Internet of Things (IoT)

• IoT system applications

• IoT system architectures

References: Wolf Text: Chapter 8

ARM SoC/IoT Presentations

https://pixabay.com/en/network-iot-internet-of-things-782707/
ARM: Making Things Smart, Connected and Interactive

26 Billion
installed units
by 2020*

Device Categories *Gartner

Internet of Things (IoT)
Safer/Smarter Automotive
What is it?
Smart Appliances Fitness / Healthcare
• “The Internet of Things (IoT) is
the interconnection of uniquely
identifiable embedded computing
devices within the existing
Smart
Portable and
Wearable
Internet infrastructure.”
Farming Electronics wikipedia.org

Resource
• Buzzword, trend, convenient
Smart
Management categorisation, industrial and
Lighting
consumer

Industrial Internet Smart Home

Machine to Machine
Internet of Things (IoT)
Safer/Smarter Automotive

Why Now?
Smart Appliances Fitness / Healthcare • Embedded chips are
becoming:
• Cheaper (<50c)
Portable and
Smart
Farming
Wearable • Smaller (<1mm2)
Electronics
• Lower power (µW)
• Commoditised HW and SW
Resource
Smart
Lighting
Management • Communication is growing
faster (broadband)
• New socio-economic demands
Industrial Internet Smart Home (globalisation, competition, mobility)
Machine to Machine
Internet of Things (IoT)
Safer/Smarter Automotive
Socio-Economic Benefits
• Automation (higher productivity)
• Smart monitoring, control and
Smart Appliances Fitness / Healthcare
maintenance (higher efficiency, lower
cost, higher quality, better
optimisation/outcomes)

Smart
Portable and
Wearable
• Better safety (early warning)
Farming Electronics • Higher responsiveness (dynamic
response to varying demands)
• Huge and varied applications in
Smart
Resource
Management industry, agriculture, health,
Lighting
transport, infrastructure, smart
living, consumer etc.
Industrial Internet Smart Home
Machine to Machine
 IoT system applications
• Soft real-time networked
embedded system.
• Input devices: tags, sensors, etc.
• Output devices: motor controllers,
displays, etc.
• Examples:
• Computer-readable identification
code for objects.
• Appliances controlled by cell phone
interface. https://www.quora.com/How-is-IOT-useful-in-manufacturing
• Sensor network with analytics.

Computers as Components 4e © 2016 Marilyn Wolf

Devices
• People: • Objects:
• Interior: temperature sensor, etc.
• Implanted devices in the body.
• Exterior: RFID, etc.
• Wearable devices on the body.
• Environmental: camera, motion
• Environmental devices outside the
sensor, etc.
body.

https://www.meddeviceonline.com/doc/how-revolutionizing- https://communicationandmediastudies.wordpress.com/2012/10/24/
healthcare-0001 the-internet-of-things/
Computers as Components 4e © 2016 Marilyn Wolf
 Connecting the Physical and Digital Worlds

Sensing and Controlling Wireless Network Cloud

• Integrated sensors, memory • High throughput networks • High performance efficient servers
and processing • High capacity storage
• Low power wireless
• Low power systems • Software as a service
networks
• Little Data • Big Data

Things (“Edge” Devices)

IoT system architectures
edge device hub UI
• Edge: I/O devices.
status request
• Cloud: centralized processing.
• Smart appliance = connected
status
appliance + network + UI.

command

Computers as Components 4e © 2016 Marilyn Wolf

IoT system architectures, cont’d.
• Monitoring system = sensors + network + database + dashboard.
edge device hub database data analysis dashboard

sample

update

alert
Computers as Components 4e © 2016 Marilyn Wolf
IoT system architectures, etc.
• Control system = sensors + database + controller + actuator.

edge sensor hub database controller edge actuator

sample

command

Computers as Components 4e © 2016 Marilyn Wolf

 Things: Basic Building Functional Blocks

M
Sense Compute Control Store Communicate
Unlock a greater potential with custom SoCs

From PCB to custom SoC Increase margins by reducing

 Cost
 Complexity
 Size

Enhance designs with greater

 Efficiency
 Reliability
 Differentiation
 IP protection
Cortex-M: Scalable, compatible and trusted
Cortex-M7
Maximum High
performance, performance
control and
DSP

Cortex-M3 Cortex-M4 Cortex-M33

Flexibility, Performance
Mainstream
Performance control and efficiency
control and
efficiency DSP with
DSP
TrustZone

