Ultra-Wide Band Communication for the Internet of Things

commonkissingcanΔίκτυα και Επικοινωνίες

17 Φεβ 2014 (πριν από 3 χρόνια και 5 μήνες)

178 εμφανίσεις

1

Ultra
-
Wide Band Communication
for the Internet of Things

The MICS UWB Network

uwb.epfl.ch


Jean
-
Yves Le Boudec (coordinator), EPFL I&C

21
-
23 January 2008

2

Abstract
:


Ultra
-
Wide

Band

communication

is

a

technology

for

low

range,

low

power

sensor

and

mobile

devices

which

employs

very

low

transmission

powers

(below

the

level

of

unintentional

emissions)

and

high

bandwidth
.

It

possesses

a

number

of

unique

features

that

make

it

very

attractive

to

many

local

applications
.

First,

ranging

with

high

accuracy

is

possible

even

indoors
.

Second,

it

is

resistant

to

multipath

fading

which

often

pleagues

indoors

communications
.

Third,

it

scales

well

in

dense

deployments
.

Fourth,

cryptographic

modulation

is

possible
.

In

this

talk,

we

describe

the

research

done

in

the

MICS

Ultra
-
Wide

Band

network,

showing

ranging,

dense

deployment

capabilities

and

medical

applications
.


3

Table of Contents

1.
The UWB Network of MICS

2.
What is UWB ?

3.
Impulse Radio UWB

4.
Low Power Medical Application

5.
Robustness to Interference

6.
Ranging

7.
Outlook

4

A Network within MICS researching on
Impulse Radio UWB

The network

CSEM, Neuchatel

Prof. Farserotu, Hai Zhan

Prof. Decotignie, Jerôme
Rousselot

ETHZ, Zurich

Prof. Wittneben, Florian Trösch,
Christoph Steiner

EPFL I&C, Lausanne

Prof Le Boudec (coordinator),
Ruben Merz, Manuel Flury

EPFL STI, Lausanne

Prof. Dehollain, James Colli
-
Vignarelli, Prakash
Thoppayegambaram

Prof. Skrivervik, Gabriela Quintero

HES SO, Yverdon

Prof. Robert, Jérome Vernez

ST Microelectronics, Geneva

Dr. J. Zory



Impulse radio

Ultra Wide Band
communication

Low power

In presence of multi user
interference

Ranging


Provide fundamental research
and proofs of concept


5

Table of Contents

1.
The UWB Network of MICS

2.
What is UWB ?

3.
Impulse Radio UWB

4.
Low Power Medical Application

5.
Robustness to Interference

6.
Ranging

7.
Outlook

6

Ultra Wide Band (UWB) Communication

Use a very large spectrum

up to Several GHzs

Very low power

Below level of unintentional
emission

Unlicensed

Co
-
exists with other technologies

Power Limits

FCC (2002) limits

peak power (0dBm per 50MHz)

mean power (
-
41.3dBm per MHz)

Europe (and CH
-
Ofcom, 2007) put
more stringent limits

US

EC

(source: FCC 2002, CH
-
Ofcom, 2007)

7

Various Uses of UWB Signals

Radar and Ranging

Radar

A very old UWB application, used
for maritime or air navigation, and
as remote speedometer

New apps: automotive security,
rescue operation

One active device analyzes echo

Target is passive and unaware of
signal

Not always low power



Ranging

From device to device

Device is active sender

Base station is receiver
/transmitter

E.g Ubisense, Cambridge UK

Low power

E.L.

E.L.

8

Various Uses of UWB Signals

Communication

Short Range Communication

Low power

Up to 30 m indoors


High data rate UWB
Communication

Wireless USB / Wireless Firewire

Uses entire bandwidth

Very large bit rate on one single
link

Peaky in frequency


Low data rate

E.g. Sensor networks

Impulse radio signals

Very large aggregate throughput

Robots with ranging needs
for collective intelligence

Source: Prof. Alcherio
Martinoli

9

Strengths and Weaknesses of UWB

High throughput

for high data rate

Shannon
-
Hartley law:

C = B log2 ( 1 + S/N )


with

C = bit rate (b/s)



B = bandwidth (Hz)

Exploited by Wireless USB /
Firewire : 100
-

480 Mb/s for
Wireless USB over 3
-
10 m


Low Power

for Low Data rate


Scalability

Sensor network with very large
bandwidth, total capacity scales
with number of nodes

