Li-Fi modulation and networked

Li-Fi attocell concept

Tutorial

Professor Harald Haas

23 December, 2013
Contributions by
Svilen Dimitrov, Sinan Sinanovic
Thilo Fath, Nikola Serafimovski,
Irina Stefan, Abdelhamid Younis,
Dobroslav Tsonev, Mostafa Afgani,
Stefan Videv, Cheng Chen,
Wasiu Popoola, Yichen Li,
Enrique Poves, John Fakidis,
Harald Burchardt,

2 23 December,
2013
 Wireless data is growing
exponentially
YouTube
In 2011, ~ 140 views for every person on earth (over
1 Trillion views
More video is uploaded to YouTube in one month
than the 3 major US networks created in 60 years
72 hours of video are uploaded to YouTube every
minute
25% of global YouTube views come from mobile
devices
Source - http://www.youtube.com/t/press_statistics

Internet video traffic is growing at 48% CAGR

Source - Cisco Visual Networking Index: Forecast and Methodology, 2010-2015

3 23 December,
2013
 leading to RF Spectrum Shortage
Global Mobile Data Traffic

4 23 December,
2013
How fix the problem?

1. Identify new spectrum, and/or

2. Enhance spectrum reuse
(smaller cells)

5 23 December,
2013
The electromagnetic spectrum

6 23 December,
2013 .
 Li-Fi

0111000110101000010
11000110101000010
Optical Attocell Concept

8 23 December,
2013
Link-Level Communication System

9 23 December,
2013
Communication Scenarios

12 23 December,
2013
State-of-the-Art: Wi-Fi

Wi-Fi Router Location

13 23 December,
2013
State-of-the-Art: Wi-Fi

Wi-Fi Router Location

14 23 December,
2013
15 23 December,
2013
Optical Attocell Network

16 23 December,
2013
Data rates per device
Space per desk 4 m2
100 employees per
floor
Assume a uniform
distribution of
employees
20m x 20m floor = 400
m2
Area can be covered
by a single Wi-Fi AP
Each Li-Fi AP can
cover around 4 m2
Wi-Fi data rate 600
Mbps
Li-Fi data rate 20 Mbps

17 23 December,
2013
Interference Scenario

desired signal
user, k

I. Stefan, et al., VTC 2013-Spring, 2013

20 23 December,
2013
Optimisation Framework

21 23 December,
2013
Rx FOV 85 - without lighting
constraint

22 23 December,
2013
Rx FOV 85 - with lighting
constraint

23 23 December,
2013
Rx FOV 45 - without lighting
constraint

24 23 December,
2013
Rx FOV 45 - with lighting
constraint

25 23 December,
2013
26 23 December,
2013
 Digital Modulation

27 23 December,
2013
Key differences to RF

System Information Signal

carried on complex
RF bipolar
electric field valued

Incoherent carried on real unipolar

Optical optical intensity valued non-negative

28 December 23,
2013
Non-linear Characteristic

29 23 December,
2013
Optical channel

30 23 December,
2013
Coherence Bandwidth of Channel

31 23 December,
2013
Path loss

bi-directional reflectance
distribution function
(BRDF)

32 23 December,
2013
 Path loss, contd
Optical path gain

Electrical path gain

Path loss distance

Detector area

Irradiance
33 23 December,
2013
Example: Aircraft cabin / LOS

34 23 December,
2013
Example: Aircraft cabin / NLOS

35 23 December,
2013
Modulation Techniques

36 23 December,
2013
Transceiver building blocks

37 23 December,
2013
 Pulsed Modulation
On-OFF Keying
Intensity
Finite slope limits
achievable data
On rates

Thres.