Cortex-M0 Cortex-M0+ Cortex-M23

TrustZone in Lowest
Lowest cost, Highest energy power & area
smallest area,
low power efficiency lowest power

ARMv7-M
ARMv8-M
ARMv6-M
ARM Cortex-M0 DesignStart Processor
• Subset of the full ARM Cortex-M0
• Low gate count, 32-bit processor, 3-stage pipeline
• Implements ARMv6-M architecture
• Can achieve around 0.9 DIPS/MHz
• Provided as synthesizable Verilog model
• CPU contained in top-level macro-cell “CORTEXM0DS” (instantiated in the
SoC system model) and submodule “cortexm0ds_logic” (pre-configured,
obfuscated)
• Top-level macro-cell implements memory and system bus interface
compatible with AMBA 3 AHB-Lite specification, including interrupt and
event inputs, 3 status outputs, and an event output.
• DesignStart Kit includes:
• Simulation testbench, a set of AHB-Lite peripherals, example SoC systems
IoT Demo Architecture
The “THING”
Sense/Compute/Control/Store/Communicate
The “NETWORK” The “CLOUD”
Sensor:
Accelerometer Low-power Server
Bluetooth
Cortex-M0 Radio
Sensor: Software,
SoC
Ambient Light Bluetooth Storage

User Interface
Display received
data in a PC
• Periodically sense, encrypt, send ambient light Labview window.
and acceleration data to “The Cloud”.
• Based on ARM SoC LiB and Cortex-M0 CPU.
• Platform: FPGA board
(Digilent Nexys4 DDR, Numato Labs Mimas V2)
IoT SoC Application
• Periodically capture sensor data
• Read ambient light sensor data
• Read X-axis/Y-axis/Z-axis acceleration data
• Sample at 1Hz frequency (timer interrupt-driven)
• Encrypt sensor data
• Tiny Encryption Algorithm (TEA)
• Encrypt before sending (for debug - decrypt back to original data if Switch 1 on)
• Transmit data via Bluetooth to server
• Simulated wireless network and “Cloud” server
• For debug (Switch 0 on) transmit via hard-wired USB to server
• Display sensor data in server terminal window
IoT Demo Sensors & Communication
• ADXL362 3-axis Accelerometer
• 12-bit X/Y/Z axis values + 12-bit temperature
• On Nexys4 DDR board
• SPI interface
• PmodALS Ambient Light Sensor
• Vishay Semiconductor TEMT6000X01 ambient light sensor
• Texas Instruments ADC081S021 analog to digital converter
• SPI interface
• PmodBT2 Bluetooth Interface
• Roving Networks RN-42 Bluetooth (2.1, 2.0, 1.2, 1.0)
• UART interface
IoT SoC Hardware
• ARM Cortex-M0 “Design Start” CPU
• Program and data in distributed/block RAM in FPGA
• Peripherals: basic I/O, timer, UART, SPI (all except SPI in the SoC LiB)
• Sensors accessed via SPI
• Wireless and wired communication via UARTs
• CPU and peripherals interconnected via AHB-Lite bus

System on Chip
ARM Cortex-M0
Processor
ARM AMBA 3 AHB-Lite System Bus
Control signals
32-bit Address bus
32-bit Data bus

Program
SPI Timer UART GPIO 7-segment
Memory SPI UART
Peripheral Peripheral Peripheral Peripheral Peripheral
(On-Chip) Peripherals Peripherals

PmodBT2
PmodALS ADXL362 UART to USB LEDs 7-Segment
Bluetooth
Light Sensor Accelerometer Converter Switches Display
Radio
Things: The Security Challenge
Flexible embedded device security

• Security important as more embedded devices become connected

• Even the smallest of devices need to

• Safely store and process secrets Non-trusted
• Have secure communications (i.e., encryption)
• Offer trust in the integrity of the device and its software
• Be able to isolate trusted resources from non-trusted
Trusted
• Reduce attack surface of key components

… without compromising on latency, determinism or footprint.

ARM TrustZone Technology
Bringing ARM security extensions to the embedded world

• Optional security extension for the ARMv8-M architecture

• Security architecture for deeply embedded processors
• Enables containerisation of software
• Simplifies security assessment of embedded devices.

