Wireless Summit 2012, Oslo29 April 2012
Advanced Radio Schemes
- with potential in Industrial Wireless
Pål Orten,
ABB Corporate Research and UniK, University of Oslo
© ABB Group
April 24, 2012 | Slide 1
There is not ONE Performance of Wireless

All industrial WSN standards are based on IEEE802.15.4

WirelessHART, ISA 100, WIA-PA

Same radio  Same fundamental performance limitations

IEEE802.15.4 is quite simple compared to many other radio systems

Challenges of industrial wireless – calls for more advanced radio schemes

Outline:
© ABB Group
April 24, 2012 | Slide 2

Basic performance measures and what is possible

Principles of advanced schemes
When is it possible to communicate?
Shannon limit
8
Shannon Limit
Asymptote -1.6 dB
W/C
6

No limit on input

AWGN channel

To reach the limit:
4

2
0
-2
© ABB Group
April 24, 2012 | Slide 3
0
2
Eb/N
[dB]
E
/N
0
b
0
4
6

infinite bandwidth

infinite code word
length
Can be modified for
modulated systems, finite
bandwidth and code
word lengths
Bit Error Rate – PB
(or equivalently Packet Error Rate)

PB over a link decreases with
increasing signal-to-noise ratio
(”waterfall” - curve)

Will never be zero

Varies with various radio
parameters

Modulation method

Error control coding

Channel influences

Receiver quality,
implementation loss (normally
negligible)

Interference
Packets are typically retransmitted until correctly received
© ABB Group
April 24, 2012 | Slide 4
(depending on application)
What about latency?

Latency depends on

Number of hops from source to destination, h

Available resources for packet scheduling, r

Number of retransmissions, TR

Latency Δ = f1(TR, r, h, …)

TR = f2(PB)  Reduced error rate gives lower latency, better
spectrum utilization, room for more nodes
Wired/fieldbus protocol
GW
WSN protocol
Node
Node
Node
Multihop
Node
meshed
Node
Node
Node
© ABB Group
April 24, 2012 | Slide 5
Node
Node
Node
networks
Advanced Error Control Coding
10
10
10
-1
Capacity Limit
Uncoded
Sequential limit
Convolutional Coding
Concatenated Coding
DVB-RCS Turbo
-2
-3
BER
BER
10
10
10


some redundancy

Receiver complexity
Convolutional
-4

MPEG packet lengths
(188 bytes)

R=1/2 (code rate, =
number of info bits
relative to total number
of bits transmitted)
Turbo
-5
Concatenated
-6
RS+Convolutional
10
Adds
-7
0
2
4
6
8
E /N [dB]
10
12
14
Ebb/N
0 0
Already used in

© ABB Group
April 24, 2012 | Slide 6
Mobile communications, satellite communications, WiMax, …
Hybrid ARQ schemes – incremental redundancy

An error detecting code tells if there were errors in a packet

With errors a retransmission occurs

Retransmitting the same information is equivalent to
repetition coding – most simple error control possible

Incremental redundancy – transmitting additional
redundancy to combine with previous bits in receiver
d
Information
© ABB Group
April 24, 2012 | Slide 7
r1
r2
rn
Redundancy

First transmission, d (+ headers etc)

Retransmission 1: r1 (+ headers etc)

Retransmission 2: r2 (+ headers etc)

Retransmission n: rn (+ headers etc)

 rate R=d/(d+∑ ri) code instead of a repetition code

 #retransmitted bits can be less than d
Fading channels

Delayed reflections add at the receiver
(constructive and destructive)

Gain varies over time and channel
smears out transmitted symbols

Consectutive symbols overlap/interfere

Increases error rate

Limits transmission rate
Why are psalms so slow?
© ABB Group
April 24, 2012 | Slide 8
Channel Adaptation
Adapt to channel conditions

High modulation order & Low code rate


Low modulation order & Low code rate


 High spectral efficiency
 Robust in bad channels
Parameters to adapt

Transmit power

Modulation size

Code rate
6
Spectral Effeciency [bit/s/Hz]

5
4
Unconstrained
QPSK
16-QAM
64-QAM
3
2
1
0
-2
0
2
4
6
8
Capacity
Limit
E
/N
(dB)
E /N
b 0
b
10
0
Used in many modern
wireless systems
© ABB Group
April 24, 2012 | Slide 9
IEEE, Golldsmith&Chua 1998
Scheduling and Multiuser Diversity

Utilizing the fact that all users have some periods of good
channel conditions

Schedule traffic when the channel is good, over time everyone
gets to transmit under equal channel conditions

Packet is most likely lost in a fade

Requires channel prediction (adds power consumption,
signaling overhead, node to be more awake)

Fairness is an issue to be resolved
Many users fade
independently, high
probability that one or
more users have a good
channel at any time
 Multiuser diversity
© ABB Group
April 24, 2012 | Slide 10
Interference Cancellation
© ABB Group
April 24, 2012 | Slide 11

Example of successive interference cancellation (CDMA) with two
users

Procedure may be repeated in several stages (and generalised for
many users)

Typically starting with strongest users
Spectrum usage ....
© ABB Group
April 24, 2012 | Slide 12

Bandwidth is a scarce resource – expensive (ref auctions for
3G bands ~10 years ago)

Existing spectrum policies – similar to a fragmented disk

Licensed and unlicensed bands – is there something
inbetween?  Cognitive Radio
Cognitive Radio – non static spectrum allocation

dynamic spectrum allocation

wide-band spectrum sensing

real-time spectrum allocation and
acquisition

infrastructureless mesh networks

coexistence of systems

need sharing rules/etiquettes – ref.
cocktail party analogy

primary and secondary users

paradigm shift - from static to
dynamic
Need flexible radios talking many protocols – Software Defined Radio
© ABB Group
April 24, 2012 | Slide 13
Software Defined Radio (SDR) Principle
LNA
AD
AD

Analog components removed

© ABB Group
April 24, 2012 | Slide 14
DSP/
uC
Reconfigurable FPGA
Costly, non-linear, production inaccuarcies,
temperature sensitive, stability, ...

Conversion to digital domain at higher frequency

AD and processing challenging

For full potential – implement many standards
Asic/
FPGA/
DSP/
uC
Traditional
Radio Design
Ultimate
SW Defined
Radio
Conclusions

Some wireless systems perform better than others

Advanced radio schemes can will improve performance

Comes at more advanced processing

Critical industrial applications (closed loop control) may
justify the extra cost
Thanks for your attention!
© ABB Group
April 24, 2012 | Slide 15
© ABB Group
April 24, 2012 | Slide 16