IoT Needs Good Neighbours - Cognitive Radio Turns Enemies into Friends

www amihotechnology.com © AMIHO Technology 2015 -- CONFIDENTIALwww amihotechnology.com © AMIHO Technology 2016
IoT Needs Good Neighbours
Cognitive Radio Turns Enemies into Friends
Steve Clarke, Technical Director
Presentation for Embedded World
25 February 2016

Founded in 2009 in Cambridge,
AMHIO Technology designs and
develops intelligent wireless
communications for smart meters and
IoT applications and networks,
including S/W stacks, hardware IP and
embedded modules.
Formerly a senior engineer at the BBC
and lecturer at Salford University,
Steve Clarke has designed systems
from bus telematics to multi-award
winning AV processors. He is
Technical Director and co-founder of
AMIHO Technology.
About the Author and Amiho

An explosion of connected devices and applications is
occurring in machine communications that makes up the
Internet of Things
Industrial, Scientific and Medical (ISM) radio spectrum is
used for the majority of these communications
Users are searching for reliable communications, which
is encouraging selfish behaviour
Sub 1GHz ISM bands are under threat
Cognitive radio makes better use of shared spectrum
To cope with the increase of users, we need these
radio devices to be Good Neighbours
The Everything-Connected World

The Future is Rosy Noisy
The New World of
Connected Devices
50bndevices
That’s more than
100k! devices / km2
Inner London
population density
> 10,000/ km2
600M
smart
meters
Updates as
frequent as
every 15
seconds
2MHzofspectrumin
the868MHzISMband
Keyless entry
systems
Remote
sensors
Wearabletech
Alarms
Cable
replacement
Wireless
Thermostats
Doorbells
Barrier control
Traffic light
priority
Baby monitors
RFID

ISM – Industrial, Scientific and Medical
Used by Short Range Devices, divided into 2 bands:
Sub- GHz used for longer-range and older applications,
characterised by:
Shared spectrum, with many unrelated users / uses
Often just pairs of devices
Often unidirectional
Often fairly dumb devices
Transmit any time (Aloha) and accept collisions
Lack of standards
> 1GHz – occupied by WiFi, Bluetooth and others
Typically shorter range, higher data rate
Unlicensed Spectrum – ISM Bands

Simple ISM band systems expect lightly used channels
so as to allow data to be received reliably
Defence against noise is usually through the use of
multiple transmissions of the same message
Classical communications theory shows that transmit-
at-will (Aloha) protocol gives low channel capacity
Increasing the number of users exponentially reduces
reliability of communications
Fixed frequency systems are susceptible to jamming:
Frequency hopping and spread-spectrum are 2
techniques to reduce the effect of this
Deficiencies of Simple Systems
Example - a simple ISM
transmitter may send a
message with probability of
correct reception of 0.9.
If we send the message 3
times, we get overall failure
probability of 0.13
i.e. 1 in a 1000
If the noise level doubles, the
probability of each message
failing is 0.2, giving an overall
failure rate of 0.23
i.e. 1 in 125

Large-scale networks have appeared in the sub-1GHz ISM bands
– they represent a major nuisance for existing users. These
include both public and private networks:
Smart Metering – private utility networks, predominantly
using the Wireless Meter-Bus protocol
Sigfox – subscription-based public networks, using ultra
narrow-band transmissions
LoRa – both public and private networks, using spread-
spectrum transmissions
All of these are set to use large amounts of ISM spectrum
The New, High Volume ISM Users

Both designed for low data-rate applications, typical 3-30
transmissions / day, to cover medium distances (several km) in a
single hop.
Both use aggressive techniques to maintain the network.
Sigfox uses ultra narrow-band transmissions to increase the
sensitivity of the receiver, with frequency agility to prevent
transmissions being continually blocked.
Channels are a few-hundred Hz wide, ultimately allowing more
energy to be put into the ISM band by a large number of devices
with narrow channels
Sigfox is only available for subscription networks
Sigfox and LoRa Basics (1)

