Internet of Things

IoT devices and platforms

1
Book
 Levels

•IoT devices,
•IoT network,
•IoT services platform, and
•IoT applications.

• Each level has its own medium and protocols.

• The two main requirements for “Things” in IoT are sensing and addressing.
Sensing is essential to identify and collect key parameters for analysis, and
addressing is necessary to uniquely identify things over the Internet.
While sensors are very crucial in collecting key information to monitor and
diagnose the “Things,” they typically lack the ability to control or repair such
“Things” when overhaul is needed
This raise the question: why spend money to sense “Things” if they cannot be
controlled?
Actuators have been introduced to address this important question in IoT.

With this in mind, the key requirements for “Things” in IoT now consist of
sensing, actuating, and unique identification
 The massive growth of connected things and objects will be sensors and
actuators, so it is important to understand their relationship in the
architecture.
It is important to understand the principles of what is being measured, and
why.

One should ask, "What type of sensor or edge device should I consider for the
problem I'm trying to solve?“

• An architect should consider aspects of cost, features, size, usable life, and
precision when deploying an IoT solution.

• Additionally, the power and energy of edge devices are rarely addressed in
IoT literature, but are critical to building reliable and long lasting
technology.
 Overview

• Device (E.g. IRIS sensor node vs Apple

iPhone)
• Hardware components
• Platforms: Overview of popular hardware
platforms
• Operating systems
• Summary

6
 IRIS Sensor Nodes are true IoT devices focused on sensor data collection and autonomous
network communication.

IRIS sensor node vs Apple iPhone

Apple iPhones can be part of IoT applications as smart controllers or user interface systems that
interact with and manage IoT devices and data, though they are not "IoT devices" in the classic
sense of solely collecting and transmitting data autonomously.

7
 IRIS sensor node vs Apple iPhone 5
• Three-axis gyro
• Accelerometer
• Proximity sensor
• Ambient light sensor
• GPS
• Microphone, Camera
• Touchscreen
IRIS Sensor node iPhone 5

• 8 MHz ≈1 : 125 • A6: 1 GHz, 2CPUs, 3GPUs

• 250 kbps ≈1 : 1200 • 300 Mbps (802.11 n)
• 8 kB RAM ≈1 : 125000 • 1 GB RAM
• 128 kB Flash ≈1 : 500000 • 64 GB Flash
For sensor nodes the reverse of Moore's Law is valid
“cheaper with the same performance”

8
 Hardware
components

9
 Hardware components

Sensor(s)
RAM
Flash
ADC Memory

Controller Radio
IO

Power unit

ADC = Analog-to-digital converter

1
0
 Hardware components

Sensor(s) Actor(s)
RAM
Flash
ADC Memory

Controller Radio
IO

Power unit

ADC = Analog-to-digital converter

11
 Hardware components

Sensor(s) Location finding system

RAM
Flash
Mobilizer
ADC Memory

Controller Radio
IO

Power unit

ADC = Analog-to-digital converter

1
2
 Hardware components

• A (sensor) node consists of several components

• Controller: Central processing unit (CPU)
• Memory for application and data
• Usually different memory types are used for application and data
• Communication interface
• Wireless radio interface with antenna
• Sensors/Actors
• Interface to the real world.
• Monitoring, changing of real world phenomena.
• Power unit
• Power supply for all the components
• Mobilizer
• Location finding system

13
 Hardware components

• Requirements of the application are design decisions for

all components
• Size (form factor)
• Energy consumption
• Costs
• Sensors/Actors

• Any idea how?

14
 Controller

• Four important factors for the controller

• Number of transistors -> size, cost, power
• Number of clock cycles -> power
• Time to develop -> cost, acceptance
• Nonrecurring engineering cost (NRE) -> cost, acceptance

• Ideal: Minimize all factors at the same time!

15
 Controller
• The controller is the Central Processing Unit (CPU) of
the device
• Different architectures possible
• Microcontroller (MCU)
• Resource constrained compared to desktop processors
• Software controlled, general purpose

• Digital Signal Processor (DSP)

• Processing of large data streams, parallelizing
• Hard-wire basic functions

• Field-Programmable Gate Arrays (FPGA)

• Special hardware, expensive
• Limit coding to configuration

• Application-Specific Integrated Circuits (ASIC)

•Special hardware designed only for one application
• Possibly embedding several MCU or DSP cores

16
 Controller [Adams]

Time to Perfor- Price Develop- Power Feature Summary

Market mance ment Ease Flexibility

MCU Excellent Fair Excellent Good Fair Excellent Good

DSP Excellent Excellent Good Excellent Excellent Excellent Excellent

FPGA Good Excellent Poor Excellent Poor Good Fair

ASIC Poor Excellent Excellent Fair Good Poor Fair

17
 Controller
• Tasks of the controller
• Running of (real time) data processing and communication
protocols

• Perform and control the application program

• Energy management of the node

• Different operation modes available (active, idle, listen, sleep, etc.)

