A DISTRIBUTED SYSTEM FOR THE INTEGRATED

MANAGEMENT OF HETEROGENEOUS WIRELESS

NETWORKS

Nikolaos Koutsouris, George Koundourakis, Louisa Papadopoulou, Dimitris Kouis, Vera

Stavroulaki, Nikolas Mitrou

School of Electrical and Computer Engineering, National Technical University of Athens, 9 Heroon Polytechneiou Street,

Zographou 15773, Greece

Panagiotis Demestichas

Department of Technology Education & Digital Systems, University of Piraeus, Piraeus, Greece

Keywords: Composite Radio, GPRS, WLAN, DVB-T.

Abstract: In a composite radio environment, different wireless access techn

ologies can be co-operating components of

a combined heterogeneous infrastructure. The exploitation of a wireless system, operating in a composite

radio context, requires upgraded service and network management capabilities. This paper presents an

integrated management system and gives evidence of its capability of optimising service delivery and traffic

distribution in a prototype composite radio environment comprised of three different wireless network

technologies, i.e., GPRS, 802.11b WLAN, and DVB-T.

1 INTRODUCTION

Wireless/mobile communications continue to attract

immense research and development effort

(Varshney, 2000). Major technical evolutions

include the migration towards 2.5G and 3G mobile

communications, the introduction of Broadband

Radio Access Networks (BRAN) (Varshney, Vetter,

2000), and the advent of Digital Video Broadcasting

(DVB). Moreover, the composite radio (CR) or

“wireless beyond 3G”, concept has emerged, in an

attempt to exploit further the potential of these

individual wireless technologies. It assumes that

mobile, BRAN and DVB networks can be co-

operating systems in a CR infrastructure. Users,

equipped with multi-mode or reconfigurable

terminals, access their services through the most

appropriate, in terms of cost and Quality of Service

(QoS), radio network.

This paper presents an integrated management

syste

m that allows the provision of enhanced

applications in composite radio environments

(CREs) (work conducted within the IST Project

CREDO). The paper gives measurement results of

the system’s operation that prove its efficiency and

its ability to optimize the use of the composite radio

infrastructure. The rest of this paper is organized as

follows. Section 2 describes the network architecture

of the CR infrastructure. Section 3 presents the

management platform that is essential for the

operation of different wireless segments as parts of

the CRE, while section 4 introduces the terminal

functionality for enabling the exploitation of the

potential of a CRE architecture. Section 5 presents

results from various test case scenarios. Finally,

concluding remarks are given in section 6.

2 NETWORK ARCHITECTURE

The composite radio environment examined by the

project consists of three different radio access

technologies, GSM/GPRS, IEEE 802.11b WLAN

and DVB-T. This section demonstrates the general

network architecture for the exploitation of all these

wireless systems operating in a CR context. As

depicted in Figure 1, the private networks of all the

Koutsouris N., Papadopoulou L., Koundourakis G., Kouis D., Stavroulaki V., Demestichas P. and Mitrou N. (2004).

A DISTRIBUTED SYSTEM FOR THE INTEGRATED MANAGEMENT OF HETEROGENEOUS WIRELESS NETWORKS.

In Proceedings of the First International Conference on E-Business and Telecommunication Networks, pages 32-39

DOI: 10.5220/0001396400320039

 SciTePress

involved radio access technologies are

interconnected either through a specific router or by

means of the public IP network (GSM/GPRS case).

Figure 1: Composite Radio Environment Architecture

The functionality of the platform includes the

following features:

 Management systems for each radio access

technology. These systems, called Network

and Service Management Systems (NSMSs)

are located in the relative subnets, but they

can inter-communicate and cooperate.

 Appropriate terminals, capable of

communicating over different wireless

technologies. These multimode terminals are

equipped with the required intelligence for

taking decisions, performing measurements

and interacting with the local NSMS. The

management system of the terminals used in

the specific project, called Terminal Station

Management System (TSMS), as well as the

protocol implemented for the interaction

with the NSMS are presented in section 4.