LoRa uses CSSS (Chirp Sequence Spread Spectrum) modulation
to spread the data through a wide channel.
Bits are represented with codes that are sequences from 64 to
4096 symbols.
This redundancy improves sensitivity by up to 20dB
The redundancy comes at price – bit rates range from 37.5 kb/s
to 18 bits / sec
Transmissions with different code sequences can occupy the
same band and still be discriminated from each other
LoRaWAN is a proprietary network protocol used by the
LoRa Alliance. It can be implemented both as a subscription and a
private network.
Sigfox and LoRa Basics (2)

600M meters in Europe are being replaced.
Radio communications used for almost all gas and water
meters
The majority of radio communications uses Wireless Meter-
Bus, a sub 1GHz simple transmission protocol.
The system was designed for the frequent transmission of
short messages every 10-15 seconds
Most devices are unidirectional and communicate at full
power
Hard-to-reach ‘dark meters’ represent a major problem
that is being solved through the use of high power
transmissions and lower frequency (169 MHz)
communications
Smart Meter Networks

What do we mean by ‘Cognitive’?
‘Cognition is a black-box of what we colloquially call
memory, attention, language/communication and decision
making’
Dr KMJ Diederen, Neuroscientist, University of Cambridge
What is Cognitive Radio?

The term Cognitive Radio was first used in 1989 by J Mitola, to
describe a radio that senses its environment and adapts its
behaviour to best suit that environment.
In radio terms, this might include:
The frequency spectrum and division of channels within it
Intelligent sensing of channel occupancy and of the channels to
use
Power levels needed
Modulation or coding schemes
It was envisaged to allow devices to use then relatively lightly-used
spectrum without interfering with existing users.
True cognitive radio, by its nature, is cooperative.
What is Cognitive Radio (2)

Much of the effort goes into choosing the channel to be
used
Try to send messages in a way that is least likely to cause
collisions
Sensing can (and should) be done at both ends of the
communication link.
Devices build up a map of how spectrum is used, such as
a database of the most frequently used channels
Features such as power control are ‘self-interested’
techniques that save energy and coincidentally minimise
interference
How Clever Does it Need to be?

2.4 GHz standard, developed for short range (<10m)
small networks at data rates from 100kbps – 2Mbps
Adaptive power control
Adaptive channel – the next channel used depends on
clear-channel assessment at both ends
Error correction in packets is added adaptively
according to error rates
Data rate has back-off to achieve shortest possible
packets
Modulation scheme, and thus the bits/symbol, is
selected according to background noise
Phasing of packet slots in multi-device networks is
chosen to minimise collisions from timing drifts
Cognitive Examples - Bluetooth

Wireless Meter-Bus has added higher layers for some
systems – notably the Italian gas (CiG) specification
UN/TS11291, using 169 MHz
CiG specifies listen-before-talk and power control
LoRaWAN implements
adaptive data rate
adaptive channel hopping
power control
Its intelligence lies mainly in the gateway
LoRaWAN does not implement listen-before talk
Semi-Cognitive Examples

Some networks – such as metering – are infrastructure and
could (as in the UK) use dedicated frequencies
New ISM band applications will have to consider the large
amount of noise generated by other users
All users of shared spectrum should strive to minimise the
interference they generate
Just shouting louder will not work – networks need to be
cooperative
Cognitive Radio makes better, less selfish use of
spectrum and is better able to cope with interference
Making ISM Bands Work

www.amihotechnology.com | +44 1223 422345
Steve Clarke
Technical Director
sclarke@amiho.co.uk

IoT Needs Good Neighbours - Cognitive Radio Turns Enemies into Friends

More Related Content

What's hot (20)

Similar to IoT Needs Good Neighbours - Cognitive Radio Turns Enemies into Friends (20)

Recently uploaded (20)

IoT Needs Good Neighbours - Cognitive Radio Turns Enemies into Friends