• Conceptually similar and compatible with existing TrustZone technology

• New architecture tailored for embedded devices
 Preserves low interrupt latencies of Cortex-M
 Provides high performance cross-domain calling.
ARMv8-M Additional States
Existing handler and thread modes mirrored with secure and non-secure states
 Secure and Non-Secure code run on a single CPU
• For efficient embedded implementation. Handler
Mode

 Secure state for trusted code Thread

• New Secure stack pointers for robust operation Mode
• Addition of stack-limit checking.
ARMv7-M
 Dedicated resources for isolation between domains
• Separate memory protection units for Secure and Non-secure Non-secure Secure
• Private SysTick timer for each state. Handler Handler
Mode Mode
Non-secure Secure
 Secure side can configure target domain of Thread Thread
interrupts. Mode Mode
ARMv8-M
ARMv8-M Programmers’ Model Memory Map
Secure state view
ROM tables Non-secure MPU alias  Secure memory view permits
System control Non-secure SCB alias access to Secure Flash, RAM,
Non-secure SysTick alias and peripherals.
Off-chip Device
Debug
Off-chip RAM Security Attribution Unit
Peripherals
Secure MPU  Load/store access to all regions
Secure SCB is possible from Secure state.
Secure Peripherals NVIC
RAM Secure SysTick

Secure RAM
ITM/DWT/FPB  Security of regions can be
configured using the
Flash
Security Attribution Unit (SAU).
Secure Flash Secure vector table
ARMv8-M Interrupt Security
High-performance interrupt handling with register protection
 Subject to priority, Secure can interrupt Non-secure Interrupt
Non-secure and vice versa Running Secure
• Secure can boost priority of own interrupts Code
• Uses current stack pointer to preserve context.
Push All Registers
 Uses ARMv7-M exception stacking
mechanism Pop All Registers Zero All Registers
• Hardware pushes selected registers. Switch to Switch to
Secure Non-secure
 Non-secure interruption of Secure code
• CPU pushes all registers and zeroes them Run Non-Secure
• Removes ability for Non-secure to snoop Handler
Secure register values. Return from Interrupt
Security Defined by Address
All transactions from core and debugger checked Request from CPU
 All addresses are either Secure or Non-secure.
System
Level
Control
 Policing managed by Secure Attribution Unit (SAU) Security
• Internal SAU similar to MPU Attribution
• Supports use of external system-level definition Unit (SAU)
• E.g. based on flash blocks or per peripheral.
Non-Secure Secure
MPU MPU
 Banked MPU configuration
• Independent memory protection per security state.

 Load/stores acquire NS attribute based on address

• Non-secure access attempts to Secure address = memory fault. Request to System
Cross-Domain Function Calls
An assembly code level example
Non-secure memory Secure memory (Non-secure callable)
NonSecureFunc: SecureFunc:
BL SecureFunc Call SG
<Non-secure code> Enter Secure state
<Secure code>
Return to NS BXNS lr

• Guard instruction (SG) polices entry point

• Placed at the start of function callable from non-secure code.
• Non-secure  secure branch faults if SG isn’t at target address
• Can’t branch into the middle of functions
• Can’t call internal functions.
• Code on Non-secure side identical to existing code.
 Cortex-M23: Imagine the possibilities

32-bit Small area,

performance low cost
TrustZone Long battery
for ARMv8-M life

27 © ARM 2017
Cortex-M23 enhancements over Cortex-M0+
Cortex-M23
TrustZone
Stack limit checking
Hardware divide
Exclusive memory accesses
Enhanced debug
Cortex-M0+ ETM
NVIC (max 32 IRQs) NVIC (max 240 IRQs)

2.50

0.98
MPU (PMSAv6) MPU (PMSAv8)

2.46

0.95
AHB Lite AHB5
WIC WIC
Fast I/O bus Fast I/O bus
MTB MTB
Serial wire / JTAG Serial wire / JTAG
ARMv6-M ARMv8-M baseline CoreMark®/MHz DMIPS/MHz
New or updated
 Addressing diverse embedded and IoT opportunities

Automotive Industrial Smart Agriculture Wearables Smart Home

building Healthcare
lighting automation

Logistics
Retail Environmental

Cortex-M33
32-bit Digital signal
performance TrustZone Configurability, processing
for ARMv8-M extensibility

29 © ARM 2017
Cortex-M33 enhancements over Cortex-M4
Cortex-M33
TrustZone
Stack limit checking
Co-processor interface
Enhanced debug
+13% +20%
Cortex-M4 MTB
ETM ETM
NVIC (max 240 IRQs) NVIC (max 480 IRQs)
MPU (PMSAv7) MPU (PMSAv8)
AHB Lite AHB5

1.50
3.86
3.40

1.25
FPU FPU
SIMD/ DSP SIMD/ DSP
WIC WIC
Serial wire / JTAG Serial wire / JTAG
ARMv7-M ARMv8-M mainline

New or updated CoreMark®/MHz DMIPS/MHz

IoT systems and networks
• OSI model for networks.
• Internet protocol.
• IoT networking concepts.
• Example networks:
• Classic Bluetooth, Bluetooth Low Energy.
• 802.15.4 and Zigbee.
• Wi-Fi.

