Simple Antenna Diversity techniques with inherit directional information for SDMA operation

Simple Antenna Diversity techniques with inherit directional information for SDMA operation

Project group 997:Julien GonidecThibaut IngrainFrançois NetMauro PelosiAurélie Villemont

Supervisors: Patrick Eggers

Chenguang Lu

Censor: Jesper Ø. Nielsen

2

Why diversity techniques ?

Introduction

Wireless technologies comparison depending on data rate and mobility

3

Why WLAN ?

A widespread technology Problems of security New localisation services Convergence of technologies

Introduction

4

Choices made

802.11G standard Open office environments Jitter diversity Implementation of diversity techniques only

at the base station Algorithm will provide directional information

Introduction

5

Experiment process

Study the recquired theory Apply the Jitter Diversity algorithm to deduce

tendencies Model a more realistic channel model and

apply the jitter diversity on it Study the gain provided by the diversity See how the algorithm can provide

directional information

Introduction

6

Antenna Array

Ordered repetition in space of identical radiating elements

5 degrees of freedom:– Geometry– Element spacing– Excitation signal amplitude– Excitation signal phase– Element pattern

Theory

7

Antenna Array: features

Portability– Compact configuration

Adaptability– Feeding network– Dynamic beamforming

Theory

8

SDMA

Multiplexing technique used together with another multiple access process

Spatial separation of the users at the BS

Advantages: - improves antenna gain

- reduces interferences

- improves capacity

Theory

9

SDMA

Principle: concentrate on the desired user and nullify the others

Directional information

Theory

10

Diversity techniques

Temporal based techniques– Time diversity– Frequency diversity

Antenna pattern techniques– Space diversity– Polarisation diversity– Angle diversity

Direction information bearing techniques– Phase diversity– Beam diversity– Jitter diversity

Theory

11

Jitter diversity

Aim: reduce deep fading effects

Principle: slightly move the antenna beam

Advantages: simple and easy to implement

Theory

12

Jitter diversity algorithm

INITIALISATION

Full environment scanto find the global maximum

JITTER PROCESS

Find nearest local maximum

CONDITION

end of operation ?

END

CONDITION

- power level under threshold ?- time deadline reached ?

UPDATE

- full scan again- local scan using previous steps

data

YES

NO

NO

YES

Theory

13

Fading types

Large- and medium-scale fading– Pathloss– Shadowing

Small-scale or multipath fading– Frequency selective fading– Flat fading– Fast fading– Slow fading

Theory

14

Large- and medium-scale fading

Pathloss: average power decay caused by distance d between Tx and Rx

Where γ is the pathloss exponent

Shadowing: absorption by the local surrounding media

PL d d

Pathloss Shadowing Multipath fadingRapidity of fluctuations of signal’s strength

Theory

15

Small-scale fading

Frequency

selective fading

Flat fading

Relationship between signal’s bandwidth and coherence bandwidth

Time-dispersive nature of the channel

Fast fading

Slow fading

Relationship between signal’s and channel’s time-rate of change

Time-varying nature of the channel

Theory

16

Small-scale fading

Multipath time

delay

Flat fading

Frequency selective fading

Bc

Bs

Bc

Bs

Doppler spread

Fast fading

Slow Fading

Tc

Ts

Tc

Ts

Theory

17

Models of small-scale fading

Rayleigh fading: NLOS between Tx and Rx

Rice fading: LOS between Tx and Rx

Nakagami-m fading:

2

_ 2 2.exp

2Rayleigh z

x xp x

2 2

_ 02 2 2

( ).exp .

2Rice z

x x v x vp x I

2 1 2

_

2 1 p .exp

(m) 2

m m

Nm z mp p

m x mxx m

Theory

Rician Factor K

18

Models of small-scale fading

Rician PDF

Theory

19

Models of small-scale fading

Nakagami-m PDF

Theory

20

Scatterers repartition: Lee’s model

Scatterers uniformly spaced on a circle centred on the MS

Useful for correlation calculation Not the best model for indoor description

Theory

21

Scatterers repartition: GBSB model

Spatial scatterer density function

GBSB Elliptical Model

Theory

22

Scatterers repartition: Saleh-Valenzuela model

Accurate indoor channel representation Clustered scatterers

Extended model including AOA

Theory

23

Choices for our model

Rayleigh fading

Clustered scatterers

Elliptical repartition

Theory

24

Jitter diversity simulation in a simplified environment

Steps of the simulation

Modelling a simplified indoor channel Generation of an ideal antenna pattern Jitter process description Results and tendencies

Monte-Carlo simulations

the user’s location is randomly defined at each step

Simulation 1

25

Environment implementation (1)

Clustered scattering Investigations concentrated on rays from an unique cluster AOA power distribution approximated by a Laplacian

distribution PowerLaplace_a(θAOA)

Environment response

Where The amplitude is defined by

The phase is defined by

,, , . er AOAj xAOA er AOAer x x e

_

1

., . 1 ,

1 ,R

R Laplace a AOAk

er i AOAk i AOAkN

n AOAkn

N Px temp x

temp x

, 2 . 2 ,er AOA AOAx temp x

Simulation 1

26

Environment implementation (2)

“a“ parameter controls the shape of the environment

10-6 < a < 10-1

BWenv: half-power width of the mean environment response

Simulation of various type of environment by varying the a parameter

Simulation 1

27

Antenna pattern

Choice of an ideal beam pattern (no side and back lobes)