 Content servers for retrieving information

relative to the applications and services

provided.

 IPv4 backbone solution, selected for reasons

explained below. Consequently, a Mobile

IPv4 infrastructure is employed for the

mobility management, especially during

inter-system handovers. The home network

(Figure 1), hosts the Home Agent (HA),

while the Foreign Agents (FAs) are located

in the corresponding subnets (WLAN and

DVB-T). Moreover, the HA has been

properly modified (with advanced tunneling

functionality), thus enabling it to cooperate

with the GPRS Network Address

Translation (NAT) gateway. Also, proper

modifications to the software of the DVB-T

FA for enabling the establishment of the

return channel were realized. The return

channel is required due to the unidirectional

nature of the DVB-T functionality. In the

specific case, the wireless medium that acts

as the missing uplink is the GPRS or the

WLAN network.

It is obvious that IPv4 is selected everywhere,

although IPv6 would be more convenient for the

whole architecture because there is no need for

including foreign agents and the NAT gateway is not

necessary. The reasons for choosing IPv4 are the

following (as studied at the time of the

implementation of the project):

 IPv4 is much more widely deployed and

multiple commercial products and networks

are based on Mobile IPv4. On the other

hand, IPv6 networks are still in development

and Mobile IPv6 is not standard yet.

 The commercial GPRS segment and the

commercial DVB-T products used do not

support IPv6.

 The applications’ clients and servers used

are also IPv4 based.

3 MANAGEMENT SYSTEM

FUNCTIONALITY

This section intends to provide useful information

related to the Network and Service Management

System (NSMS) introduced previously. As

presented in Figure 2, NSMS consists of two main

modules, namely Session Manager and Network

Manager.

Network and Service Management System (NSMS)

Network Manager

Monitoring and

Configuration

Monitoring and

Configuration

Resource

Brokerage

Resource

Brokerage

Service

Management

Service

Management

Session Manager

radio access

technology

independent

radio access

technology

dependent

Mid-term

functionality

Short-term

functionality

Network and Service Management System (NSMS)

Network Manager

Monitoring and

Configuration

Monitoring and

Configuration

Resource

Brokerage

Resource

Brokerage

Service

Management

Service

Management

Session Manager

radio access

technology

independent

radio access

technology

dependent

Mid-term

functionality

Short-term

functionality

Figure 2: Management system architecture

Session Manager is the module responsible for

the interface with the terminal, that is, the interface

between NSMS and TSMS. Additionally, Session

Manager issues recommendations to the terminal on

the best network and the provided QoS level. These

recommendations are based on lookup operations

and/or “light” optimization problems.

On the other hand, Network Manager is

responsible for the monitoring of the managed

network infrastructure. It also assesses the relevant

network and service-level performance, and

dynamically finds and imposes the appropriate

A DISTRIBUTED SYSTEM FOR THE INTEGRATED MANAGEMENT OF HETEROGENEOUS WIRELESS

NETWORKS

traffic distribution, through which the service

management requests or new service area conditions

are handled in the most cost-efficient manner. As

depicted in Figure 2, Network Manager includes the

following entities: Monitoring and Configuration,

Resource Brokerage, and Service Management.

3.1 Session Manager

As already mentioned, Session Manager is the

NSMS component responsible for performing all

operations concerning the communication between

the NSMS and the terminal. It also holds

information about the active terminals that are

served by each network, and also about the quality

level assigned to them. Based on that information,

and on consequent calculations, Session Manager

issues recommendations to the terminals on

choosing the best available network for the provision

of a particular service. Thus, Session Manager

addresses a short-term optimization problem,

targeted to the assignment of the user terminal to a

specific network. The solution of this optimization

problem enables the sophisticated selection of the

appropriate radio technology, for a specific user,

through which services can be obtained efficiently in

terms of cost and QoS, in near real time.