Resistance to
Channel
Impairments

Multiple paths are distinguishable

Suitable for indoors, terrain with
obstacles, metallic environment

High Resolution in time domain

Ranging

with cm accuracy indoors

Secure ranging

Short range

10 m to 30 m

Source: Mohammad Abualreesh

10

Table of Contents

1.
The UWB Network of MICS

2.
What is UWB ?

3.
Impulse Radio UWB

4.
Low Power Medical Application

5.
Robustness to Interference

6.
Ranging

7.
Outlook

11

Impulse Radio UWB Uses Short Pulses

Pulses are narrow in time,
wide in frequency


Pulse duration order of 1 ns



Source: Gabriela Quintero

Features

Low power

Duty cycle at 1 Mb/s = 1 %

Robust against multi
-
user interference

High precision ranging



12

Impulse Radio UWB Uses Time
Hopping

Time Hopping Sequence: […, 2, 5, 4, 7 …]

Pulses appear random unless you know THS

THS is predictible to user who knows the key ; e.g.: MAC address


Transforms packet collision into symbol collision

Increaed bit error rate instead of packet loss


Software
-
like flexibility in hardware

When a pulse is sent can easily be changed by modifying a few values
in the system

Change the time hopping sequence

Change the modulation rate


13

Multipath Propagation

Signal propagation subject to reflections

Pulses are attenuated / modified but still distinguishable

Very little destructive interference

Channel response

Received signal

14

Table of Contents

1.
The UWB Network of MICS

2.
What is UWB ?

3.
Impulse Radio UWB

4.
Low Power: Medical Application

5.
Robustness to Interference

6.
Ranging

7.
Outlook

15

Body Area Network with UWB

Requires very low power

Very bad transmission
channel


UWB body area network
prototype developed at ETH /
Prof A. Wittneben’s group

Ear to ear communication

Focus on low power and
point to point link



16

Wireless BAN Communication for less
than 1 mW

Bursts of 500 bits/ms

Average Data Rate of 500 kbits/s

Peak Data Rate of 50 Mbits/s

Simple Tx and Rx Structures

Mainly Analog Processing

Estimated Power Consumption < 1mW

Analog Part

Rx Chain
Energy Detection

Tx Chain
UWB Pulse
Generator

1% duty cycle
500 kbits/s

< 0.3 mW

Digital
Baseband

ADC
Clock Synthesis
Synchronization
Decoding
Error Correction
MAC

< 0.7 mW

Sampling at

200 MHz

Low Cost

Low Power

Low Complexity

Ultra
-
Wideband Radio

17


Body Area Network UWB
Test Bed

Ear
-
to
-
Ear Channel

18

GUI for
UWB test
-
bed

Average
transmit power
-
45 dBm

Ear
-
to
-
ear
channel with
artificial
water
-
bucket
-
head

BER at
-
45dBm
is 0.04,
capacity is 480
Mb/s

transmit

receive

19

Relevant Publications

F. Troesch, C. Steiner, T. Zasowski, T. Burger, and A. Wittneben, "Hardware
Aware Optimization of an Ultra Low Power UWB Communication System,"
IEEE
International Conference on Ultra
-
Wideband, ICUWB 2007
, Marina Mandarin,
Singapore, Sept. 2007.

C. Steiner and A. Wittneben, "On the Interference Robustness of Ultra
-
Wideband Energy Detection Receivers,"
IEEE International Conference on
Ultra
-
Wideband, ICUWB 2007
, Singapore, Sept. 2007.

20

Table of Contents

1.
The UWB Network of MICS

2.
What is UWB ?

3.
Impulse Radio UWB

4.
Low Power Medical Application

5.
Robustness to Interference

6.
Ranging

7.
Outlook

21

Robustness to Interference

From Theory to Practice

In Theory, UWB transmission
is robust to interference from
other UWB systems

Due to large bandwidth


This makes UWB systems
potentially
scalable,
well
adapted to dense deployments

Throughput per node constant
with number of nodes N

Contrast to narrowband
systems:


»

N
-
1/2

In practice, this requires
careful system design

MAC

Signal Acquisition

Accommodate multipath



22

PHY
-
Aware MAC

Classical organization of a
network

E.g. WiFi, Bluetooth

PHY transmits packets

MAC avoids collisions

i.e. MAC = mutual exclusion


This is not efficient for UWB

Mutual exclusion divides
throughput linearly…

… but most collisions are at
pulse level

Rate reduction is small

The optimal is: Allow
interference and manage it !

Requires MAC to be PHY
aware

Data

THS(A), Code =
R
i

ACK

THS(A),Code =
R
N

Data

THS(A),Code =
R
j

Idle

THS(B), Code =
R
N

NACK

THS(A),Code =
R
N

Incremental Red.

