INTERNET ACCESS QUALITY MONITOR

Bruno P. Ramos, Vasco N. G. J. Soares and Alexandre J. P. D. Fonte

Superior School of Technology, Polytechnic Institute of Castelo Branco, Castelo Branco, Portugal

Keywords: Internet Quality of Service, Connection Speed, Distributed Systems, .NET Remoting Layer.

Abstract: Assessing the perceived Quality of Service (QoS) offered on Broadband Internet accesses from end-User

standpoint is important not only to monitor the performance, but also to assist end-Users to quantify the

effective quality offered by their Internet accesses and compare it with the quality parameters specified in

the Service Level Specification (SLS) of the connectivity service contracted to the ISPs (Internet Service

Providers). Other key benefits are end-Users being aware about the Geographic quality distribution of the

Internet accesses, pinpointing congestion links or areas, and being able to make price-performance tradeoffs

and subscribe better ISPs. In this paper, we present the design principles of a practical and comprehensible

Internet Access Quality Monitoring (IAQM) system and some aspects regarding its deployment and

security. IAQM aims at supporting accurate assessment of the performance of Internet Broadband accesses

and satisfaction of end-Users at large scale. A collection of tests are planned to measure several Quality of

Service metrics of an Internet access, such as (but not limited to) download and upload rates, latency, jitter

or DNS (Domain Name System) lookup times.

1 INTRODUCTION

As the Internet is growing it is increasingly changing

the social relationships between people, the business

relationships between customers and enterprises and

the relationships between citizens and governments.

Major advantages of the Internet are the numerous

powerful communication and multimedia tools, such

as electronic mail, instant messaging or Internet

phone, it makes available to users for fast share and

access to information in any geographic location.

High speed accesses to the Internet are thus

recognized as an essential tool for fast success. An

important fact is the recent exponential increasing of

the number broadband accesses to the Internet at

home, enterprises or governmental institutions.

One key factor driving this increasing is the

emergence of mission critical and multimedia

applications, such as VoIP (Voice Over IP), video

streaming or virtual reality, that are increasingly

demanding stringent bandwidth guarantees and so

quality of service.

Internet Service Providers (ISPs) are also

contributing for this change as they are offering

better quality of service and capacity to end-Users.

First, it is becoming common the negotiation of

Service Level Agreements (SLA) or private

agreements for increasing the quality of service and

interconnection capacity between ISPs. Second, ISPs

are also providing end-Users better Internet accesses

and services. For instance, regarding the ADSL

technology one technique for improving the lines

capacity that has been adopted by ISPs is Dynamic

Spectrum Management (DSM) by using automated

intelligent control of ADSL (Asymmetric Digital

Subscriber Line) lines parameters (DSM, 2003,

Song, 2002).

Unfortunately, most of ISPs only reveal publicly

few or partial measurements regarding the quality

and reliability of their networks due to business

constraints. The IAQM (Internet Access Quality

Monitor) project aims at answering to common

issues and concerns raised by end-Users about the

perceived Quality of Service and parameters of the

Internet accesses contracted to ISPs.

There has been relatively little work (e.g.,

(ANACOM, 2005, TCP/IQ, 2007) on the design and

deployment of tools for assessment of the

performance of Internet accesses and satisfaction of

end-Users at large scale. Most of this work relies on

an approach in which software agents installed at the

Internet accesses autonomous and asynchronously

collects measurements that can be related to the

parameters of the Internet links contracted from

197

P. Ramos B., N. G. J. Soares V. and J. P. D. Fonte A. (2008).

INTERNET ACCESS QUALITY MONITOR.

In Proceedings of the Fourth International Conference on Web Information Systems and Technologies, pages 197-201

DOI: 10.5220/0001532101970201

 SciTePress

ISPs, such as maximum download and upload rates.

But, unfortunately none has rigorously providing

measurements about the effective quality of service

of the Internet accesses and end-Users satisfaction at

large scale.

In contrast to previous approaches, IAQM

introduces significant improvements in the way it

quantifies the quality of the Internet accesses and

end-Users satisfaction. More specifically, IAQM

system follows the client-server architecture model.

