IoT System Architectures and Standards
Syllabus
This module will cover the following aspects
• Key considerations for IoT architectures
• Cloud, fog, and edge paradigms
• The role of gateways in IoT
• IoT internetworking approaches
• Standards that enable practical IoT deployment and interoperability
2 © 2020 Arm Limited
The IoT architectural landscape
• Thousands of new applications exist, spanning countless domains (verticals).
• Each application has its unique requirements → combining these leads to
systems that are complex, difficult to manage, and often proprietary.
• Defining a unified architecture is challenging and interoperability problematic, if there are too many standards to choose from.
• Efforts by multiple entities to define common frameworks, including international standardization bodies, multi-national collaborative research projects, industry consortiums, and large commercial actors.
• Device/protocol documentation is scattered and often difficult to navigate.
• We will focus on the key principles different architectural patterns share and examine some examples.
3 © 2020 Arm Limited
Key considerations for IoT architectures
What application domains should be covered?
Where to place the “intelligence”?
What networking structure should be employed?
How to modularize systems, so as to manage complexity and enable programmability? What are the cost and scalability implications?
4 © 2020 Arm Limited
A layered view to IoT architecture
At a high level, stakeholders may converge to a shared vision
Applications (different scales)
Platform
(API, protocols, data processing, device management)
Infrastructure/Networking fabric (connectivity, storage)
Physical world (sensors, actuators)
Security & Privacy
5 © 2020 Arm Limited
This approach enables to break up complexity, share resources more easily, and promote interoperability
Advantages of the layer approach
• Allows IoT device manufacturers to focus strictly on improving their performance, power consumption, etc. – expose only well-defined interfaces to software platforms.
• Easier to share and partition strictly the network and computing resources (slicing); reducing the burden on service providers to build and manage networks – Infrastructure/Network as a Service (IaaS/NaaS)
• Enables software/app developers to build applications without having to understand the specifics of a device – Platform as a Service (PaaS)
6 © 2020 Arm Limited
Security challenges
How to secure the entire ecosystems, from hardware to application?
• Hardware isolation (Arm TrustZone)
• Middleware (Speculative Store Bypass Barrier – SSBB)
• Network isolation (Software-defined Networking – SDN)
• Data confidentiality in transit (Transport Layer Security – TLS)
• Software isolation (containers)
End-to-end security not straightforward
7 © 2020 Arm Limited
A practical network-centric view
IP/ MPLS
M2M/ NB-IoT
LTE/ Wi-Fi mesh
Deterministic Ethernet
8 © 2020 Arm Limited
Cellular
PLC
ZigBee
End points
Area Core Data
networks network
center
Cloud vs. Fog vs. Edge
The information processing view
Cloud computing
Cloud dominated the networked systems landscape until recently
All intelligence on powerful servers, including relational databases, control functions, data analytics engines, web interfaces, etc.
End devices merely information gatherers
Might not scale as the number of IoT devices grows, and applications continue to diversify and generate more data
9
© 2020 Arm Limited
Example: Smart metering
Electricity meter
Home hub
Cloud Infrastructure
Energy supplier
• Sensing performed by simple sensors
• Information relayed by home hub over cellular network
• Data processed in the cloud by different stakeholders
Gas meter
Core network/ Internet
Radio access network
Network operator
Third parties
10 © 2020 Arm Limited
Cloud vs. Fog vs. Edge
The information processing view
Pushing some of the intelligence closer to the device, for e.g., to access networks or gateways
This includes data aggregation, compression, (partial) processing, making localized decisions
IoT devices kept simple, no direct communication with end servers, still battery powered
Resource management implemented across different network layers – management could be regarded as an application
Fog computing
11 © 2020 Arm Limited
The role of gateways in fog architectures
Data filtering and processing (for e.g., aggregation of summaries, compression, etc.)
Protocol translation and interfacing between different connectivity technologies
Data flow multiplexing, packet routing
Security (for e.g., data encryption, firewalling)
Scalability problem: as the number of devices grows, so will the number of gateways that are required
12 © 2020 Arm Limited
Example: Home automation
Home appliances
Cellular
Cloud Services
• Gateway performs protocol
translation
• Incorporates basic network intrusion detection system
• Cloud services continue to perform analytics
Fiber
Gateway
Wi-Fi
Internet
13 © 2020 Arm Limited
Smartphones as gateways
The fitness and healthcare domain
• Embed multiple networking technologies (Wi-Fi, 3G/4G, Bluetooth/BLE, NFC, etc.)
