Multi-Standard Convergence in Mobile Terminals

courageouscellistAI and Robotics

Oct 29, 2013 (3 years and 10 months ago)

95 views

15

March
200
5

Multi
-
Standard Convergence in
Mobile Terminals



(Master Thesis)



Presenter: Shakeel Ahmad

venue: GK Workshop Waldau, Germany.

Supervisors: M.Sc. Chunjiang Yin

Prof. Dr. Hermann Rohling


Department of Telecommunication

Technical University Hamburg Harburg


Presenter:
Shakeel Ahmad

2

Contents


Introduction and Motivation


Tasks


Convergence Manager


Simulation Scenario and Considerations


Standard Selection Algorithms


Implementation in MLDesigner


Simulation Results


Conclusions



Presenter:
Shakeel Ahmad

3

Introduction & Motivation


An increased demand of mobile Internet


A wired
-
like Internet service while on move… a big challenge


The challenge seems hard to be met with pure 3G deployment


WLAN a good candidate but suffers from low coverage



In 4G system one proposal considers the convergence of the
existing wireless standards


Convergence Manager


Presenter:
Shakeel Ahmad

4

Introduction & Motivation (2)


Overall performance can be improved


P
otential

benefits

for

end
-
u
sers,

n
etwork

o
perator

and

the

s
ervice

p
rovider


Presenter:
Shakeel Ahmad

5

Tasks


Definition

of

a

simulation

scenario

deploying

multi
-
standard

convergence


Implementation

of

simulation

scenario

in

MLDesigner


A

quantitative

analysis

of

the

potential

benefits

offered

by

multi
-
standard

convergence

only

in

mobile

terminals


Standard

selection

algorithms

and

comparison


Delay

performance


Request

discarding

rate

performance


Presenter:
Shakeel Ahmad

6

Convergence Manager


Function:


Enables the convergence of wireless standards.


Location:


A crucial issue from architecture point of view


Some proposed locations for Convergence Manager (CM):


Only in Mobile Terminal


Only in the network side


Can be split into a network and Mobile Terminal part

Standard Selection


Presenter:
Shakeel Ahmad

7

Simulation Scenario & Considerations


A busy road about 1.5km long in a city and is crossed by some other roads


Technological Scenario


Two standards, HSDPA (for UMTS) and HL2 (WLAN) were considered.


HSDPA is available throughout and HL2 is available at crossings (100 m radius)


User Scenario


Uniform spatial distribution along the road, moving with a constant speed.


Users make request for a service according to Poisson process


Service Scenario


Generic file download service



Presenter:
Shakeel Ahmad

8

Contents


Introduction and Motivation


Tasks


Convergence Manager


Simulation Scenario and Considerations


Standard Selection Algorithms


Implementation in MLDesigner


Simulation Results


Conclusions



Presenter:
Shakeel Ahmad

9

Switched Algorithm


Upon arriving a request, the highest data rate bearer is selected.


Download starts immediately and can take place via multiple standards


Switching between standards while mid of file download (complications involved)

Standard Selection Algorithms

UMTS Only

UMTS Only

UMTS / HL2

UMTS / HL2

Request

Request

Request

Request

Request

Request

Request

Request

Request

Request

time

time

time

Switched

Current

Location

UMTS Only

Road

Three algorithms for standard selection were considered [1]



[1] MultiStandard approach for enhanced communications service


provision to rail commuter
IST Mobile Communications Summit,

Lyon, France
.



Location Algorithm


Uses additional knowledge of users‘s mobility, geographical coverage of standards and
the mean data rate, to make more intelligent decision


Download may not start immediately and always takes place via single standard

Current Algorithm


Upon arriving a request, the highest data rate bearer is selected.


Download starts immediately and takes place via single standard

Initial


delay


Presenter:
Shakeel Ahmad

10

Implementation in MLDesigner

Simulation Scenario


Presenter:
Shakeel Ahmad

11

Implementation in MLDesigner (2)

An instance of Mobile Terminal


Presenter:
Shakeel Ahmad

12

Contents


Introduction and Motivation


Tasks


Convergence Manager


Simulation Scenario and Considerations


Standard Selection Algorithms


Implementation in MLDesigner


Simulation Results


Conclusions



Presenter:
Shakeel Ahmad

13

Multi
-
Standard Single
-
User Case


Request arrival Process is Poisson with mean rate of 1/10 requests per second, File
Size Fixed


It is supposed that user is moving with constant speed (10 m/s), Red Signal On
Probability=0.0


Mean Initial Delay Vs File Size & Mean Total File Download Time Vs File Size


Presenter:
Shakeel Ahmad

14

Multi
-
Standard Single
-
User Case (2)


Request arrival is Poisson with mean rate of 1/10 requests per second, File Size
Fixed


It is supposed that user is moving with constant speed (10 m/s), Red Signal On
Probability=0.0


Mean Request Discarding Rate Vs File Size


Presenter:
Shakeel Ahmad

15

HotSpot Effect


Request arrival Process with mean 1/10 requests per second, File Size Fixed


It is supposed that user is moving with constant speed (10 m/s),
Red Signal On
Probability =0.1 (Red Signal is On for 15 sec)


Mean Total Download Time Vs File Size

Crossing points are
good place to install
HL2 APs


Presenter:
Shakeel Ahmad

16

Multi
-
Standard Multi
-
User Case


Request arrival Process with mean 1/10 requests per second, File Size = 125 KB


Users spatial distribution is uniform along the road,


It is supposed that users are moving with constant speed (10 m/s), Red Signal On
Probability =0.0


Mean Total Download Time & Mean Request Discarding Rate Vs numUsers


Presenter:
Shakeel Ahmad

17

Problem with Location Algorithm


Proposed Solution:

History Based Location Algorithm


Mean data rate for the current file download via
standard x is inferred from previous file
download via standard x.

UMTS Only

UMTS Only

UMTS / HL2

UMTS / HL2

Request

Request

Request

Request

time

Location

UMTS Only

Road




Wrong data rate estimation leads to wrong standard selection


Presenter:
Shakeel Ahmad

18

Performance of History Based Location Algorithm


Improvement by applying History Based Location Algorithm


Mean Total Download Time & Mean Request Discarding Rate Vs numUsers


Presenter:
Shakeel Ahmad

19

Conclusion


Benefits of Convergence Manager


Improved QoS and System performance


Easy to realize


no standardization efforts



Comparison of standard selection algorithms


Switched algorithm lower bound of performance
-

Location algorithm closest
match


Location Algorithm


Poor performance in multi
-
users


History Based Location Algorithm
-
one Possible Solution



Presenter:
Shakeel Ahmad

20

Conclusion (2)


Modular and Extendable Simulation Setup


A useful off
-
shelf component


The framework can be easily extended e.g., for other wireless standards, new
standard selection algorithms and new scheduling policies can be incorporated
and tested easily



Future Work


The assumption that CM knows precisely about user‘s mobility, location and
geographical coverage of wireless standards may not be realistic


Source of
error.


New standard
-
selection algorithms for different scenarios and services


Presenter:
Shakeel Ahmad

21



Thanks for listening