It proposes a thin client –a software agent– which

basically performs the measurement tests regarding

the performance of the Internet access; and a central

server that is responsible not only for composing and

scheduling of the tests and controlling the IAQM

clients activity, but also for the subsequent analysis

of the individual user results and, above all, crossing

these results in order to allow users to compare the

quality among the available services, ISPs and

geographic areas. Others key concerns are on the

selection/invitation of end-Users to participate in the

study and on the type of tests and their scheduling

methodology.

In this paper, we present the base design

principles of IAQM and some aspects regarding its

deployment and security. We believe that IAQM

deployment path is as interesting as the envisioned

end state. That is, it is expected that IAQM would

offer several benefits to users of broadband accesses,

such as auditing the effective quality of service

offered by its ISP, the ability to make price-

performance tradeoffs and finding better ISPs to

subscribe, and providing a powerful tool for

diagnosing and troubleshooting of the Internet

accesses.

The rest of the paper proceeds as follows.

Section 2 motivates IAQM system and briefly

describes its architecture. In Section 3, we present

the IAQM architecture in detail. In Section 4, we

discuss the security issues regarding the IAQM

utilization. Section 5 reviews related work and

Section 6 concludes.

2 MOTIVATION AND SYSTEM

OVERVIEW

There are numerous factors that are driving the

growth of the number of the Internet accesses and

end-Users (eTForecast, 2007). The low cost of

Internet accesses and devices will continue to be the

main growth factor. However, there are other key

driving factors, such as the emergence of a wide

range of services, the impact of broadband

technologies on Internet accesses (e.g., cable modem

and ADSL connections) and Web-driven

productivity gains.

Recent statistics revealed that in many countries

there are more than 600 Internet users per 1000

people (eTForecast, 2007). Among these Users, it is

likely that a significant fraction has contracted its

own Internet accesses. For instance, in Portugal one

of top 15 countries in terms the number of Internet

users, this growth trend continues at a rate of about

30% a year and it estimates that there are already

more than 1.8 million of broadband customers,

among an universe of about 10 million of people

(ANACOM, 2007). Inclusively, it is expected that at

the end of 2007, 50% of Portuguese homes would

have a basic access to the Internet.

This motivated the need to analyse if there are

also a corresponding increasing in end-Users

satisfaction and in the quality of service of the

Internet accesses. Moreover, none work has

rigorously providing measurements about the

effective quality of service of these accesses and

users satisfaction.

The aim of the IAQM project is therefore to

design and provide a client-server application that

allows assessing and studying the underlying

perceived quality of basic connectivity services

offered on the Internet access methods from multiple

ISP, such as Asymmetric Digital Subscriber Line

(ADSL) and Cable, available to Home users.

The IAQM framework focuses on two distinct,

but complementary evaluations (not limited to):

• Overall evaluation and comparison of the

effective quality of service offered on

broadband accesses (ADSL and Cable) from

multiple ISPs.

• Evaluation of specific services provided by ISPs

and violation rates of SLAs, including the

assessing of geographic distribution of quality

(i.e., throughout districts or regions) and

eventual disparities among broadband accesses

from multiple ISPs.

To assess quality of service there is need to

collect data at each broadband access. The IAQM

includes two main software components. First, an

IAQM software agent is installed on computers of

end-Users invited (or that wish) to participate in the

study. This software component is launched

automatically at periods of time pre-defined and

accesses to a central server from which it downloads

the tests to run. Upon the measurements tests are

completed, it sends the measurements to the central

server. On the other hand, the central server, besides

controlling the agents activities and gives them tests

WEBIST 2008 - International Conference on Web Information Systems and Technologies

198

to run, is responsible for storing and processing the

measurements samples collected during the tests.

3 CLOSER LOOK AT THE

SYSTEM ARCHITECTURE

Figure 1 shows IAQM system components: the

IAQM Client, the IAQM Server and the IAQM Web

Server. Both the IAQM Client and the IAQM Server

were developed with the C# programming language

(MSDN, 2007), and the communication between

them is implemented using the Microsoft .NET

Remoting Layer infrastructure (Rammer, 2002). The

IAQM Web Server relies on the ASP.NET web

application framework.

Figure 1: IAQM System architecture.

The set of IAQM Client functionalities include

the monitoring and displaying of the End-user’s

download and upload rates on real time basis on its

GUI in a graphic representation (see Figure 2), and,

above all, the execution of a pre-defined set of tests

to quantify the quality of the access. IAQM Server

supplies this tests that are performed in a transparent

way, in order to be not intrusive and to not degrade

the user’s Internet experience. Each test has its own

specification, such as the type of the test and the

time of day to be executed.