Amplitude of the pattern

“α“ parameter controls the antenna beamwidth

sin,

0AOA BO AOA BO

AOA BO

ifa

otherwise

Simulation 1

28

Transfer function

At each realisation all beam’s orientation are performed

Discrete transfer function

Influence of the environment width on the fades

1

, , . ,AOAN

BO AOAn AOAn BOn

h x er x a

Simulation 1

29

Jitter process

We want to compare 3 different algorithms: JRDA (Jitter with respect to the Reference Direction Algorithm) BPP (best possible process algorithm) FB (fixed beam algorithm) as a reference

Explanation of the JRDA process

1. Reference direction θrefk is found at the kth step

2. is compared to and

3. The orientation of the maximum value is chosen θpathk

4. is the whole of the collected h module

,k refkh x ,k refk jitth x

,k refk jitth x

, pathh x

Simulation 1

30

JRDA results

Simulation 1

31

Standard deviation of the JRDA

Simulation 1

32

Total power gain of the JRDA

Total power gain at the 1% level of probability:

We define the total power gain at the 1% level of probability as the difference between the cumulative density values of

and at the 1% level of probability

( , )path dBh x ( , )FB dB

h x

_1%JRDATPG

Simulation 1

33

Diversity gain at the 1% level of probability : Definition of the normalised power :

_( , ) ( , ) ( , )path path pathdB norm dB dBh x h x h x We define the diversity gain at the 1% level

of probability as the difference between the cumulative density values of and at the 1% level of probability

_( , )path dB normh x

_( , )FB dB normh x

Diversity gain of the JRDA (1)

Simulation 1

34

Diversity gain of the JRDA (2)

Simulation 1

35

Simulations with a more realistic channel model

Simulation aim:

• Derive a more realistic channel model

• Compare the efficiency of diversity techniques in the aforementioned channel model

• Provide simple Directional Information

Simulation 2

36

Environment description:modified elliptical model

Simulation 2

37

Beam scanning power at the base station over spatial iterations

Simulation 2

38

Location of the maxima of the beam scanning power over spatial iterations

Simulation 2

39

Jittering algorithms

Optimum jittering We do periodical updating of the global maxima on

the beam scanning power

Dominant jittering We do not have updating; we only initialise the

algorithm with an angular value corresponding to the angular center of gravity of the dominant cluster

Simulation 2

40

Reference algorithm

Dominant fixed beam algorithm (no diversity)

• In this case we choose a fixed beam orientation for

our antenna, which will remain the same for all the

spatial iterations of the mobile station; the algorithm

is first initialised with the angular position of the

center of gravity of the dominant cluster

Simulation 2

41

Diversity gain calculation

Simulation 2

42

Simulation sets

Dominant cluster’s angular width variation Scatterer’s complex gain variation Scatterer’s distribution variation Modified elliptical model with 4 clusters

Simulation 2

43

Modified elliptical model with 4 clusters

Simulation 2

44

Simulation set 1

•Scatterer’s complex gain variation

•50°< Dominant cluster < 90°

Simulation 2

45

Simulation set 2

•Scatterer’s distribution variation

•50°< Dominant cluster < 90°

Simulation 2

46

Simulation set 3

•Scatterer’s complex gain variation

•60°< Dominant cluster < 90°

Simulation 2

47

Simulation set 4

•Scatterer’s distribution variation

•60°< Dominant cluster < 90°

Simulation 2

48

Simulation set 5

•Scatterer’s complex gain variation

•70°< Dominant cluster < 90°

Simulation 2

49

Simulation set 6

•Scatterer’s distribution variation

•70°< Dominant cluster < 90°

Simulation 2

50

Simulation results

All the previous results show that we get a fair diversity gain even when we change the width of the dominant cluster.

Though the optimum jittering algorithm shows the best performances, the dominant jittering, that has a lower complexity, behaves also good. This would suggest the sub-optimality of the dominant jittering, that though has no updating part in the jittering process, leads to good results.

Simulation 2

51

Simple directional information

This information will be provided by the dominant jittering algorithm

The steering beam path on the beam scanning power over spatial iterations is then investigated

Simulation 2

52

Steering beam over spatial iterations

Simulation 2

53

Directional information results

As we can see from the previous figure, the dominant jittering is nearly confined in the dominant cluster neighbourhood

If sometimes the jittering path escapes from the dominant cluster bounds, it means that the algorithm simply finds less fading in that direction

A suggestion for future investigations would be a sort of inter-cluster hand-over, because during the moving of the mobile station the dominance of a new cluster may replace the previous one

Simulation 2

54

Conclusions

A simplified channel model has been investigated, with the main conclusion that the jitter diversity takes advantage of a spread angular repartition of the rays together with the use of a narrow antenna beamwidth

A more realistic channel model has been studied, with the definition of a modified elliptical model including the effect of scatterers clustering

A fair diversity gain has been found nearly in all the simulations, suggesting the sub-optimality of the dominant jittering algorithm

Simple directional information has been provided

Conclusion

55

Future work

Introduction of a realistic antenna pattern Variation of the input parameters of the

simulations in order to have more average information

Extension of directional information and study of a possible inter-cluster hand over

Conclusion