Computers as Components 4e © 2016 Marilyn Wolf

 Network Abstractions: OSI model
• International Standards
Organization (ISO) application end-use interface
Open Systems presentation data format
Interconnection (OSI)
to describe networks: session application dialog control

• 7-layer model. transport connections

• Standard way to network end-to-end service

classify network reliable data transport
data link
components and
operations. physical mechanical, electrical

Computers as Components 4e © 2016 Marilyn Wolf

OSI layers
• Physical: connectors, bit formats, etc.
• Data link: error detection and control across a single link (single hop).
• Network: end-to-end multi-hop data communication.
• Transport: provides connections; may optimize network resources.
• Session: services for end-user applications: data grouping,
checkpointing, etc.
• Presentation: data formats, transformation services.
• Application: interface between network and end-user programs

Computers as Components 4e © 2016 Marilyn Wolf

PHY and MAC
• PHY = physical layer. • MAC = media access control.
• Circuitry to transmit and receive • Provides link-level services.
bits.

Computers as Components 4e © 2016 Marilyn Wolf

Internet Protocol (IP)
Internet = network of networks: transports data from one network to another.

application application

presentation presentation

session Internet Protocol session

transport transport

network network network

data link data link data link

physical physical physical

Node A Router Node B

Network links
Computers as Components 4e © 2016 Marilyn Wolf
IoT networking concepts
• Edge device may not run IP
protocol.
• IP connection may be provided by message
hub or gateway.
• Non-IP networks are known as edge
networks.
Internet
• Ad hoc network is self-organized--- gateway
not set up by system administrator.
• Ad hoc network services:
• Authentication of eligibility to join
network. device device
• Authorization for access to given
pieces of information on the network.
• Encryption and decryption. device

Computers as Components 4e © 2016 Marilyn Wolf

Network topologies

star tree mesh

Computers as Components 4e © 2016 Marilyn Wolf

Routing
Routing Routing Routing
table table table
• Routing discovery determines
routes between
source/destination pairs.
msg
• Routing is driven by routing
tables at the nodes.
Routing Routing
table table

Routing
table

Computers as Components 4e © 2016 Marilyn Wolf

QoS
• Many networks support synchronous and asynchronous
communication.
• Asynchronous: data records, etc.
• Synchronous: voice, etc.
• Quality-of-service (QoS): bandwidth and periodicity characteristics.
• Admission control ensures that network can handle the QoS demands
of a request.

Computers as Components 4e © 2016 Marilyn Wolf

Synchronization and beacons
• Many network operations require nodes to be synchronized.
• Synchronization can be performed using beacon.
• Beacon transmission marks the beginning of a communications interval.

beacon synchronous asynchronous …

Computers as Components 4e © 2016 Marilyn Wolf

Communcations energy
• Communications energy is a
large part of node energy
consumption. sleep
• Comm energy consumption
depends on many factors and
parameters. idle
• Generally evaluated for a set of
use cases.
• We can use power state machine receive transmit
to model communications
energy cost.

Computers as Components 4e © 2016 Marilyn Wolf

Communications power state machine
example
step state time energy
sleep
1 sleep 1 ms 1 nJ
1 µW
2 idle 10 µs 0.5 nJ
3 receive 50 µs 1.25 nJ
idle 4 transmit 50 µs 1.75 nJ
50 µW
5 receive 50 µs 1.25 nJ
6 transmit 50 µs 1.75 nJ
receive transmit total = 7.5 nJ
25 mW 35 mW

Computers as Components 4e © 2016 Marilyn Wolf

Bluetooth
• Introduced in 1999, originally for telephony applications.
• Classic Bluetooth operates in instrumentation, scientific, and medical
(ISM) band in the 2.4 GHz range.
• Bluetooth networks organized as piconet.
• One master, several slaves.
• Slave can be active or parked.
• A device can be a slave on several networks simultaneously.

Computers as Components 4e © 2016 Marilyn Wolf

Bluetooth stack
• Transport protocol:
• Radio, baseband layer, link manager, logical link control and adaptation
protocol (L2CAP).
• Middleware:
• RFCOMM for serial port, service discovery protocol, Internet Protocol, IrDA,
etc.
• Applications.