THS(A)

B

A

C

Our experimental MAC

Interference,
not collision

23

DCC
-
MAC

A PHY aware MAC protocol,
designed to be robust to
interference

DCC= dynamic channel coding

Key features of design

One time hopping sequence
per destination (private time
hopping sequences)

Interference mitigation at
pulse level

Mutual exclusion for a single
destination only

Rate adaptation



DCC MAC

CA/CDMA
-
like

802.11
-

like

N nodes in a chain

24

Signal Acquisition

Signal acquisition is difficult for
Impulse Radio UWB

Signal is intermittent

Interferences are allowed

Classical methods based on
gaussian noise hypotheses do not
apply

Power Independent Detection
(PID) is robust to interference


even if interfering power is larger
than intended signal

uses thresholding


25

Private Time Hopping Sequences

Common Time Hopping Sequence
in preamble

Many useless acquisitions

One Private Time Hopping
Sequence per destination

Acquisition is private, only
intended receiver decodes

Requires source to know
sequence of destination

E.g. linear congruence seeded
with MAC address of destination

26

Private Sequences Avoid the Ad
-
Hoc
Collapse

Ad
-
hoc collapse

Many TCP connections in an
ad
-
hoc

Collapses with 802.11 and
other protocols

Due to collisions

No good solution known to this
problem


With private sequences, the
ad
-
hoc collapse goes away

Nodes acquire only packets
destined to self

27

Accommodate Multipath

Assume modulation is pulse
position

With interferers and multipath,
received signal looks like


28

Idea: (Rake receiver)

Estimate channel during signal
acquisition phase

Look for pattern of pulses in the
received signal
-

correlation

Use thresholds to avoid near end
effects


Similar ideas apply to energy
detectors

0

29

Performance Evaluation of IEEE 802.15.4a

Standard for Impulse Radio
UWB, Low Data Rate

MAC influenced by narrow
band tradition

2 THSs in total

Makes some compromises to
ease implementation

Bursts of pulses

Q: how does it perform with
respect to interference
robustness ?

Multiple transmissions in same
network

Transmissions from
neighbouring, non
coordinated network


We simulated the standard in
detail, with interferers, and
compared its performance
against two benchmarks

Benchmark 1: Destructive
collision

Packet lost when two
transmissions overlap

ALOHA performance

Typical of narrowband
systems

Benchmark 2: Perfect capture

Packets compete during
signal acquisition and
transmission

Only one succeeds

Typical of ideal UWB system

30

IEEE 802.15.4a is not Robust to Interference

Performance is close to
destructive collision

Does not exploit UWB benefits
well

Possible fixes

Compress bursts

Private time hopping
sequences

Benchmark 2: Perfect capture

Benchmark 1: Destructive collision

802.154a, no interference

802.154a, with interference

31

Interference Testbed

Goal:

Implement and test multi
-
user
impulse radio system

In presence of multi
-
user
interference


Real hardware, still
programmable in matlab

A coordinated effort of the
MICS UWB network

Ruben Merz (coordinator)

James Colli
-
Vignarelli

Gabriela Quintero

Prakash Thoppayegambaram

Jerome Vernez

Jean
-
François Zürcher

32

Interference Testbed (EPFL, HES SO)

Video by Jerome Vernez, HES SO (Yverdon)
33

Relevant Publications

El Fawal, Alaeddine ; Le Boudec, Jean
-
Yves, “A Robust Signal Detection
Method for Ultra Wide Band (UWB) Networks with Uncontrolled
Interference”, In:
IEEE Transactions on Microwave Theory and Techniques
(MTT),

vol. 54, num. 4, part 2, 2006, p. 1769
-
1781

Radunovic, Bozidar ; Le Boudec, Jean
-
Yves, “Optimal Power Control,
Scheduling and Routing in UWB Networks”, In:

IEEE Journal on Selected
Areas in Communications
, vol. 22, num. 7, 2004, p. 1252

Merz, Ruben ; Widmer, Jörg ; Le Boudec, Jean
-
Yves ; Radunovic, Bozidar,
“A Joint PHY/MAC Architecture for Low
-
Radiated Power TH
-
UWB Wireless
Ad
-
Hoc Networks”, In:
Wireless Communications and Mobile Computing
Journal, Special Issue on Ultrawideband (UWB) Communications,

vol. 5,
num. 5, 2005, p. 567
-
580

Flury, Manuel ; Merz, Ruben ; Le Boudec, Jean
-
Yves, “Managing Impulsive
Interference in Impulse Radio UWB Networks”, In:
ST Journal of Research,
2007