18
 Controller

• Energy management of controller

• Example: Power state machine of StrongARM-1100

19
 Controller: Overview of some popular MCUs

Atmega1281 TI MSP430 PIC 18F6720 ARM7TDMI-S ARM926EJ-S

8 bit RISC 16 bit RISC 16 bit RISC 32 bit RISC 32 bit RISC

RAM 8 kB 10 kB 4 kB 98 kB 1 MB

128 kB Program 128 kB Program 512

512 kB Data kB File system 1 kB
Flash (Serial Flash) 48 kB 512 kB 8 MB
EEPROM
4 kB EEPROM

Max. Freq. 8 MHz 8 MHz 20 MHz Up to 72 MHz 400 MHz

Power- Active: 8 mA Active: 1.8 mA 20-144 mA

consumption Active: 8 mA
Sleep: < 8 µA Sleep: 5.1 µA Sleep: 65 µA

TelosB Sun SPOT

Platform IRIS Particle MSB-A2 (Rev 8.0)
MSB-H30

Price ~12 € ~5 € ~5 €

20
 Memory
• Properties Instruction
• Often different program and data memory Memory
-> Harvard architecture
• No memory management unit (MMU) available
• Size of memory varies Controller
• May depend on the application

• Types of memory Data

Memory
• Random Access Memory (RAM)
• Store temporarily sensor data and messages
• Electrically Erasable Programmable Read-Only Memory (EEPROM)
• Stores the program code
• Operates on bytes
• FLASH Memory
• Non-volatile memory
• Stores the program code
• Stores data if RAM too small or protection against power loss required
• Operates on blocks (read, write)
• High delay and energy consumption

21
 Communication interface

• Communication interface is required to exchange data

with other devices
• Typical communication media
• Radio
• Infrared (IR)
• Directed / undirected
• Ultrasonic
• More in military applications
• Sound
• Underwater communication

• Communication interface has high energy consumption

• Radio interface consumes the most energy usually

22
 Communication interface: Radio

• ISM bands defined by the ITU.R

Frequency Range Bandwidth Center Frequency
6.765 MHz 6.795 MHz 30 kHz 6.78 MHz
13.553 MHz 13.567 MHz 14 kHz 13.56 MHz
26.957 MHz 27.283 MHz 326 kHz 27.12 MHz
40.66 MHz 40.70 MHz 40 kHz 40.68 MHz
433.05 MHz 434.79 MHz 1.84 MHz 433.92 MHz
902 MHz 928 MHz 26 MHz 915 MHz
2.4 GHz 2.5 GHz 100 MHz 2.45 GHz
5.725 GHz 5.875 GHz 150 MHz 5.8 GHz
24 GHz 24.25 GHz 250 MHz 24.125 GHz
61 GHz 61.5 GHz 500 MHz 61.25 GHz
122 GHz 123 GHz 1 GHz 122.5 GHz
244 GHz 246 GHz 2 GHz 245 GHz

23
 Communication interface: Radio

• Overview of popular radio interface technologies

• IEEE 802.15.4 / ZigBee
• IEEE 802.11b (WLAN)
• Bluetooth
• 868 MHz Derivate (e.g. TR1001)

24
 Communication interface: Radio

ZigBee / 868 MHz Near field

Bluetooth Derivative IEEE 802.11b
IEEE 802.15.4 communication

Frequency 2.4 GHz 2.4 GHz 868 MHz 13.56 MHz 2.4 GHz

Data rate 250 kbps 732 kbps-2.1 Mbps 40-150 kbps 424 kbps 11 Mbps

Communication 300 m 300 m Outdoor

30-100 m 1-100 m Outdoor <10 cm 90 m Indoor
range

Energy
consumption (send, 10-20 mA 30 mA 25 mA <15 mA 120-180 mA
receive)

Energy
consumption 20 µA
(sleep)