The optimization problem addressed by the

Session Manager relies on the following input data:

(a) the set of services the user is requesting and the

corresponding set of quality levels at which these

services are requested; (b) the profile of the user

requesting the set of services (this includes

parameters such as the maximum price that the user

is willing to pay for the requested services); (c) the

network policies, which mainly involves the cost

deriving from the assignment of user demand to

several quality levels and possible inabilities of a

network to handle a specific service.

The optimization process carried out by the

Session Manager should result in an allocation of the

requested services to specific quality levels, and to

specific networks. The calculation of these two

allocations should optimize an objective function,

which is associated with the quality levels at which

each service will be provided, and the utility

deriving from the assignment of the user demand to

high quality levels. These allocations are bound to

certain constraints, such as the capabilities of the

user terminal, or the limit to the overall price that the

user is willing to pay during usage of the composite

radio system.

3.2 Network Manager

This section presents in more detail the three entities

comprising the Network Manager. The Resource

Brokerage entity has the general functionality of

coordinating all the other entities of the NSMS so as

to handle various conditions, such as congestion in a

certain service area. Apart from this, Resource

Brokerage has also an important role as regards to

the efficient communication and co-operation of

affiliated network providers in a composite radio

environment, since it enables and assists the latter in

exchanging, and negotiating on, sets of offers.

The Service Management entity provides

optimization functionality for determining the

appropriate service configuration (allocation of

services to QoS levels) and aggregate traffic

distribution (allocation of traffic to networks). In

contrast with Session Manager short-term

optimization, this is a mid-term optimization

procedure, as it is explained in the sequence.

The functionality of the Service Management

entity is similar to the functionality of the Session

Manager, as related to the network selection of the

TSMS. The difference between the algorithms of

each entity is that Service Management provides a

decision for a redistribution of the users of a service

area, due to congestion, and not a recommendation

to a single TSMS about the best network choice. As

for the input and output data and the constraints,

Service Management uses the same information

described in the Session Manager’s section, but for

all the involved TSMSs.

The operation of Resource Brokerage and

Service Management entities is independent from

the underlying radio access technology, while the

Monitoring and Configuration entity operation

depends on the radio access technology.

Monitoring and Configuration entity provides

auxiliary functionality for handling new service area

conditions or management requests. The aim of this

entity is to provide insight on the status (offered load

and performance) of the underlying network, ensure

that the latter operates properly, and perform the

necessary configuration actions to the managed

network segments. These actions are achieved by

using the Simple Network Management Protocol

(SNMP). Apart from communicating monitoring

related information to the Resource Brokerage

entity, the Monitoring and Configuration entity also

processes the rough network parameters and

compares them with corresponding thresholds. In

case some thresholds are exceeded for a number of

sequential updates, the entity is responsible for

triggering a redistribution request to Resource

Brokerage.

ICETE 2004 - WIRELESS COMMUNICATION SYSTEMS AND NETWORKS

3.3 Operation Sample

 WLAN network status acquisition phase

takes place (step 3). The status of the

network (e.g., traffic carried by cells) in the

affected service area regions is obtained. It

is noted that this information may already be

available through monitoring.

The sample operation of the network and service

management system, provided in this sub-section

(depicted in Figure 3), presents the interactions

between the various components of NSMS and the

terminals. It should be mentioned that the scenario is

initiated from the WLAN network, but could be

similar if it was initiated from the GPRS or the

DVB-T network. The following steps are identified:

 Offer exchange phase takes place (step 4).

The WLAN Network Manager requests for

offers from the co-operating (GPRS and

DVB-T) networks. These offers should

contain cost and capacity information.

 The Monitoring and Configuration entity of

the WLAN NSMS identifies a new

environment condition (e.g., degradation of

service quality or increased traffic load),

which may require redistribution of the

traffic load to the radio systems (step 1). It

triggers (Alarm Request) the Resource

Brokerage entity functionality (of the same

NSMS). The latter forwards this request to

the Session Manager, in order to obtain the

current status of the users (per service and

user class) in the affected service area.

