Föreläsning 2 - The common architecture types of IoT-based systems

Question 1

What are the layers of IoT architecture in a 6-layer model? -

Answer 1

• The 3-layer architecture (perception layer, processing and storage layer, and application layer).
• The 4-layer architecture (adds the business layer to the 3-layer architecture).
• The 5-layer architecture (adds the network layer to the 3-layer architecture).
• The 6-layer architecture (adds the adaptation layer to the 3-layer architecture).

Question 2

Explain the 3-layer IoT architecture -

Answer 2

• The perception layer consists of the physical objects and the sensor device.
• The processing and storage layer oversees analysis and stores the data gathered by sensors (Cloud computing, machine learning, and database software are being used to best handle the collected information.).
• The application layer provides the service requested by customers.

Question 3

Explain the 4-layer IoT architecture -

Answer 3

The business layer fosters innovation by leveraging IoT data to identify new business opportunities, develop new products or services, and improve existing offerings.

Question 4

Explain the 5-layer IoT architecture -

Answer 4

The network layer transfers information from the perception to the processing layer. The transmission medium can be wired or wireless, and technology can be 3G, UMTS, Wi-Fi, Bluetooth, infrared, ZigBee, etc,. Depending upon the sensor devices.

Question 5

Explain the 6-layer IoT architecture -

Answer 5

The adaptation layer is an adapter that facilitates the interoperability of IoT heterogeneous devices.

Question 6

Cloud-based IoT architecture -

Answer 6

Has unlimited processing and storage capabilities. Cloud models include: IaaS, PaaS, SaaS.

Question 7

Service-Oriented IoT architectures -

Answer 7

They put the service at the centre of the IoT systems design.

Question 8

Microservice-based architectures -

Answer 8

Microservices are small, independent applications that handle one specific task. They can be deployed, tested, and scaled separately from other parts of the system. Microservice architecture builds systems using many of these small services, based on Service-oriented Architecture (SOA) and virtualisation. This helps improve scalability and overall system quality.

Question 9

Monolithic vs microservices -

Answer 9

• Monolithic is when the whole application is built as one single unit. All parts are connected, so it’s easier to develop at first but harder to scale and update.
• Microservices break the app into small, independent services. Each service does one job and can be updated, tested, and scaled separately, which makes it more flexible for larger systems.

Question 10

Microservice advantages -

Answer 10

• Organizational alignment (teams can focus on smaller, independent services, speeding up development).
• Composability (easily compose a microservice to get new functionalities).
• Ease of deployment (allowing changes to be made to one service without affecting the whole system. Rollbacks are simpler, reducing downtime risk).

Question 11

How to design a microservice -

Answer 11

1. Bounded contexts (identify different areas within your domain, each microservice should handle a distinct function. Only share necessary data between them, keeping other data within their context.).
2. Loose coupling (minimize dependencies between microservices, ensure changes to one service dont impact others, and each service should know as little as possible about others.).

Question 12

Hidden models -

Answer 12

represent the internal data structure and logic within each microservice that are not directly exposed to other services. These models are hidden from the outside world and encapsulated within each microservice. E.g a user model that stores authentication data, used only by the authentication service and not shared with other services.

Question 13

Shared models -

Answer 13

Shared models represent data structures or schemas that are shared and used by multiple microservices. These models serve as a common language or contract between services, facilitating communication and integration. E.g. a booking and inventory service shares a common model with the number of available cars, so both services know how many available cars.

Question 14

Why should you avoid shared DB integration? -

Answer 14

• Schema changes affect everyone → One service changes the database = others may break.
• Blocks system evolution → Hard to switch from SQL to NoSQL if all services rely on the same DB.
• Limits tech choices → You’re stuck with DBs/languages that work well together (e.g., MySQL + Java).

Question 15

Client-side service discovery -

Answer 15

Each microservice registers itself with the service registry(as when they are available). Service registry responds with the instance of the requested service to the client. Fewer network calls (just query service registry). Coupling between client and service registry.

Question 16

Service-side service discovery -

Answer 16

Client sends a request to API gateway or load balancer. The load balancer or API gateway uses the Service registry to discover services. Separation of logic from client. Load balancer needs to be managed and replicated. Additional network hop.