Flury, Manuel ; Merz, Ruben ; Le Boudec, Jean
-
Yves ; Zory, Julien,
“Performance Evaluation of an IEEE 802.15.4a Physical Layer with Energy
Detection and Multi
-
User Interference”, In:
IEEE International Conference
on Ultra
-
Wideband (ICUWB 2007)
, 2007


34

Table of Contents

1.
The UWB Network of MICS

2.
What is UWB ?

3.
Impulse Radio UWB

4.
Low Power Medical Application

5.
Robustness to Interference

6.
Ranging

7.
Outlook

35

Impulse Radio UWB enables low cost
ranging at high precision

Short pulses can easily be located by receiver

Basis for radars

Can be used at low cost in all sorts of equipments with UWB

2 techniques are researched in the MICS UWB Network

Geo
-
regioning

High resolution ranging

36

Geo
-
Regioning

A method for location finger
-
printing

Idea: channel impulse
response is correlated in
space

Method:

Learning phase:

send test signals to base
station from various locations

Analyze correlations (e.g.
covariance matrix, delay
profile)

Tracking Phase

Mobile sends beacons to base
station

Real time correlation is
performed




Channel response

37

UWB Geo
-
Regioning Demonstration

Developed by Prof. A. Wittneben’s
group / ETHZ

Channel impulse responses from region 22 to RX

38

Relevant Publications

C. Steiner, F. Althaus, F. Troesch, and A. Wittneben, "Ultra
-
Wideband Geo
-
Regioning: A Novel Clustering and Localization Technique,"
EURASIP Journal
on Advances in Signal Processing, Special Issue on Signal Processing for
Location Estimation and Tracking in Wireless Environments
, Nov. 2007.

C. Steiner and A. Wittneben, "Clustering of Wireless Sensors based on Ultra
-
Wideband Geo
-
Regioning,"
Asilomar Conference on Signals, Systems, and
Computers
, Pacific Grove, USA, Nov. 2007.

39

High Resolution Ranging

Accurate ranging = estimation
of distance

Based on time of arrival of
signal

Idea:

mobile sends UWB pulses to
one or several base stations

detect
first

pulse at receiver


How:

Estimate
both

channel
response and time of arrival of
first pulse

Not always strongest

Remove noise and
interference by modified
Prony algo

40

Experimental setting











Quiet room at EPFL (not
anechoic
)

Experiment implemented by
Hai Zhan (CSME)

True distance is 48.8 cm


estimated distance is 50.0 cm



Ranging Through Obstacles and With
Interferers

Non severe non light of sight
ranging is possible

E.g. through wood or
cardboard


The modified Prony algorithm
finds the first pulse

Sent signal contains a train of
encoded pulses

Received signal contains
many replicas due to multipath

Strong pulses help find weak

Click on figure for video

Video by Hai Zhan, CSEM

41

Relevant Publications

Zhan, Hai ; Farserotu, John ; Le Boudec, Jean
-
Yves “A Novel
Maximum Likelihood Estimation Of Superimposed Exponential
Signals In Noise And Ultra
-
Wideband”,

PIMRC 07, 2007


Zhan, Hai ; Ayadi, Jaouhar ; Farserotu, John ; Le Boudec, Jean
-
Yves, “High
-
Resolution Impulse Radio Ultra Wideband”,

In: The
2007 IEEE International Conference on Ultra
-
Wideband, ICUWB
2007, 2007




42

Table of Contents

1.
The UWB Network of MICS

2.
What is UWB ?

3.
Impulse Radio UWB

4.
Low Power Medical Application

5.
Ranging

6.
Robustness to Interference

7.
Outlook

43

Impulse Radio UWB is a key technology for
the Internet of Things

Unique features

Indoors ranging

Resistance to multiuser
interference

Scalable total throughput

Very low power


Practical developments are
only starting

Standard based
implementations can be
improved


Potential areas of future
research

Secure ranging

Very short signal time

High throughput ranging

Frequent position updates for
distributed robot control

44

Thank You

Special thanks go to all who
helped prepare this presentation

Jerome Vernez

Hai Zhan

Ruben Merz

Christoph Steiner

And to all other contributors of the
MICS UWB network who make this
project such a great fun

Manuel Flury

James Colli
-
Vignarelli

Jean
-
Dominique Decotignie

Catherine Dehollain

John Farserotu

Gabriela Quintero

Stephan Robert

Jérome Rousselot

Anja Skrivervik

Prakash Thoppayegambaram

Florian Trösch

Armin Wittneben

Julien Zory