 Based on the WLAN network status, and the

offers provided by the GPRS and DVB-T

networks, the Service Management entity of

the WLAN NSMS decides on the

assignment of services to quality levels, and

of traffic to networks (step 5).

 Acceptance phase takes place (step 6).

During this phase, the three co-operating

Network Managers are accepting the

solution proposed by the WLAN Service

Management entity, in Step 5.

 The Session Manager and the terminals are

notified about the decision of the WLAN

Network Manager (step 7).

 The Session Manager collects the

aforementioned information (step 2) and

triggers the WLAN Network Manager

functionality (Distribution Request).

 Redistribution phase takes place (step 8).

GPRS Network

Manager

GPRS Network

Manager

DVB Network

Manager

DVB Network

Manager

Session

Manager

(W)

Session

Manager

(W)

Resource

Brokerage

(W)

Resource

Brokerage

(W)

Service

Management

(W)

Service

Management

(W)

Monitoring &

Configuration

(W)

Monitoring &

Configuration

(W)

WLAN Network Manager

Distribution Request

TSMS

Alarm Request

4.Offer exchange phase

3.Network status acquisition phase (monitoring) and

load evaluation

2.User related

information collection

1.Identification of new

environment condition

5.Assignment of services

to quality levels and of

traffic to networks

6.Acceptance phase

7.Session Manager Notification phase

8.Distribution phase

GPRS Network

Manager

GPRS Network

Manager

DVB Network

Manager

DVB Network

Manager

Session

Manager

(W)

Session

Manager

(W)

Resource

Brokerage

(W)

Resource

Brokerage

(W)

Service

Management

(W)

Service

Management

(W)

Monitoring &

Configuration

(W)

Monitoring &

Configuration

(W)

WLAN Network Manager

Distribution Request

TSMS

Alarm Request

4.Offer exchange phase

3.Network status acquisition phase (monitoring) and

load evaluation

2.User related

information collection

1.Identification of new

environment condition

5.Assignment of services

to quality levels and of

traffic to networks

6.Acceptance phase

7.Session Manager Notification phase

8.Distribution phase

Figure 3: Management system sample operation

A DISTRIBUTED SYSTEM FOR THE INTEGRATED MANAGEMENT OF HETEROGENEOUS WIRELESS

NETWORKS

4 TERMINAL FUNCTIONALITY

The Terminal Station Management System (TSMS)

resides in the user terminal and controls its operation

within the CREDO system (Catalina, 2003). It is

necessary for the exploitation of the benefits offered

by the composite radio environment. TSMS is

responsible for the following tasks:

 It receives service start and stop requests

from the CREDO applications. In this way it

can keep track of all the currently running

applications. The communication with the

applications is based on message exchange

between the TSMS and the server through a

specific interface.

 It monitors the terminal status: TCP/IP

status, network interface status, application

status, etc. Concerning the status of the

network interface, the TSMS monitors IP

and link layer parameters, related to each

radio technology.

 It reports all the gathered information to the

Session Manager on the NSMS.

 Together with NSMS it selects the best

access network to use at each moment.

 It manages the terminal network

configuration. It configures the network

drivers and the TCP/IP stack according to

the decisions it takes.

Additionally it has a user interface, which allows

the configuration of the TSMS. The user can select

his/her preferences and see the current status of the

terminal and the network, as reported by the terminal

monitoring system and the TSMS.

Finally, there is a module responsible for the

communication between the terminal and the NSMS,

through a specific protocol, implemented for this

purpose. The TSMS – NSMS interactions, governed

by this protocol, include the following messages:

 Service Contract Information Request and

Reply. These messages are used once, at

start-up, in order to specify the set of

services to which a user is registered.

 Service Request and Reply. Through these

messages, the terminal reports to the NSMS

its current status (serving network, available

networks, services used, request for a new

service, etc) and the NSMS indicates by its

response the list of the preferred networks,

towards guiding the terminal in network

selection. The messages are sent

periodically (acting also as keep – alive

probes), but also whenever a change in the

current terminal status occurs (either in the

network availability or in the services used).