• Run full TCP/IP stacks, thus maintain end-to-end connectivity with cloud
• Can connect to multiple devices within close proximity simultaneously
• Ability to enforce secure transport (e.g., TLS/HTTPS)
• Sufficient computing power to pre-process/augment collected data
14 © 2020 Arm Limited
Example: Wearables
Cloud-based fitness tracking service
• Smartphone communicates over BLE with wearable devices
• Performs minimal information pre- processing
• Relays data to cloud-based services
LTE
BLE
Core network/ Internet
Cloud-based image manipulation/facial recognition/navigation services
15 © 2020 Arm Limited
Cloud vs. Fog vs. Edge
The information processing view
Edge computing
Pushing compute power, communication capabilities, intelligence down at device level
Processing as much as possible where data is collected
Transmitting only key information or summaries
Enabling new applications: automotive IoT, virtual/augmented reality, in-ear computing
16
© 2020 Arm Limited
Example: Hearables
Cloud-based services
• Hardware: Low-power chips specialized in computationally intensive tasks (Arm Ethos)
• Software: AI libraries optimized for constrained devices (uTensor)
• Neural networks: compressed/pruned models
LTE
Core network/ Internet
17 © 2020 Arm Limited
Choosing the right IoT architecture
Performance and cost remain the dominant architectural drivers
Fog Nodes
Thousands
Millions
Cloud servers
• What are the application requirements?
• What data needs to be acted on locally?
• Where is most of the computing power?
• How much networking infrastructure should be deployed/used?
Edge Devices
• Where are the trust boundaries? Billions
18 © 2020 Arm Limited
Standards for IoT
Multiple regulation bodies and industry alliances are standardizing the means by which devices can interact with each another and with gateways or cloud services.
The Institute of Electrical and Electronics Engineers (IEEE)
19
© 2020 Arm Limited
• •
•
Primarily dealing with defining protocols for (wireless) access networks
Targeting the Industrial, Scientific, and Medical (ISM) bands (e.g., 2.4GHz, 5GHz, 900MHz in some regions, etc.)
From an IoT perspective, the most relevant technologies include
• IEEE 802.15.4, on which ZigBee builds, and
• IEEE 802.11ah (HaLow) that is an amendment to the IEEE 802.11 specification (typically
used for Wi-Fi) that enables low-power wide-area networking
Standards for IoT
Multiple regulation bodies and industry alliances are standardizing the means by which devices can interact with each another and with gateways or cloud services.
The 3rd Generation Partnership Project (3GPP)
20
© 2020 Arm Limited
• •
Focuses on specifying cellular network architectures and protocols (e.g., GSM, 3G, 4G-LTE, etc.) Developing standards for cellular communications tailored to IoT applications
• LTE-M – compatible with existing LTE networks, easy to roll out, limited to 1Mb/s speeds
• NB-IoT – deployed in same or different frequency bands, lower capacity (200Kb/s), different
modulation and coding schemes, and does not require gateways.
Standards for IoT
The Internet Engineering Task Force (IETF)
• Focuses on specifying protocols that are used across the Internet; these standards are known as Requests for Comments (RFCs)
• IoT relevant standards include
• Addressing/internetworking for low power devices (IPv6 over Low-Power Wireless Personal Area Networks
– 6LoWPAN)
• Routing (Routing Over Low-power and Lossy networks – ROLL)
• End-to-end communications (Constrained Application Protocol – CoAP)
• Security (Datagram Transport Layer Security – DTLS)
• Software updating (Software Updates for Internet of Things – SUIT)
• Also
• Example: The JavaScript Object Notation (JSON) Data Interchange Format – RFC 8259
21 © 2020 Arm Limited
offers experience-based guidance
Standards for IoT
Industry alliances
• Bluetooth – wireless personal area networks (WPANs); defines application profiles
• ZigBee – WPANs building on IEEE 802.15.4; inexpensive consumer/industrial applications
• LoRaWAN – LPWAN based on chirp spread spectrum technology
Collaborative associations
22
© 2020 Arm Limited
• •
The Alliance for IoT Innovation (AIoTI) – European Commission framework supporting interaction between IoT players to drive innovation, standardization, and policy.
Open Connectivity Foundation (OCF) – Industry-led framework aiming to develop IoT standards, interoperability guidelines, and provide a device certification program.
Standards for IoT
Other standardization bodies relevant to IoT
• National Institute of Standards and Technology (NIST) – works on a range of science, technology, and engineering topics
• Example: Advanced Encryption Standard (AES)
• International Organization for Standardization (ISO) – promotes a broad range of proprietary, industrial, and commercial standards
• Example: Internet of Things (loT) – Reference Architecture (ISO/IEC 30141:2018) • International Telecommunication Union (ITU) – recommendations, reference models
• Example: ITU-T Y.4000/Y.2060 – Overview of the Internet of things
23 © 2020 Arm Limited
Choosing among IoT standards is not straightforward
24 © 2020 Arm Limited