p erformance e valuation of b ase s tation a pplication o ptimizer nana ginzbourg instructor: dr....

36
PERFORMANCE EVALUATION OF BASE STATION APPLICATION OPTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

Upload: anis-brown

Post on 01-Jan-2016

217 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

PERFORMANCE EVALUATION OF BASE STATIONAPPLICATION OPTIMIZER

Nana Ginzbourg

Instructor: Dr. Ronit Nossenson

Internal Instructor: Dr. Tami Tamir

Page 2: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

OVERVIEW Introduction

LTE Network

NS 2 Simulator

Base Station Application Optimizer

Implementation Model – network topology

Traffic generation and cache implementation

Simulation

Trace analyzers

Results

Conclusions and Future work

Page 3: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

INTRODUCTION

Cellular operator take advantage of new technologies and offer rich services to advanced handset

Backhaul lines are still limited to 2 Mbps capacity

Leads to backhaul bottleneck problem

Page 4: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

POSSIBLE SOLUTIONS

Backhaul lines upgrade

Application solution Data reduction No need in upgrade Integrated solutions

Base Station Application Optimizer Analyze and optimize user data in application

level Prevent unnecessary data travelling through

backhaul network Less painful upgrade

Page 5: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

PROJECT GOALS

Implement Base Station Application Optimizer over an LTE network in NS-2 simulator

Evaluate solution performance in various simulations: Realistic mixture of applications Different cache hit rates Changing network load – number of users Different queue sizes

Page 6: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

LTE NETWORK

Page 7: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

BASE STATION APPLICATION OPTIMIZER

Page 8: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

NS 2 SIMULATOR

Open Source event driven simulator History:

1989 – REAL network simulator 1995 – VINT Project with DARPA

Support a big variety of protocols and keeps growing

Simulator Implementation: Backend in C++ - compiled hierarchy Frontend in Otcl – interpreter with similar

hierarchy Simulation scripts - .tcl

Page 9: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

NS 2 SIMULATOR

Page 10: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

BASE STATION APPLICATION OPTIMIZER IMPLEMENTATION

Open source: http://code.google.com/p/base-station-

application-optimizer/

Four stages: Network Topology definition Traffic generation Configurations and simulations over different

parameters Trace analysis

Page 11: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

NETWORK TOPOLOGY DEFINITION

Wired Nodes eNode-B Serving Gateway PDN-GW Server

Wireless Nodes User equipment

DropTail queuing algorithm Oversubscription factor at S1 interface is 1/3

Page 12: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

TRAFFIC GENERATION APPLICATIONS DISTRIBUTION

Page 13: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

TRAFFIC GENERATION IN NS 2

GeneratorPacket Size

(bytes)

Protocol

Rate(kbps)

Parameters

VoIPCBR200UDP64

WebPareto1040TCP300ON – 200msOFF – 2000msShape – 1.3

Video

Pareto1300TCP600ON – 2000msOFF – 2000msShape – 1.5

FilePareto1500TCP100ON – 2000msOFF – 200msShape – 1.7

Page 14: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

TRAFFIC GENERATIONCACHE IMPLEMENTATION

Page 15: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

CONFIGURATION AND SIMULATIONS

All simulation parameters are configurable Command line Configuration.tcl

Run simulations using main.tcl and override relevant parameters Number of end users Buffer size Hit rates per application Users distribution per application

Page 16: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

TRACES ANALYSIS

Page 17: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

TOTAL BANDWIDTH ANALYZER

Output: total number of bytes per application type over

S1 interface

Implementation: Combination of shell and awk scripts Summarize packets sizes over S1 interface using

source and destination id Use flow id to separate results per application

type and hit rate scenarios

Page 18: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

BANDWIDTH PER SECOND ANALYZER

Output: Total number of bytes per second per application

type over the S1 interface

Implementation: Awk script Use an array[simulation_length_seconds] Summarize packet sizes according to source and

destination nodes, flow id and packet size.

Page 19: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

END-TO-END DELAY AND DROPPED STATISTICS ANALYZER

Output: End-to-end delay Dropped packets percentage

Implementation: Array per application type Packet id indicates the index in the array Send event -> save start time Receive event -> save end time Dropped event -> update with -1

Page 20: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

RESULTS

Primary Simulation 20, 50 and 100 users User Equipment links buffer size – 10 packets

Cache Hit rate impact of results Standard hit rates: 20% web, 40% video and files Increase and decrease hit rate by 10%

Buffer Size impact on results Increase and decrease buffer size in 25% Keep medium hit rate

