IoT Architecture

• scale

• constrained devices

• security constrained communication

• latency

• data

• legacy device support

IT: several thousand devices

IoT: few thousand to few million devices

IT: powerfull devices and networks

IoT: limited devices and lossy networks with low data rates

huge amount of data in IoT

core functionalty of IoT

IT: protected by firewalls and physical secure

IoT: wirelessly connected and easily accessable

Industrial, Science and Medical used spectrum of unlicensed spectrum

IT: upgrade or access denied

IoT: long lifetime of devices therefore legacy protocols and interfaces

• oneM2M IoT

  • launched 2008

• IoT World Forum

  • launched 2014

• telco standardisation organisations

• Application layer: industry specific

• Service layer: horizontal framework including hardware and physical network

• Network layer: devices, communication liks and gateways

Horizontal framework and Restfull APIs allow applications to interface with the IoT devices

IoT World Forum

companies like

• Cisco

• IBM

• Rockwell Automation

7 layers

1. Physical devices and controllers: things

2. Connectivity: reliable and timely transmission of data

3. Edge Computing: fog layer, first step to work with data, mostly reduce data flow

4. Data Accumulation: captures data and stores it so it is usable by applications, converts eventbased to query based

5. data abstraction layer: reconciliation of multiple data formats to complete data sets

6. application layer: interprets data, may monitor, control and provide reports

7. collaboration and process layer: delivers benefits of IoT

1-4 are domain of the OT

4-7 are IT

seperates core IoT and data managment into parallel aligned stacks

core IoT functional stack constists of things, communication network and application

because analytics and apllication do not only exist in the cloud but are deployed thoughout the archtecture

each layer can be broken down further

communications layer:

• access network:

• gateways and backhaul: Ethernet, WiFi (2.4, 5 or sub 1GHz), WiMax, Cellular

• network transport: mostly open protocols to be able to scale and have security

• operations and managment sublayers: CoAP (similar to Http but not as fat), MQTT (open protocol for M2M uses broker based architecture)

applications layer:

• vertical specific apps: controls behaviour of things

• analytics: collects and processes data, displays information

layer 1: things, self contained or interacting with an external system

• low or high power, mobile or static, low or high data generation, simple or rich data, report range, density per cell

layer 2: wifi, lora, routers, IP, UDP, TCP

1. Requirements for mobility and data transmission

2. Formfactor and capabilities of the device

3. Range between device and collector

4. Quantity of objects inside

5. Costs

• point-to-point: one deivce can only talk to one gateway, in IoT context mostly means that each device has established connection with gateway

• point-to-multipoint: one device can talk to multipe, one gateway talks to more than one thing, nodes may have different functionalties (thing can only send, gateway can accumulate and forward)

• (clustered) star topology: needed when not all devices implement the full protocol stack and therefor can not form a peer-to-peer network. Each thing can only communicate with full stack device and not with each other. Clustered star have some full stack devices connected to each other, otherwise it is just a full funcitional device in the center.

• mesh: more than one path between nodes, fault resistance, but higher energy consumtion.

An MQTT system consists of clients communicating with a server, often called a "broker". A client maybe either a publisher of information or a subscriber. Each client can connect to the broker.Information is organized in a hierarchy of topics. When a publisher has a new item of data todistribute, it sends a control message with the data to the connected broker. The broker thendistributes the information to any clients that have subscribed to that topic. The publisher does notneed to have any data on the number or locations of subscribers, and subscribers in turn do not haveto be configured with any data about the publishers.

data analytics:

• process data collected by smart objects

• provide view releated to IoT systems

• value of IoT

network analytics:

• fault/event managements

• performance management

• detailed analysis about network performance

this stack contains of the cloud, fog and edge

cloud beeing the same as the cloud in traditional computing,fog being a middle layer between edge (the device) and cloud, this may be a router or some dedicated near by device to handle some of the data and logic to have a lower latency and edge beeing the device itself.

cloud, fog and edge

data management in traditional IT systems is very simple

Endpoints communicate with a server other IP

the links to the servers typically have a high enough bandwith to answer the requets they get from clients.

minimises latency, conserves network bandwith

backhaul from gateway can be unreliable

faster responses to data with fog as nearer to the device

reevaluate and reformat data for higher levels

filter data and reduce traffic

asses data for laters and fast responses