Lecture Notes Architecture

• no standard

• different challenges than IT

• data type may be inportant

how the data is transported, collected, analysed and acted on

main challenges:

• scale

• security

• constraint and networks

• data

• legacy support

• only IPv6

  • otherwise not enough addresses

• many millions of routable IP endpoints

• endpoints often not behind a firewall

  • also physical accessable

• require consistens mechanisms of authentication, encryption and intrusion prevention

• most things are designed for a single job

• can be deployed anywhere —> lossy network

• unstrucktured data

• need to filter data

• store hug amount of data

• long ass device support as machinery is only upgraded if totaly necessary

• 
  

machine to machine communication

authority: European Telecommunication Standards Institue

common service layer that allows communication with application servers

focuses on services, applications and plattforms

horizontal plattform architecture

restfull APIs allow interoperabilty of networks normally not able to communicate

application layer protocols

industry specific

individual data models

framework accros vertical industires

physical network the IoT nodes run on

management protocols

devices

endpoints

IEEE 802.15.4, 801.11ah, 1910

gateway devices

7 layer

includes edge comupting and data storage

technical perspective

each layer has specific functions and security

decomposes the problem into smaler parts

identify different technologies for each layer

parts can come from different vendors

things (endpoints, sensors, actuators)

generate data

beeing controlled over the network

reliable and timly transmission of data (between layer 1 devices & network and network & processing at layer 3)

all network elements (no distinction between last mile and backhaul)

data reduction

converting data flows into information

makes data ready for processing and storing

filter and aggregate data

stores data so that it is usable by application when needed

recompiles dataformats

ensures consistent semantics

confirms that data is complete

interprets data

provides repots/alerts

consumates and shares application layer data

OT: data in motin

IT: data at rest

due to complexity data managment and application need to be deplyed thoughout the layers

ˆThings layer: At this layer, the physical devices need to t the constraints of the envi-ronment in which they are deployed while still being able to provide the information needed.

Communications network layer: When smart objects are not self-contained, they need tocommunicate with an external system. In many cases, this communication uses a wirelesstechnology. This layer has four sublayers.

Application and analytics layer:At the upper layer, an application needs to process thecollected data, not only to control the smart objects when necessary, but to make intelligentdecision based on the information collected and, in turn, instruct the things or othersystems to adapt to the analyzed conditions and change their behaviors or parameters.

• Access network sublayer: last mile

• Gateway and backhaul network sublayer: multiple smart object into area, gateway needs to forward data to central station for data processing

• network transport sublayer: e.g. IP, UDP

• IoT network managment sublayer: protocols for data center to exchange data with sensors

• battery-pwoered or power-connected

• mobile or static: thing can move or not

• low or high reporting frequency: how often does data need to be sent

• simple or rich data: quantity of data exchange each report cycle

• troughput: reporting frequncy and data quantity

• report range: distance to the gateway

• object density per call: nodes per gateway

3 sub layers

• access network: last mile

• backhaul network: link between access and core, also the small subnets at the edge

• core network: interconnects networks, path between different subnets

depending on use case different technology

• range

  • personal area network (pan): few meters (bluetooth)

  • home area network (han): few tens of meters (zigbee, bluetooth low energy)

  • filed area network (fan): several tens to several hundred meters

  • local area network (lan): up to 100 m (when standard networking technologies are used)

  • metropolitan area network and wide area network

  • w at the beginning to indicate wireless

more throughput or longer range means more power needed

responsible to forward data from the sensor to the central station

gateway is most often static or position relative is static

dedicated short range communication: sensor and gateway are on the same thing and move together

network protocol should be used that allows for flexible communication

best if open and standard based

must be scalable and secure

IP does all this with UDP or TCP as layer above

transmission between smart object and other systems (may be http)

http is a fat protocol not designed for constrained devices

CoAP (constrained application protocol) solves this. uses get, post, put and delete within a smaller header and runs in UDP adds observation

observation allows streaming of state changes as they occur

MQTT (message queue telemety transport) uses broker based architecture

• analytics applications: collects data from multiple smart objects processes it and displays it

• control applications: controls behaviour of smart object (lights on/off, heating on/off) can even be automated based on time/sensor data/manual input

Data analytics:

• simple: dashboard displays current data

• complex: from current data a trend is extrapolated

network analytics:

• loss of conectivity

complexity of analytics depending on use case and need

• amount of data generated is biggest challenges

• most of the data is polling data with very little use on its own

  • combinded with other data it can paint a picture

• processing is not happening on the device but in the cloud

  • long latency

  • needs huge bandwith

data managment as close to the edge as possible

any device with a computing, storage and network connectivity can be used

minimizes latency and network traffic

characteristics:

• contextual location awarness and low latency

• geographic distribution

• deployment near IoT endpoints

• wireless commiúnication between fog and IoT endpoint

• use for real time interactions

each layer to compliment each other not to replace it

time sensitive data is analysed on the edge or fog

data that can wait some seconds can be passed to aggregation node

data that is less time sensitive is passed to the cloud

any physical object that contain embedded technology that can interact with their enviroment in a meaningful way by being interconected and enableing communication among themselves or external agent

real value by conectiong sensors with actuators

ITU definition

• RFID tags

• Sensor technologies

• Smart technologies

• nanotechnologies

connects objects to databases

provide unique identification or mini database

currently Electronic Product code is the dominant standard

does not need a line of sight

range between 15 and 20 cm to about 1m newer tags up to 7m

can be deployed everywhere

function of input device

incorperate sensors and actuators to act on stimuly

three kinds

• passive: respond directly to stimuli without processing any signal

• active: can sense a signal and determine how to response to that

• autonomous: fully integrated controllers, sensors and actuators

focus on design, characterisation, production and application of structures and devices through manipulation of materials at the nanoscale

measures some physical quantity and converts measurement into digital representation

representation is passed on to another device for transformation into usefull data

categories

• active or passive: produce an energy output and need an external power supply (active) or not (passive)

• invasive or non-invasive: part of enviroment they measure (invasive) or not (non-invasive)

• contact or no-contact: need physical contact or not

• absolute or relative: measure an absolute scale or difference with a fixed variable reference

• area of application: based on the industry they are beeing used

• what they mesure

natural complement to sensors

receive some type of control signal that triggers physical effect

categorasation:

• type of motion: linar, rotary,…

• power: high or low power output

• binary or continuous: number of stable states

• area of application: based on specific industry they are used

• type of energy:

integrate and combine mechanicla elements as sensors and actuators on a very small scale

needs microfabrication technique similar to microelectronc integrated circuts

• processing unit: some type of processing unit for acquring, processing and analysind data. Most common is a micro controller because of the size, flexibiltiy, smplicitiy, low power and low costs.

• sensor or actuator: capable of interacting with physical world

• communication device: responsible for connecting device with outside world, can be wired or wireless

• power source: needs most power to communicate

network of sensors

communicate with each other in a productive manner

two communication patterns:

• event driven

• periodic