 Quality Report Request and Reply. The

terminal uses the request message in order to

report to the NSMS quality degradation

observed at the utilized services (e.g. a

major traffic load alteration sensed). The

NSMS after processing all the relative data,

instructs the terminal which is the best

action suggested in this case, by sending the

reply message.

 Handover Required Notification. This

message is sent by the NSMS and forces the

terminal to switch to another network. A

handover indication could also be included

in the service reply message, but this is sent

only after the service request from the

terminal. The handover required notification

does not require any trigger from the

terminal and covers cases where the

handover is necessary, without waiting for

the next service request.

5 EXPERIMENTS AND RESULTS

In order to evaluate the benefits gained by the

composite radio concept, several experiments and

performance measurements took place under the

framework of the project.

5.1 Test Environment Description

The overall platform used, consists of the relative

infrastructure components, described in section 2.

More specifically, the access networks comprise

one GPRS Base Transceiver Station (BTS)

connected directly to a commercial GSM/GPRS

provider, two IEEE 802.11b access points (APs)

jointly forming a single ESS (Extended Service Set)

and one IP/DVB-T multiplexer feeding a DVB-T

modulator. Concerning the GPRS network, the CS-2

coding scheme is used and up to four non-dedicated

time slots are used for packet switched traffic (the

rest is only voice traffic). As for the IP traffic over

DVB-T, a separate PID (Packet Identifier) has been

allocated. The local NSMS entities are Windows

PCs located at the corresponding subnets, while the

application server (also a Windows PC) is at the

home agent’s subnet, in order to avoid additional

delays. The multimode terminal used for the

experiments is a desktop Linux PC, equipped with

an IEEE 802.11b access card, a DVB-T receiver

card and interconnected to a GSM/GPRS phone

through the serial interface. The terminal possesses

the full TSMS functionality described in section 4.

Moreover, software applications have been

developed for simulation of the terminal behaviour;

ICETE 2004 - WIRELESS COMMUNICATION SYSTEMS AND NETWORKS

these applications have been used as virtual

terminals in order to evaluate the NSMS

functionality, because of the lack of equipment for

more real terminals. Finally, a traffic generator tool

has been used for simulating congestion situations. It

is a software engine that runs over Linux systems

and produces UDP packets creating background

traffic in a specific radio segment (Loukatos, 2002).

During the experiments, three types of

applications have been tested and provided in two

different quality levels (high – low):

 Video Streaming Service, for retrieving

streamed MPEG-4 encoded content from the

application server. This service is accessible

only through DVB-T and WLAN due to

high bandwidth requirements (nominal bit

rate 512 Kbps for the high quality level and

128 Kbps for the low one).

 Sports Event browsing, for acquiring

information for the Olympic games from the

application server. The bandwidth needed is

64 Kbps for the high quality level (not

through GPRS) and 32 Kbps for the low one

(served also by GPRS).

 Generic Internet Service Provision (GISP),

for web browsing. This service is not

decisive for the experiments due to its low

demand in bandwidth (32 Kbps and 8 Kbps

for each quality level). It is supported by all

the examined radio access technologies.

Two typical scenarios are selected from the

experiments and presented in the following

subsections. Both scenarios are related to handling

of congestion situations. Their difference lies on the

results view aspect. Scenario 1 presents results of a

more user-centric approach, while scenario 2 of a

network-centric one.

5.2 Scenario 1

This scenario demonstrates the benefits deriving

from the diverse radio networks interworking during

high traffic situations (inside hot-spots regions).

More specifically, when a stand-alone radio network

cannot confront aggregate IP-based services requests

from users inside a specific area, it is more

preferable to divert the exceeding traffic to an

affiliated segment rather than to reject it or

downgrade the QoS levels provided. During the

experiments and validation of the composite radio

framework, many scenarios of this type where

tested. The most representative scenario refers to the

case where the terminal accessing the video

streaming service through the WLAN network faces

a congestion situation (simulated by a 4.4 Mbps

UDP stream injected to the WLAN by the traffic

generator tool). Consequently, the quality of the

videos begins to downgrade as the aggregate traffic

exceeds the alarm threshold (4.8 Mbps). At the same

time, the NSMS located in the WLAN segment

triggers the optimization process for solving the

problem. After the completion of this process, the

terminal is forced to handover to the DVB-T

network with return path GPRS. During the

handover procedure, the video continues without any

interruption and after its completion it regains the

best quality.

