no
internet connection, making it difficult to send data to a remote
server. Instead of relying on a cloud, multiple nearby devices
(sensor nodes) work together to process data locally.
How It Works:
1.Sensors collect data (e.g., temperature, soil moisture in a
farm).
2.Nearby sensors share and process the data together instead
of sending it far away.
3.Decisions are made locally without needing an internet
connection.
Advantages:
 Works even without internet – Since the data is processed
locally, there is no need for a strong network.
Faster processing – Avoids delays caused by sending data over
long distances.
Saves bandwidth – Reduces the amount of data sent over
networks, saving costs.
Best for:
Farms & agriculture – Since farm data (like soil moisture) doesn’t
need instant updates, local processing is enough.
Large-scale systems in remote areas – Places where setting up
internet infrastructure is expensive or difficult.
Example:
 Imagine a smart farm where different sensors check soil
moisture. Instead of sending data to a cloud server, nearby
sensors share and process the data together to decide when to
water the crops.
 IoT Device Design and Selection Considerations
When designing an IoT device (like a smart sensor), the processor is the most
important part. It affects how well the device works, how much power it uses, and
how affordable it is. Here are the key factors to consider when selecting a
processor for an IoT sensor node:
1. Size
• The smaller the device, the less power it consumes.
• Small-sized processors are better for IoT applications like smartwatches and
wearables.
2. Energy Consumption
• A processor that uses more energy needs frequent battery replacements.
• Low-power processors are better for long-lasting IoT devices, especially those
used in remote areas.
3. Cost
• Cheaper processors allow more sensors to be used at a lower cost.
• Example: Affordable fire and gas sensors can be installed in more locations,
improving safety.
4. Memory
• Memory affects data processing, storage, and filtering.
• Devices with more memory can do more, but they are also more expensive.
5. Processing Power
• Stronger processors can handle complex tasks like image or video processing.
• Simpler processors are enough for basic tasks like temperature sensing.
6. I/O Rating (Input-Output)
• Determines how a processor connects with sensors and devices.
• Newer processors use 3.3V instead of 5V, so older sensors may need extra
components to work.
7. Add-ons
• Extra features like USB, Wi-Fi, Bluetooth, and built-in timers make the processor
more useful.
• Having these features built-in makes it easier and faster to develop IoT devices.
Processing Offloading:
In IoT (Internet of Things), tiny sensors are used to collect data (like
temperature, fire detection, or security monitoring). However, these small
devices usually don’t have enough power or storage to process large
amounts of data. Instead of handling everything on the device itself, the data is
sent to stronger computers for processing.
How Does Processing Offloading Work?
1. Sensing Layer (Collecting Data)
1. Sensors detect changes in the environment (fire, movement, temperature,
etc.).
2. These sensors are connected to a processor using wired or wireless
connections.
2. Local (Edge) Processing
1. If quick action is needed, some processing is done locally (on the device
or nearby computers).
2. Example: A fire alarm system that immediately sounds an alert when
smoke is detected.
3.Fog Computing (Nearby Processing)
•If the device cannot process the data alone, it sends the data to a fog node (a
nearby, more powerful computer).
•This fog node helps process the data faster without relying on the Internet.
•Example: Smart cameras in a factory analyzing footage in real time to detect
issues.

4.Cloud Computing (Remote Processing)

•For large-scale analysis, the data is sent to the cloud, which has huge storage
and processing power.
•Example: A weather monitoring system that collects temperature and air
quality data from multiple locations and analyzes it in the cloud.
Why is Processing Offloading Important?
Saves battery on small devices.
Makes IoT devices cheaper by reducing hardware costs.
Allows faster processing with powerful cloud or fog-based
computers.
From the Diagram
 This diagram explains how data is collected, processed, and stored in
different locations, from local devices to the cloud.

1.Data Collection (On-site)

•Sensors (such as temperature and camera sensors) collect data from the
environment.
•Examples: A fire detection system using temperature sensors, or a surveillance
system using cameras.
2.Local Processing
•The collected data is processed at a nearby computing device called a sensor
node to reduce unnecessary data transmission.
•This step helps in quick decision-making (e.g., triggering an alarm if a fire is
detected).
3.Edge Processing
•Local network clusters (multiple interconnected devices) further process the data
before sending it to distant locations.
•This reduces the load on the internet by filtering out unnecessary information.
4.Fog Processing
•A server near the data source (e.g., in a building or a city) processes data
before sending it to the cloud.
•It may store the data in a local database temporarily.
5.Cloud Processing (Global)
•If needed, data is sent to cloud servers and databases for long-term storage
and advanced analysis.
•The internet is used to transmit data from local servers to cloud-based
databases.
6.Communication Methods
•Different communication methods are used, such as wired/wireless
connections, short-range wireless, and long-range wireless/backbone
networks.
•Short-range is used for nearby devices, while long-range connects to
remote servers.
In IoT, data offloading means sending the data collected by sensors to
more powerful computers for processing and storage.
1.On-Site Layer (Sensing & Local Processing)
• Sensors (Temperature, Camera, etc.) collect data from the
environment (e.g., fire detection, surveillance).
• Sensors connect to a sensor node, which does some basic local
processing (e.g., detecting temperature changes).
• Communication between sensors and the sensor node is done via
wired or wireless connections.
2.Local Layer (Edge & Fog Processing) 🌐
• Edge Processing (Local Network Clusters)
• Data is sent to edge devices (like a nearby computer) for quick local processing.
• Edge devices are within the same network and connected using short-range wireless
communication.
• Fog Processing (Gateway Servers & Databases)
• Data is processed in local servers (fog nodes) before being sent to the cloud.
• Fog computing allows for faster response times than cloud processing.
3.Global Layer (Cloud Processing & Storage) ☁️
• If deeper analysis or long-term storage is needed, data is sent to cloud servers via long-
range wireless networks (like the Internet).
• Cloud servers store and analyze data for big applications like AI-based surveillance or
global monitoring.
• This method requires high bandwidth and network infrastructure, making it more
expensive but highly scalable.
Offload location:
The location where processing is offloaded impacts the cost,
efficiency, and sustainability of an IoT system. There are four main
types of offload locations:
1. Edge Computing (Processing Near the Source)
Processing happens directly on the IoT device or very close to it
(e.g., routers, gateways).
Helps in reducing bandwidth usage and fast response times.
Useful for real-time data processing (e.g., smart cameras
analyzing video footage locally).
 Example: A smart thermostat analyzing temperature data on
the device itself before sending only important updates to the
cloud.
2.Fog Computing (Intermediate Processing Layer) ☁️
Data is processed closer to the IoT devices but not on them (e.g.,
local servers, edge devices).
Reduces network congestion by filtering unnecessary data
before sending it to the cloud.
Helps in low-latency applications (e.g., autonomous vehicles,
smart cities).
 Example: A smart traffic management system where cameras
and sensors send data to a nearby fog server to adjust traffic
signals in real-time.
3. Remote Server (Standalone, Powerful Computer)
• A single, powerful server that processes IoT data instead of edge
or fog computing.
• It can handle complex tasks but may not scale easily if more
devices are added.
• Why use it? Good for moderate IoT applications but expensive
and harder to maintain.
• Example: A security system that sends video data to a nearby
dedicated server for facial recognition processing.
4. Cloud Computing (Powerful, Internet-Based System)
• Data is sent to a cloud server over the internet, where it is
processed using advanced tools and algorithms.
• Why use it? Scalable, flexible, and doesn’t require users to
maintain hardware.
• Example: A weather prediction system that collects global
sensor data, processes it in the cloud, and provides forecasts.