Page 21: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

TOTAL TRANSFERRED KB WITH 20, 50 & 100 CONCURRENT USERS

Page 22: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

DATA REDUCTION PER APPLICATION TYPE

Page 23: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

DATA REDUCTION OVER TIME

Page 24: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

DATA REDUCTION FACTOR

UsersApplication

OPTIMIZEDKB

TraditionalKB

Reduction Factor

20Web1524871812651.19

Video4022638031612.00

File Sharing2493024990972.00

VoIP43190431901.00

Total84724215267131.80

50Web4140025198291.26

Video110089918964561.72

File Sharing74273312418151.67

VoIP1295701295661.00

Total238720437876671.59

100Web7276809643491.33

Video217619625056611.15

File Sharing149010920294261.36

VoIP3022592947930.98

Total469624357942281.23

Page 25: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

DROPPED STATISTICSUsersApplicationTraditional

KBOptimizedKB

#Drops %Dropped #Drop %Dropped

20Web00.00%00.00%

Video00.00%00.00%

File Sharing00.00%00.00%

VoIP190.05%00.00%

Total190.00%00.00%

50Web1140.07%1090.07%

Video2910.05%1880.02%

File Sharing2490.06%1050.03%

VoIP650.06%620.06%

Total7190.06%4640.03%

100Web59541.96%1650.05%

Video123331.57%6030.03%

File Sharing95281.50%3740.05%

VoIP125034.96%2630.10%

Total403182.04%14050.04%

Page 26: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

AVERAGE DELAY

UsersApplication

OPTIMIZEDKB

TraditionalKB

Reduction %

20Web0.39040.462315.56%

Video0.11390.462175.35%

File Sharing0.23600.462248.93%

VoIP0.46090.46120.06%

50Web0.36880.463620.45%

Video0.14400.463368.92%

File Sharing0.28090.463439.39%

VoIP0.46170.46280.23%

100Web0.34800.471326.16%

Video0.14480.471069.26%

File Sharing0.28320.471039.87%

VoIP0.46510.46910.86%

Page 27: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

IMPACT OF CACHE HIT RATE ON DATA REDUCTION FACTOR

UsersApplicationLow Hit Rate

Medium HitRate

High HitRate

20Web1.211.191.53

Video1.602.001.99

File Sharing2.012.002.01

VoIP1.001.001.00

Total1.611.801.87

50Web1.271.261.27

Video1.581.721.89

File Sharing1.681.671.89

VoIP1.001.001.00

Total1.531.591.72

100Web1.191.331.27

Video1.151.151.27

File Sharing1.281.361.52

VoIP0.980.980.98

Total1.191.231.33

Page 28: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

IMPACT OF CACHE HIT RATE ON DROP PACKETS PERCENTAGE

Application%DropTraditional

%DropLow Hit Rate

%DropHigh HitRate

Web1.96%0.26%0.15%

Video1.57%0.12%0.05%

File Sharing1.50%0.20%0.12%

VoIP4.96%0.46%0.23%

Total2.04%0.17%0.09%

Page 29: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

IMPACT OF CACHE HIT RATE ON AVERAGE DELAY

Application %ReductionLow Hit Rate

%ReductionMediumHit Rate

%ReductionHigh HitRate

Web19.89%26.16%25.24%

Video67.37%69.26%73.32%

File Sharing36.55%39.87%46.57%

VoIP0.40%0.86%0.98%

Page 30: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

IMPACT OF BUFFER SIZEON DATA REDUCTION FACTOR

UsersApplication Small Buffer Size

Medium BufferSize

Large Buffer Size

20Web1.531.191.53

Video1.992.001.99

File Sharing2.012.002.01

VoIP1.001.001.00

Total1.871.801.87

50Web1.271.261.27

Video1.911.721.90

File Sharing1.891.671.89

VoIP1.001.001.00

Total1.731.591.72

100Web1.261.331.28

Video1.301.151.28

File Sharing1.511.361.57

VoIP0.980.980.98

Total1.331.231.34

Page 31: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

IMPACT OF BUFFER SIZE ON DROPPED PACKETS PERCENTAGE

Application

%DropTraditionalSmall buffer

%DropOptimizedSmall buffer

%DropTraditionalLarge Buffer

%DropOptimizedLargeBuffer

Web2.17%0.09%0.11%0.05%

Video1.70%0.05%0.16%0.05%

File Sharing1.75%0.12%0.24%0.09%

VoIP4.09%0.21%0.17%0.16%

Total2.10%0.08%0.18%0.07%

Page 32: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

IMPACT OF BUFFER SIZE ON AVERAGE PACKETS DELAY

Application %ReductionSmall BufferSize

%ReductionMediumBuffer Size

%ReductionLarge BufferSize

Web25.31%26.16%25.20%

Video73.48%69.26%73.30%

File Sharing46.57%39.87%46.55%

VoIP1.08%0.86%1.00%

Page 33: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

CONCLUSIONS

Open Source implementation of LTE model Configurable Easy to install and use Provides functionalities like different application

traffic generation and users distribution Base Station Application Optimizer

implementation Trace Analyzers

Performance Evaluation: Total bytes transferred reduction factor: 1.2-1.8 Average delay reduced by 75% Packet loss percentage reduced from 2% to

0.05%

Page 34: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

FUTURE WORK

Add User mobility support:

Implement basic handover process in NS 2

Support several deployment modes of Base

Station Application Optimizer in an LTE network

Implement new traffic generator in NS 2

Page 35: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

REFERENCES Patrick Donegan, "Backhaul Strategies for Mobile Carriers", In

Heavy Reading , Vol. 4 No. 4, 2006. Ronit Nossenson, “Base Station Application Optimizer”, The

International Conference on Data Communication Networking 2010 (DCNET), Athens, Greece.

Holma, H., and Toskala, A., LTE for UMTS – OFDMA and SC-FDMA Based Radio Access, John Wiley & Sons Ltd, United Kingdom, 2009.

Qin-long Qiu, Jian Chen, Ling-di Ping, Qi-fei Zhang, Xue-zeng Pan, 2009, LTE/SAE Model and its Implementation in NS 2, 2009 Fifth International Conference on Mobile Ad-hoc and Sensor Networks, Fujian, China, pp. 299-303. LTE implementation: http://code.google.com/p/lte-model/

The Network Simulator (NS2), http://isi.edu/nsnam/ns/

Page 36: P ERFORMANCE E VALUATION OF B ASE S TATION A PPLICATION O PTIMIZER Nana Ginzbourg Instructor: Dr. Ronit Nossenson Internal Instructor: Dr. Tami Tamir

REFERENCES Introduction to Network simulator NS 2" by

Teerawat Issariyakul and Ekram Hossain, 2009 Springer Science+Business Media, LLC

Project source code and results: http://code.google.com/p/base-station-application-optimizer/

Allot Mobile Trends, Global mobile broadband traffic report, H1/2011.

GNU Awk, http://www.gnu.org/s/gawk/manual/gawk.html