Figure 4 depicts, for the whole observation time,

the traffic monitored in the WLAN and DVB-T

segments as well as the traffic generator output and

the alarm threshold. It must be noticed that the total

response time of the platform against the congestion

situation is app. 30 seconds from the point that

traffic excesses the threshold value. The initiation of

the optimization algorithm is deliberately delayed

for 15 seconds, avoiding this way false alarm events

triggered by short peak values of the aggregate

traffic. The time interval and the alarm threshold are

both configurable parameters.

0 50 100 150 200

Time (sec)

Traffic (Mbps)

WLAN AP

monitored Traffic

DVB-T Traffic

Traffic Generator

Output (UDP)

Alarm Threshold

Handover to DVB-T/GPRS

Start of Optimization Process

Figure 4: Traffic streams over time

Figure 5 displays the video’s frame rate during

the previous procedure.

0 50 100 150 200

Time (sec)

Frame Rate (fps)

Handover to DVB-T/GPRS

Start of Congestion Situation

Figure 5: Frame Rate of the video stream during the

congestion situation

A DISTRIBUTED SYSTEM FOR THE INTEGRATED MANAGEMENT OF HETEROGENEOUS WIRELESS

NETWORKS

Figure 6 demonstrates the interactions between

the various components as the terminal switches

from the WLAN to the DVB-T network. As depicted

in the figure, the following steps are visible:

 After the completion of the mid term

optimization algorithm, the NSMS sends the

handover notification message to the

terminal through the Home Agent (HA) and

the WLAN Foreign Agent (FA), because

WLAN is at this moment the serving

network.

 The terminal sends a service request through

the WLAN network. The WLAN FA

delivers this message to the Home Agent

(HA) that forwards it to the NSMS.

 The NSMS-response, suggesting of

handover to the DVB-T network with return

channel GPRS, follows the reverse way and

reaches the terminal.

 The terminal sends a MIP registration

request to the HA (for switching to the

DVB-T network) through the GRPS NAT

(return channel) and receives the reply

through the same entity. After receiving the

reply, the terminal has finished all the

required actions in order to change the

serving network from WLAN to DVB-T.

 The terminal sends a new service request

through the GPRS network, which is

delivered from the GRPS NAT to the HA

and then to the NSMS. The NSMS response

is sent through the HA and the DVB-T FA.

Figure 6: WLAN to DVB-T/GPRS handover procedure

5.3 Scenario 2

For the purposes of this scenario more than one

terminal were essential, so a set of virtual terminals

has been used. Furthermore, two user classes have

been assumed: the Gold and the Economy user class.

A user can choose to subscribe to different user

classes for different services. Users of the Gold class

are provided the service at the corresponding high

quality level, whereas users of the Economy class

can be provided the service at either of the two

reference quality levels.

Table 1 depicts the distribution of concurrent

users in the considered service area in the WLAN

network. Background traffic was simulated in the

WLAN network so as to create, over a period of

time, a hot spot.

Table 1: Demand volume in the WLAN network

Service

Generic

Internet

Sports

Event

Streaming

Video

QoS Levels

Description

High Low High Low High Low

Gold 2 0 4 0 1 0

User

Class

Economy 0 0 10 0 2 0

The normal condition of the service area is

configured to allow a cell load of 45% to 55% for

the cell (cells) covering the particular service area. A

snap shot of the WLAN cell under normal

conditions, serving the demand volume presented in

the previous, is depicted (Figure 7). The cell load

has reached 52% of its overall capacity. The

simulated traffic gradually increases.