25
 Sensors / Actors
• Interface between digital and real world

Communication with
sensor network and/or
user

Control

Actor
Sensor

Process /
Environment

26
Sensors / Actors

27
 Sensors / Actors
Example: Smart Greenhouse
Blinds
Light sensor
Lamps

Window

Plants with individual

Temperature water supply
sensor

Water pump

Humidity sensor

Watering logic

28
Sensors
Calibration?

• Different criteria for classification

• Context (activity, position)
• Used technology (infrared, ultrasonic, induction, sound)
• Technical aspects (magnetic, mechanical, electrical)
• Logical operation (integrated, active, passive)
• Measured value (temperature, pressure, acceleration)
• Spatial distance (contactless, contact, omnidirectional)
• Information dimension (one dimension, multidimensional)
• Application area (…)

• Analog-digital conversion of sensor data required

• Analog Digital Converter (ADC) required
• Often integrated into the microcontroller

29
 Actors

• Purpose
• Translate signals to actions
• Counterpart to sensors
• Event generation by actors
• Actors are usually developed for one applications
• Example: Water pump in greenhouse

Pump
Adapter to
sensor node

Power supply

30
 Power supply

• Energy is the most critical resource in a battery

operated device (sensor node, “thing”)

• Energy supply has two tasks

1)Provision of electrical energy
• Typically with batteries
• In future also fuel cells

2)Conversion of other energy forms

• Extraction of energy from the environment
• -> Energy harvesting

31
 Energy consumption of sensor nodes
• Energy characteristic of the components
• Radio interface consumes the most energy
• Ratio of energy requirements of CPU / radio interface
• Model
E(1 Instruction of CPU) : E(Sending of 1 bit) ≈ 1:1500 – 1:2900
• Send and receive operations are roughly equal expensive
• Best energy consumption reduction: switch-off radio

• Flash-Memory has high energy requirements

• Write operation ~900 times more expensive than read operation
• Processor not so critical
• Typically several power modes available
• -> Adaption of energy consumption to the operation
• Sensors / Actors
• Varies between components and difficult to predict

32
 Example: Ratio of energy consumption

• Sensor node
• iBadges
• 8-Bit RISC
Microprocessor
(ATmega128L Atmel)
• Bluetooth

• Data analysis on the sensor node or base station

• Tradeoff between communication and processing
• Ratio of energy consumption

EDSP : EATMEGA : EBT = 1 : 4.75 : 1585

• -> Reduction of energy by more processing and less

data transmission (Bluetooth)

33
 Provisioning of energy

• Batteries
• Cheap
• Easy to use 5 cm

• Rechargeable batteries available

• Lithium Ion batteries with high capacity
• Charging is simple and easy 1.4 cm

• Size of batteries is a problem

• AA battery defines the size of many devices
• Environment (temperature) has influence on the capacity

• In future fuel cell possible

34
 Energy harvesting

• Capacity of battery limits the lifetime of the device

• Battery depletion -> Device cannot work What is the
-> (Sensor) network cannot work network
• Idea: recharge the batteries during operation lifetime?
• Use energy from the environment

• Current approaches
• Photovoltaic
• Solar modules for sensor nodes
• Thermoelectric generators
• Conversion of temperature differences to energy
• Kinetic energy conversion
• Piezo-electric principle already tested for shoes
• MEMS gas turbines
• Convert air- or fluid streams

35
 Photovoltaic [Krüger]

• Solar modules for sensor nodes

• Additional controller for the energy management

Battery
Solar panel
Management unit

Sensor node with solar panel

36
 Photovoltaic

• Advantages of photovoltaic
• Energy comes for “free” and is “available unlimited”

• Disadvantages
• Solar panels are expensive (~100 €)
• Depends on season/weather
• Depends on location
• Shadow vs sun
• Angle of the sun rays important
• Solar panels are large in comparison to device

37
 Thermoelectric generators [micropelt]

• Use of temperature difference for energy conversion

based on Seebeck effect
• Use of thermo electric generators

38
 Thermoelectric generators

TE-Power PLUS Evaluation Kit

MPG-D751 Thermo generator Chip.

39
 Thermoelectric generators

• Advantages
• Cheap
• Energy comes for “free” and is “available unlimited”

• Disadvantages
• Temperature difference has to be known and large
• 35°C ~3600 mAh (or 2-4 AA Batteries)

40
 Kinetic energy conversion

• Use of kinetic energy for energy-harvesting

• Various forms of kinetic energy
• Rotation
• Linear movement
• Vibration/impacts
• Flow

41
 Kinetic energy conversion

• Advantages
• Cheap
• Energy comes for “free” and is “available unlimited”

• Disadvantages
• Continuous movement required

42
 Other approaches

• Piezo-Effect
• e.g. in shoes

• Radio waves
• Sensor prototypes exists

• Laser, acoustic, …

• What else can be used?