The IAQM Client is responsible for accessing to

the IAQM Server that indicates the tests that it must

download and execute (Figure 3). For each of the

executed tests, the client registers the results

obtained and transmits them to the server.

IAQM Server is responsible for preparing and

scheduling tests, receiving the results sent by the

IAQM clients, assuring the validity of the user’s

credentials and results, and for the analysis of the

measurements.

Figure 2: IAQM Client monitoring download and upload

rates.

Figure 3: IAQM Client downloading test information

The tests objectives are to measure several

quality of service metrics of an Internet connection,

such as (but not limited to) download and upload

rates, latency, jitter or DNS lookup times.

Diagnosing route congestion is also an interesting

feature to be explored. The influence of user’s

computer hardware and software (e.g. operating

system, firewall, anti-virus, anti-malware), and its

local area network characteristics is also considered.

Each of the tests has a particular purpose like for

example to evaluate the upload rate using FTP, or

the HTTP access time, or the round trip time using

Ping. They are executed at specific times of the day

and on specific days of the week. Furthermore, they

are applied to servers located on the network of the

ISP’s client, and located on different ISPs on the

same country. Finally, they are also applied to

international servers. The analysis of the

measurements returned by clients will allow to

INTERNET ACCESS QUALITY MONITOR

199

identify variations on the Internet access quality on

different ISPs, regions and districts a country (e.g.,

Portugal), and also to locate points of congestion on

national and international Internet accesses.

The scheduling of the tests is based on data

statistics regarding the diurnal cycle of the End-user

accesses, popular Web sites visited and age of users

as in (NetPanel Report, 2003), using a proper

methodology as the base methodology proposed in

(ANACOM, 2005).

IAQM Server analysis of the measurements is

only available for registered users and can be viewed

either by consulting the IAQM Client, or the IAQM

Web Server. By using the IAQM Client GUI a user

may check his individual results, analyzing and

comparing them through time (day, week, month

and year). A consult of the data stored on the IAQM

Web Server allows viewing the global study,

enabling users to evaluate Internet access quality on

different ISPs, regions and districts.

4 SECURITY CONSIDERATIONS

One major requirement of the IAQM system is to

protect the user’s data and to avoid to tampering

attempts on the obtained measurements. To

accomplish this requirement, the communication

between the IAQM Client and the IAQM Server is

encrypted using an asymmetric key algorithm. .NET

Remoting Layer channel codification is also used.

Another important aspect is related to controlling

where the IAQM Client is running. In order to

guarantee the validity of the measurements, IAQM

needs to check if the tests are being executed with

the same conditions that were defined on the user’s

profile. To assure this, a tracking algorithm will

evaluate the user’s IP address, geographic place, and

the type of network where it is connected. If some

incongruity appears, the tests are not executed and a

warning is sent to the user.

To participate on the Internet access quality

study, a user must access the IAQM Web Server

where will find a register form. The correct fulfilling

of the user profile (contact details, computer

hardware and software attributes, and the ISP

contracted service) will permit to download and

install IAQM Client. During execution, the IAQM

Client will only collect information about the

computer’s hardware, IP address, monitored

download and upload rates, and results from the

tests.

5 RELATED WORK

Some studies and software solutions to evaluate the

Internet access quality have been proposed, such as

the ANACOM study (ANACOM, 2005), the

Speakeasy Inc. Speed Test (SpeakEasy, 2007), the

Bandwidth Place Speed Test (Bandwidth Place,

2007), the DSLreports.com Speed Test (Broadband

Reports, 2007), the Visualware Inc. MySpeed PC

(Visualware, 2007) and the TCP/IQ Line Speed

Meter (TCP/IQ, 2007).

There are some similarities between IAQM and

these approaches. We intend to develop a system

that will also give the possibility to make a general

study about the user’s perceived Internet quality

access and satisfaction, but as a major contribution it

will focus on particular details that have not been

exploited and studied yet.