Video Streaming

26%

Sports Event

15%

Simulated Traffic

10%

Generic Internet

Figure 7: Traffic distribution: normal condition

Under loaded service area conditions the

simulated traffic has reached 30% of the cells

capacity. In other words, the cell is now

approximately 72% loaded (Figure 8).

There are two typical solutions that can be

proposed. (i) To maintain all the users at the WLAN

network. Actually, this will result to the degradation

of the quality offered (Figure 9). In fact this is the

solution that would be imposed without the

composite radio concept and the exploitation of the

NSMS functionality. (ii) To maintain all users, of

both user classes, at the high quality level by

exploiting the GPRS and DVB-T networks. In this

ICETE 2004 - WIRELESS COMMUNICATION SYSTEMS AND NETWORKS

respect, a subset of the users (specified by the mid-

term optimisation algorithm) will obtain the

corresponding service through the WLAN network,

and the remaining users will be instructed to obtain

the service either through GPRS or DVB-T.

Simulated traffic

30%

Generic Internet

Sports Event

15%

Video Streaming

26%

Load has increased

due to simulated

traffic (20% increase)

Figure 8: Traffic distribution: loaded condition

Generic

Internet

orts Event

Degradation

Video

Streaming

Degradation

Video

Streaming

Simulated

Traffic

30%

Sports Event

Figure 9: Degradation of Economy user class

As depicted in Figure 10, exploitation of the

networks of GPRS and DVB-T results in part of the

terminals being instructed to switch to these

networks. The Session Manager functionality and

the overall NSMS operation assist in avoiding

congestion while continuing to provide all users with

the highest possible quality level.

Video

Streaming

Sports Event

Simulated

traffic

30%

Generic

Internet

Video

Streaming

17%

Part of the load

transferred to GPRS

Part of the load

transferred to DVB-T

Part of the load

transferred to DVB-T

Figure 10: Exploitation of GPRS and DVB-T networks

6 CONCLUSIONS

This paper tries to validate the benefits of the

composite radio concept by demonstrating results

from the prototype architecture of the IST project

CREDO. It addresses the profits gained for the

network operator (resolution of congestion

situations) and also for the user (usage of demanding

services with the best quality level). Simulations

could estimate the efficiency of this architecture in a

more realistic environment (larger areas to cover,

many real terminals to serve) (Kontovasilis, 2003).

The structure of the architecture is flexible enough

to be adapted for handling of larger scale situations

and encompass other radio network types with

minimum effort, because of its distributed form and

technology independent optimization algorithm. The

study of such extensions and the feasibility of

producing more practical terminals are issues for

future research.

REFERENCES

Varshney, U., “Recent advances in wireless networking”,

IEEE Computer, Vol. 33, No. 6, June 2000.

3rd Generation Partnership Project (3GPP) Web Site,

www.3gpp.org, Jan. 2002.

Varshney, U., Vetter, R., “Emerging mobile and

broadband wireless networks”, Commun. of the ACM,

Vol. 43, No. 6, June 2000.

Digital Video Broadcasting (DVB) Web site,

www.dvb.org, Jan. 2002.

IST project CREDO (Composite radio for enhanced

service delivery during the Olympics) Web site,

www.ist-credo.org, Feb. 2001.

Catalina, M., Stathopoulos, P., 2003. Terminal

Management System for Optimized Service Delivery

in Composite Radio Environments. In IST Mobile

&Wireless Telecommunications Summit.

Loukatos, D., 2002. Traffic Generation and Analysis,

Emphasizing on ATM and IP Network Technologies.

Ph.D. Thesis. National Technical University of

Athens, Athens.

Kontovasilis, K., Skianis, C., Kormentzas, G., 2003.

Estimating Signalling Efficiency in a Composite

Radio Environment. In 1st International Working

Conference on Performance Modelling and

Evaluation of Heterogeneous Networks (HET-

NETs’03).

A DISTRIBUTED SYSTEM FOR THE INTEGRATED MANAGEMENT OF HETEROGENEOUS WIRELESS

NETWORKS