43
Platform
s

44
Platforms: Overview of popular platforms
Platform Microcontroller Radio chip
AVRraven ATMEGA1284p, ATMEGA3290p AT86RF230

BTnode ATMEGA128L CC1000

EyesIFX v1 / v2 MSP430F149, MSP430F1611 TDA5250

iMote, iMote2 ARM core, ARM 7TDMI, ARM 11 Bluetooth, CC2420

Lotus ARM7 Cortex M3 AT86RF231

MICA, MICA2/MICAz, ATMEGA103, ATMEGA 128, TR1000, CC2420, AT86RF230

Cricket, MICA2Dot, IRIS ATMEGA128L, ATMEGA1281
Particle, DINAM PIC18F720, PIC18F14K22 TR1001

ScatterWeb MSB MSP430F1612IPM CC1020

Shimmer MSP430F1611 CC2420

SunSPOT ARM926EJ-S CC2420

Telos, TelosB, T-Mote Sky MSP430 CC2420

Tinynode MSP430 SX1211

45
 Platforms: IRIS Motes
• Processor
• XM2110CA based on Atmel “IRIS” without batteries
ATmega1281
• 8 bit Microcontroller
• 8 MHz
• Compute power similar to 8088
CPU from the original IBM PC
(~1982), but reduced energy
• Radio chip consumption
• AT86RF230
• IEEE 802.15.4
• ZigBee compatible
• 2.4 GHz, 250 kbps,
up to 300 m (outdoor), up to 50 m
(indoor)

•Memory
• 4 kB EEPROM
• 8 kB RAM
• 128 kB program Flash Memory
• 512 kB measurement Flash Memory

46
 Platforms: IRIS Motes
• Size and weight
• 5.8 x 3.2 x 0.7 cm
• Without batteries & sensor board: 18 g

• Energy supply
• 2 x AA batteries to provide required voltage
• Microcontroller: 8 µA (sleep), 8 mA (active)
• Radio chip: 15.5 mA (receive), 17.4 mA (send), 1.5 mA (idle), 0.02 µA (sleep)

• Periphery
• Sensor boards, UART, 10bit AD-converter, Digital IO, I2C, SPI Bus, JTAG ICE, 51 Pin
Connector

• Producer
• MEMSIC (formerly Crossbow)

• Programing and accessories

• NesC (C-Derivat) -> TinyOS Operating System
• MoteWorks: Platform for developing sensor network applications

• Costs
• 134$ / without sensor board

47
 Platforms: IRIS Motes

• Sensors
• Onboard: 3 LEDs
• Additional sensor boards
• Light
• Temperature
• Humidity
• Barometric pressure
• Accelerometer “MTS400/420”
• GPS
• Analog and digital I/O interfaces
• Memory for data

48
 Platforms: Particle
• Processor
• PIC 18F720
• 20 MHz
• 8 bit
• Radio chip
• TR1001
• 125 kbps
• 868.35 MHz
• Memory
• On-Chip
• 4 kB RAM
• 128 kB Flash
• 1 kB EEPROM
• Extern
• 500 kB Flash for data
• File system: ParticleFS
• Size
• 4.5x2.7 cm incl. batteries

49
 Platforms: Particle
• Energy supply
• Flexible w.r.t. voltage: 0.9 V – 3.3 V
• AAA, 2xAAA, AA
• Periphery
• 21 pin interface with I2C, SPI, Serial (625 kbps), Parallel bus,
Analog interface, Interrupt interface, Digital I/O
• Programming
• C
• Producer
• TeCO, KIT (formerly spin-off: Particle Computer GmbH)

• Costs
• 120 € per node, with batteries 130 €

50
 Platforms: Particle

• Sensors and actors

• LCD
• Accelerometer
• Pressure sensor
• Light- / IR
• LEDs
• Temperature
“Particle LCD AddOn”
• Microphone
• Analog interface
• Digital I/O
• Speaker

“spart Sensor Board”

51
 Platforms: Arduino

• Simple platform for

sensing and control the
environment
• Open source hardware
• Hardware
• Processor:
ATmega328 (Atmel)
• Clock Speed 16 MHz
• Digital I/O Pins 14
• Analog Input Pins 6
• Flash Memory 32 KB
• SRAM 2 KB
• EEPROM 1 KB

52
 TelosB
•Processor: TelosB uses a Texas Instruments MSP430
microcontroller, which is known for its low power consumption.

•Radio: Equipped with an IEEE 802.15.4 compliant RF transceiver

(Chipcon CC2420), supporting the ZigBee protocol.