Off
Time
1 0 1 0 0 0 1 0 1

38 December 23,
2013
Single Carrier Binary

39 23 December,
2013
Single Carrier Multi-level

40 23 December,
2013
PPM: BER Performance

Spectal Efficiency:

41 23 December,
2013
PAM: BER performance

Spectal Efficiency:

42 23 December,
2013
OFDM-based OWC System

43 December 23,
2013
OFDM

44 23 December,
2013
 DCO-OFDM and ACO-OFDM Symbol
Structures

DCO-OFDM

1
2 1 2 2 1

ACO-OFDM

1 1
2 1 2 2

45 23 December,
2013
ACO-OFDM Time Domain Signal

46 23 December,
2013
 QAM vs. CAP
QAM

CAP
Not feasible if C is much
greater than the symbol
frequency

A. H. Abdolhamid, et al. A Comparison of QAM/CAP Architectures, 1998

47 23 December,
2013
OFDM Generation (Time Domain)
After the IFFT, the signal follows a zero-mean Gaussian
distribution in the time domain:

Probability density
Time-domain signal
function

Scatter plot
6

Quadrature
0

-2

-4

-6
-6 -4 -2 0 2 4 6
In-Phase

48 December 23,
2013
Multi-carrier Multi-level

49 23 December,
2013
DC Bias and Signal Power

DC

50 23 December,
2013
UPVLC Results (assuming single
colour LED)

51 23 December,
2013
Bussgang Theorem
X is a zero-mean Gaussian random variable with variance
and g(X) is an arbitrary transform on X, which could be
linear or nonlinear.

The Bussgang theorem:

g X KX Yn
EXYn 0

Then:
EX g X
K
2

E Yn2 E g 2 X K 2 2
Ebnew K 2 Ebold EYn Eg X

Nonew
N o VarYn
VarYn E Yn2 EYn
2

52 December 23,
2013
Redefintion of the Distortion
Any arbitrary distortion function g(x) can be represented with
a set of intervals I and a number of continuous polynomials
which describe the function in those intervals.

Then g(X) becomes:

I
g x ck , j x j Ux xmin ,k Ux xmax ,k
nk

k 1 j 0

where n k is the order of the polynomial in interval k, and

U(x) is the
unit step function:
0, x 0
Ux
1, x 0
53 December 23,
2013
Examples
3-bit DAC:

g x 3U x 1.5 U x
1U x U x 1.5
1U x 1.5 U x
3U x U x 1.5

Clipping and LED

current-to-light
conversion:

g x 3U x 3 U x

1x 2 7 x 5 U x 1 U x 3
1U x U x 1

54 December 23,
2013
Closed Form Solutions
The equations from the Bussgang analysis become:

1
I nk
d j 1Dt , xmin,k , xmax,k ,0,
K
2 c
k 1 j 0
k, j
dt j 1 t 0

I nk
d j Dt , xmin,k , xmax,k ,0,
Eg X ck , j
k 1 j 0 dt j t 0

d j m Dt , xmin,k , xmax,k ,0,

I nk nk
E g 2 X ck , j ck ,m
k 1 j 0 m 0 dt j m t 0

Tsonev, et al., JLT, 2013

55 December 23,
2013
Channel Capacity: Optimisation
Frameworks

56 23 December,
2013
Results

57 23 December,
2013
Spectral Efficiencies

58 23 December,
2013
Spectral Efficiencies, contd

59 23 December,
2013
Implications on Dimming

60 23 December,
2013
Optical Output Power

62 23 December,
2013
Spectral Efficiency of OFDM

63 23 December,
2013
Theoretical Capacity limits, contd

For 10 dB dynamic range:

(1): with optimisation, DC
bias power not included
(2): without optimisation, DC
bias power not included
(3): with optimisation, DC
bias power included
(4): without optimisation, DC
bias power included

22 July 2013
 Spatial Modulation: How does it
work?
Spectral Efficiency: Im
Signal Constellation

Im 01(00)

10(00) 00(00)
11(00)
00 (Tx0) Re
Im
01 (Tx1) Re
01(11)
10 (Tx2)
10(11) 00(11)
11(11)
11 (Tx3) Re

Spatial Constellation
66 The University of Edinburgh 12/23/2013
Spatial Modulation OFDM

67 23 December,
2013
 Thank You!

68 23 December,
2013