Computers as Components 4e © 2016 Marilyn Wolf

Bluetooth protocol
• Every Bluetooth device has a 48-bit Bluetooth Device Address.
• Every device has a Bluetooth clock.
• Transmissions alternate between master and slave directions.
• Two types of packets:
• Synchronous connection-oriented (SCO) packets for QoS-oriented traffic.
• Asynchronous connectionless (ACL) packets for non-QoS traffic.
• SCO traffic has higher priority than ACL packets.

Computers as Components 4e © 2016 Marilyn Wolf

Bluetooth Low Energy
• Designed for very low energy operation such as button-sized battery.
• Goal: minimize radio on-time.
• Part of Bluetooth standard but deviates from Classic Bluetooth in several ways.
• Advertising transmissions can be used to broadcast, discover devices, etc.
• Connections can be established.
• Attribute Protocol Layer allows devices to create application-specific protocols.
• Generic Attribute Profile Layer (GATT) defines basic attributes for all BLUE
devices.
• Pairing devices uses a short-term key to send a long-term key.
• Bonding: storing long-term key in device database.
• Optional data encryption using AES.

Computers as Components 4e © 2016 Marilyn Wolf

 scanning
Bluetooth Low Energy (BLE)
active
Link-level state machine
passive

advertising standby initiating

connected

master

slave

Computers as Components 4e © 2016 Marilyn Wolf

802.15.4 and ZigBee
• 802.15.4 defines MAC and PHY layers.
• Supports full-function and reduced-function devices.
• Either star or peer-to-peer topology.
• Communication performed using frames.
• Optional superframe provides a beacon mechanism and QoS.
• ZigBee is a set of application-oriented standards.
• NWK layer provides network services.
• APL layer provides application-level services.
• Supports many different topologies.

Computers as Components 4e © 2016 Marilyn Wolf

Wi-Fi
• Originally designed for portable and mobile applications.
• Has been adapted for lower-energy operation.
• Supports ad hoc networking.
• Network provides a set of services:
• Distribution of messages from one node to another.
• Integration delivers messages from another network.
• Association relates a station to an access point.

Computers as Components 4e © 2016 Marilyn Wolf

IoT Systems Databases
• Database holds data about devices, helps to analyze data.
• Relational database management system:
• Domain1 X domain2 X … -> Range.
• Database organized into records or tuples:
• Attribute: table column.
• Record: table row.
• One column is the primary key---uniquely identifies a record.

Computers as Components 4e © 2016 Marilyn Wolf

Database example devices
name id (primary address type
key)
door 234 10.113 binary
record refrigerator 4326 10.117 signal
table 213 11.039 MV
chair 4325 09.423 binary
faucet 2 11.324 signal

device_data
signature device time value
(primary key)
256423 234 11:23:14 1
252456 4326 11:23:47 40
663443 234 11:27:55 0
Computers as Components 4e © 2016 Marilyn Wolf
IoT Management - Timewheels
• Used to manage timing of events in the system.
• Timewheel is a time-sorted set of events.
• Event placed in proper spot in timewheel queue upon arrival.
• When current time is equal to time of event at head, event is processed.

8:00 AM 7:45 AM 7:30 AM

… lights on
toaster on radio on

head pointer

7:28 AM

Computers as Components 4e © 2016 Marilyn Wolf

Timewheel state diagram

time = current_event.time()
get next event process event

Computers as Components 4e © 2016 Marilyn Wolf

Example: smart home
• Performs a variety of services:
• Remote or automatic operation of lights and appliances.
• Energy and water management.
• Activity monitoring.
• Activity monitoring can help elderly, people with special needs:
• Reports on daily activities.
• Alerts for out-of-the-ordinary activity.
• Recommendations.

Computers as Components 4e © 2016 Marilyn Wolf

Example smart home
camera
• Cameras can identify resident bedroom 1
and their activity.
• Faucet, door sensors can identify dining/living
activity but not who performs room
door sensor
the activity.
camera

faucet faucet sensor

sensor bath
kitchen

Computers as Components 4e © 2016 Marilyn Wolf

Use case: activity monitoring
resident door monitor hall camera bathroom tap TV

Computers as Components 4e © 2016 Marilyn Wolf

Use case: light control
resident console door monitor light

Computers as Components 4e © 2016 Marilyn Wolf

Smart home object diagram
sensor console

hub timewheel database

1
actuator
event queue

1
n

event

Computers as Components 4e © 2016 Marilyn Wolf