•Memory: Features 10KB RAM and 48KB of flash memory.

•Sensors: Integrated with temperature, humidity, and light sensors.

•Power: Typically powered by AA batteries or a USB connection,

designed for low energy consumption to extend operational periods.

•Usage: Widely used for its compact size and integrated sensors,
making it suitable for environmental data collection and wireless
sensor network research.
Operating systems

54
 Operating systems

• An operating system (OS) acts as a resource manager

• OS multiplexes system resources in
• time and
• space

• Properties of an OS
• Architecture
• Programming model
• Scheduling
• Memory management and protection
• Communication protocols
• Resource sharing
• Support for real-time applications

55
Operating systems: Properties

56
 Operating systems: TinyOS

• Open source, flexible, component based, and application-

specific operating system
• Footprint of 400 bytes
• Monolithic architecture using component model
• Components have three computational abstractions
• commands
• events
• tasks
• The TinyOS component library includes
• network protocols
• distributed services
• sensor drivers
• data acquisition tools

57
 Operating systems: Contiki

• Lightweight open source OS written in C

• A typical Contiki configuration consumes 2 kB of RAM
and 40 kB of ROM
• Event-driven kernel
• Preemptive multitasking
• Stack-less and lightweight threads: Protothreads
• Modular architecture
• Dynamic memory management
• Rich set of communication protocols -> IPv4, IPv6,
TCP
• Additional stack: Rime

58
 Operating systems: MANTIS

• MultimodAl system for NeTworks of In-situ wireless

Sensors (MANTIS)
• MANTIS Operating System (MOS)
• Layered architecture
• MOS kernel handles only Layer 3:
the timer interrupt
• Preemptive multitasking Layer 2:
• Number of threads fixed
• Default thread stack in Layer 1:
MOS is 128 bytes
• MOS kernel needs 500 bytes
• Dynamic memory management
Layer 0:

59
 Operating systems: Nano-RK

• Preemptive multitasking real-time OS for WSNs

• Easier to program, since known from traditional OS
• Priority scheduling at two levels
• Nano-RK uses 2 kB of RAM and 18 kB of ROM
• Nano-RK supports hard and soft real-time applications
• Monolithic kernel architecture
• Static memory management
• Lightweight networking protocol stack

60
 Operating systems: Summary

OS/ Architecture Programming model Scheduling Support for Real-time

Feature Applications
TinyOS Monolithic Primarily event Driven, FIFO No
support for TOS threads
has been added

Contiki Modular Protothreads and events Events are fired as they No

occur. Interrupts execute
w.r.t. priority.

MANTIS Layered Threads Five priority classes and To some extent at process
further priorities in each scheduling level
priority class (implementation of priority
scheduling within different
processes types).

Nano-RK Monolithic Threads Rate monotonic and rate Yes

harmonized scheduling.

LiteOS Modular Threads and Events Priority based round robing No

scheduling.

µkleos Microkernel Threads Tickless, preemptive No

scheduling with priorities.
 Operating systems: Summary

OS/ Memory Management and Protection Communication Protocol Resource Sharing

Feature Support
TinyOS Static memory Management with memory Active Message Virtualization and
protection. Completion Events

Contiki Dynamic memory management and linking. No uIP and RIME Serialized Access
process address space protection.

MANTIS Dynamic memory management supported but use At kernel level COMM layer. Through semaphores.
is discouraged, no memory protection. Networking Layer is at user
level. Application is free to
use custom routing
protocols.

Nano-RK Static memory management and no memory Socket like abstraction for Serialized access through
protection. networking. mutexes and semaphores.
Provide an implementation
of priority ceiling algorithm
for priority inversion.

LiteOS Dynamic memory management and it provides File based communication. Through synchronization
memory protection to processes. primitives.

µkleos Static and dynamic memory management, no 6LoWPAN w/ TCP/UDP Through mutexes.
memory protection.
 Summary

• Embedded systems are “different”

• Important criteria -> Hardware
• Nodes have reduced resources
• Energy is the resource
• Radio interface is the energy consumer
• Radio switch-off is the best way to save energy
• Typical dimensions of resources
• Memory (8 KB – 1 MB RAM)
• Operating frequency (8 MHz – 400 MHz)
• Radio interface (40 kbps – 2 Mbps)
• Energy (8 µA – 100 mA)
• Various platforms are available
• Development tools are not standardized

63
 Summary

• Operating systems are also “different”

• Desktop OS vs Embedded OS vs WSN OS
• Energy efficiency
• Real-time capability
• Programming model
• Hardware is more visible

64