Question 17

Publish/subscribe IoT architecture -

Answer 17

MQTT client (publisher) → Topic → MQTT broker → Subscriber.

Question 18

Architectural distribution patterns, name three -

Answer 18

With distribution, we refer to the deployment of a software system to hardware.
* Centralized architecture (all parts of the system run on a central server),
* Distributed architecture (system is split across multiple machines that work together)
* Decentrilized architecture (No central control, all nodes is independent, but can connect to others).

Question 19

Advantages and challenges on a cloud-centric approach -

Answer 19

• • → scalability (can handle massive IoT networks). Advances AI and Big data capabilities.
• Challenges → High network dependency (not ideal for remote aread). Privacy and security risks (data stored in third-party clouds).

Question 20

What is an edge-centric approach? -

Answer 20

processing happens closer to the data source (edge device, gateways, local servers) instead of cloud. Reduces latency and bandwidth costs by analyzing data locally before transmitting only essential information. E.g. Autonomous vehicles are making real-time driving decisions using onboard AI.

Question 21

Edge-centric approach advantages and challenges -

Answer 21

• Advantages → faster real-time processing, lower cloud storage and bandwidth costs.
• Challenges → required powerful edge devices, increasing hardware costs, complexity in managing distributed edge nodes.

Question 22

What is an hybris (cloud-edge) approach? -

Answer 22

Combines cloud and edge computing for optimized performance, security, and scalability. Frequently used in industrial, smart city, and healthcare applications where real-time response is critical, but long-term data analysis is needed. E.g. Industrial automation with edge-based safety monitoring and cloud-based AI-driven process improvements.

Question 23

Hybrid approach advantages and challenges -

Answer 23

• Advantages → balance real-time performance and long-term analytics, more control over security than pure cloud solutions.
• Challenges → complexity in implementation and management, required seamless synchronisation between edge and cloud.

Question 24

Over the top (OTT) deployment -

Answer 24

Over-the-Top (OTT) IoT deployment delivers IoT services independently of the network, running on existing internet infrastructure, similar to how Netflix and YouTube operate without traditional cable or satellite.

Question 25

Key characteristics of OTT -

Answer 25

1. Network-Agnostic: IoT services don’t require tight integration with specific networks (e.g., cellular, Wi-Fi). Devices can use any available connectivity to send/receive data. 2. Cloud-Centric: Most OTT IoT platforms are cloud-based, offering remote access, scalability, and flexibility. Data processing and management occur in the cloud, not on local infrastructure. 3. Scalability & Flexibility: OTT platforms easily support a wide range of IoT devices across locations and industries without the need for heavy investment in network infrastructure.

Question 26

Advanatges of OTT iot depolyment -

Answer 26

* Interoperability – Easier to integrate with different IoT devices, platforms, and business applications. * Lower Costs – No need for proprietary infrastructure; utilizes existing public networks and cloud platforms. * Faster Time to Market – Organizations can launch IoT solutions quickly without complex network integrations.

Question 27

IoT platform security and privacy challenges -

Answer 27

* Critical Requirement: Security and privacy are essential for IoT platforms, with 57% of organizations citing security as the greatest challenge. * Industrial Considerations: Sectors like oil, gas, and transportation need segmented architectures to secure operational data (e.g., Purdue Model). * OTT challenges → vulnerabilites (iot platforms are susceptible to attacks like DoS, Data control (Ownership and management of data are unclear, raising legal and financial concerns), and Privacy risk (increased exposure of sensitive data).

Question 28

What is middleware? -

Answer 28

IoT systems involve challenges such as different devices, networks and application making it hard to standardize. New technologies make the systems more complicated. Middleware acts as a glue that connects applications, devices and communication interfaces. It ensured consistent standard for different protocols and data formats → easier for components to communicate.

Question 29

Middleware architecture -

Answer 29

1. Interface protocol. 2. Device abstraction. 3. Centralized management and control. 4. Application abstraction.

Question 30

Middleware vs IoT platform -

Answer 30

Middleware provides interfacing across the IoT stack but does not replace an IoT platform. It pervades the entire architecture, ensuring seamless interaction between system components.