Except the ANACOM study, the other presented

software solutions access external servers to

measure only the Internet connection speed. IAQM

aims to do a more complete study, measuring several

quality of service metrics of an Internet connection,

on download and upload accesses to servers located

on the user’s ISP and on others ISPs from the same

country or located on foreign countries, using more

accurate measurement methods.

Online internet connection speed tests from

Speakeasy Inc. Speed Test and DSLreports.com

Speed Test, do not permit to cross and compare the

individual user results with the results experienced

by other users. Since it is the user that starts the test

whenever he wants to, the test can be influenced by

Internet traffic characteristics, so it is not possible to

have a correct idea of the connection performance

though time. Line Speed Meter and ANACOM

approaches permit to analyze the Internet

performance connection over a period of time.

ANACOM concentrates its study only on

Portuguese ISPs, Line Speed Meter provides

comparison of global Internet connection speeds

throughout the world. However, further work can be

done since Line Speed Meter does not have the

possibility to directly compare country’s ISPs as

ANACOM does. But ANACOM’s study can be

enhanced, by examining the variation of Internet

access quality along a country, comparing it along

regions and districts. The number of users, their

profile, and how they were selected to run the tests is

also a weak point of these studies.

Furthermore, none of the above-mentioned

systems takes into consideration the user’s computer

hardware and software, and the user’s local area

network characteristics. However, it can be

WEBIST 2008 - International Conference on Web Information Systems and Technologies

200

interesting to evaluate if these aspects affect the

accuracy of the tests, so IAQM will also analyze

them.

6 CONCLUSIONS

In this paper, we presented the base design

principles of IAQM and some aspects regarding its

deployment and security. We believe that there are

several advantages in the use of the IAQM system,

namely in obtaining a more comprehensive view of

the real conditions and quality that End-users usually

experiment. First, IAQM will comprise large enough

number of quality metrics, which would allow to

capture and a give broader characterization of the

effective conditions and quality of the Internet

accesses when comparing with previous work.

Second, our IAQM system will enhance the

gathering of this data and the amount of information

about the Internet accesses through the use of a

specific set of tests to be performed during each

period of time, the level of communication with

servers within the ISP network or within an external

or international network. Finally, as the IAQM

server has the ability to control the exact periods of

time when the clients should perform the tests, and

their geographic location, it would allow obtaining

more consistent results. The association of the

geographic location to the quality of each access will

allow building a map with the distribution of levels

of quality of service or asymmetries throughout a

country and identifying eventual bottlenecks in the

Internet (e.g., internal links, access links or peering

links).

In short, we believe the envisioned end state of

IAQM will provide a useful study of the quality of

service on broadband Internet accesses that would

for sure benefit the customers.

REFERENCES

DSM Draft Report. December 8, 2003, San Diego, CA.

ANSI T1E1.4/2003-018R9.

Song, K B, Chung, S T, Ginis, G, Cioffi, J. Dynamic

Spectrum management for next-generation DSL

systems. IEEE Communications Magazine, 40 (10),

101–109, 2002.

eETForecast, http://www.etforecasts.com/, accessed at

November, 2007.

ANACOM, 2007, Serviço de Acesso à Internet,

http://www.icp.pt/txt/template12.jsp?categoryId=2502

02, accesssed at October 2007.

Net Panel Report, Analysis Report of Internet Market,

May 2003, http://netpanel.marktest.pt/, accessed at

October 2007.

MSDN, “Visual C# Developer Center”,

http://msdn2.microsoft.com/en-

us/vcsharp/default.aspx, accessed at October 2007.

Rammer, I., 2002, Advancing .NET Remoting, Apress,

USA, 1º Edition.

ANACOM, 2005, “Avaliação do Serviço de Acesso à

Internet – Banda Larga: ADSL e Cabo”,

http://www.anacom.pt/template12.jsp?categoryId=176

822, accessed at September 2007.

SpeakEasy, “Communications Simplified”,

http://www.speakeasy.net, accessed at October 2007.

BandwidthPlace, “Speed Test”,

http://bandwidthplace.com, accessed at October 2007.

Broadband Reports, “Speed Test”,

http://www.dslreports.com, accessed at September

2007.

Visualware, “My Speed PC”, http://www.myspeed.com,

accessed at October 2007.

TCP/IQ, “Line Speed Meter”, http://www.tcpiq.com,

accessed at October 2007.

INTERNET ACCESS QUALITY MONITOR

201