George Kormentzas, Ilias Maglogiannis*, Dimitris Vassis ... · proposed framework for designing a Regional Area Network (RAN) serving seven actual conﬁdential applications. Finally,

A modelling and simulation framework for compoundmedical applications in regional healthcare networks

George Kormentzas, Ilias Maglogiannis*,Dimitris Vassis, Dimitris Vergados, Angelos RouskasDepartment of Information and Communication Systems Engineering,University of the Aegean, GR-83200, Karlovassi, GreeceFax: +30-2273082009 E-mail: [email protected]: [email protected] E-mail: [email protected]: [email protected] E-mail: [email protected]*Corresponding author

Abstract: Regional healthcare information networks have already started togrow across Europe, in order to cover local healthcare provision needs,especially in isolated regions, where there is often no availability of centralgeneral hospitals. The paper discusses a modelling and simulation frameworkfor the design of regional healthcare information networks running compoundmedical and QoS – sensitive applications. The proposed framework decomposesthe compound medical applications into combinations of elementary trafficprofiles, assesses appropriate values to the traffic parameters of the assignedmodels and defines suitable simulation scenarios. The simulation results areanalysed and finally lead to reliable bandwidth estimations of the links of thehealthcare information network under design. The proposed framework has beenthoroughly validated through its application for the design of a healthcarenetwork in the islands of the North Aegean Sea, running actual compoundmedical applications in the context of a national research project.

Keywords: integrated applications; modelling and simulation; network design;regional healthcare information networks.

Reference to this paper should be made as follows: Kormentzas, G.,Maglogiannis, I., Vassis, D., Vergados, D. and Rouskas, A. (2005) ‘A modellingand simulation framework for compound medical applications in regionalhealthcare networks’, Int. J. Electronic Healthcare, Vol. 1, No. 4, pp.427–441.

Biographical notes: George Kormentzas is currently Lecturer at the Universityof the Aegean, Department of Information and Communication SystemsEngineering. He received his Diploma in Electrical and Computer Engineeringand his PhD in Computer Science both from the National Technical Universityof Athens (NTUA), Greece, in 1995 and 2000, respectively. His research interestsare in the fields of traffic analysis, network control, resource management andquality of service in broadband networks. He has over 30 publications in books,journals and international conference proceedings in the above areas. He is amember of the Technical Chamber of Greece.

Dr Ilias Maglogiannis received his Diploma in Electrical and ComputerEngineering and his PhD in Biomedical Engineering from the NationalTechnical University of Athens (NTUA) Greece in 1996 and 2000 respectively.From 1996 until 2000 he worked as a researcher in the Biomedical Engineering

111234567891011123456789201112345678930123456789401234567118

Int. J. Electronic Health, Vol. 1, No. 4, 2005 427

Copyright © 2005 Inderscience Enterprises Ltd.

IJEH-7-Kormentzas correx 3/23/05 5:19 PM Page 427

Laboratory in NTUA. Since February 2001 he has been a lecturer in theDepartment of Information and Communication Systems Engineering at theUniversity of the Aegean. His published scientific work includes four lecturenotes (in Greek) on biomedical engineering and multimedia communicationstopics, 15 journal articles and more than 30 conference papers.

Dimitrios Vassis received his Diploma in Electrical and Computing Engineeringfrom NTUA in 2001. He is a PhD student in the Department of Information andCommunications Systems Engineering at the University of the Aegean. Since2002, he has been a member of the Technical Chamber of Greece.

Dimitrios Vergados received his BSc in Physics from the University of Ioanninaand his PhD in Integrated Communication Networks from the NationalTechnical University of Athens, Department of Electrical Engineering andComputer Science. His research interests are in the area of communicationnetworks, wireless networks, heterogeneous communications networks, militarynetworks, adhoc networks, computer vision systems and high performancecomputing. He has participated in several EC projects and has had severalpublications in journals, books and conference proceedings. He is Guest Editoron the IEEE Network Journal and Reviewer of several journals. DimitriosVergados is a member of the IEEE and AFCEA.

Dr Angelos Rouskas received his five-year Diploma in Electrical Engineeringfrom the National Technical University of Athens (NTUA), his MSc inCommunications and Signal Processing from Imperial College, London, and hisPhD in Electrical and Computer Engineering from NTUA. He is AssistantProfessor in the Department of Information and Communication SystemsEngineering of the University of the Aegean (UoA), Greece, and AssociateDirector of the Computer and Communication Systems Laboratory. His currentresearch interests are in the areas of resource management of mobilecommunication networks, mobile and adhoc networks security, and pricing andcongestion control in wireless and mobile networks and he has had severalpublications in the above areas.

1 Introduction

During recent years, healthcare providers are continuously focusing on the prevention and early detection of disease and the enhancement of primary and emergency care(Demiris, 2004). Furthermore, computer and network based systems are expanding inorder to support more healthcare activities. In such a dynamic environment, informationand communication technologies (ICT) are starting to play a significant role in thepractice of healthcare at all levels (Kumar, 2004; Li and Rubin, 2004; Wickramasingheand Goldberg, 2004). At a regional level the provision of health telematic services to thelocal population requires the existence of regional healthcare information networks(OHIHHC; Pouloudi, 1999).

111234567891011123456789201112345678930123456789401234567118

428 G. Kormentzas, I. Maglogiannis, D. Vassis, D. Vergados and A. Rouskas


More specifically, regional healthcare information networks have already started togrow across Europe, in order to cover local healthcare provision needs, especially inisolated regions (i.e. isolated areas in Greece, in Scandinavian countries and in Germany),where there is often no availability of central general hospitals. The regional networks can fill this gap, connecting local healthcare service providers with central regional and/orperipheral hospitals and thus making possible tele-consultation, tele-diagnosis andexchange of views between remote located doctors in certain patient treatment cases(Professionals and Citizens Network website; Tsiknakis et al., 2002; US DHHS website).The type of information exchanged in such healthcare networks, corresponds to sensitivepersonal real-time data; therefore Quality of Service (QoS) issues during their design mayarise (Mundy and Chadwick, 2004).

One of the most critical issues of efficient and Quality of Service (QoS) assurednetwork design is the correct estimation of the required bandwidth. In this context, it isessential for the network designer to know the traffic characteristics of the applications,which are planned to run over the network under design. The precise prediction of theproduced traffic load can lead to the establishment of networking links of appropriatecapacity, which can guarantee the requested QoS. However, in many cases, the exacttypes of applications, as well as their detailed traffic parameters, are unknown to thenetwork designer. The emergence of new applications after the deployment of a networkand the confidential nature of some special-purpose applications (e.g. healthcare) are themain reasons for this fact.

The design of healthcare networks, which can guarantee QoS for special-purposeconfidential applications, seems to constitute an emerging grand challenge for networkdesigners and planners due to its complexity (arising from the special healthcareinformation networks characteristics) and its significant economical and social impact (foran inappropriately designed regional healthcare network, significant investment is neededfor its redeployment). The adaptation of protocols, such as MPLS (Swallow, 1999), RSVP (Braden et al., 1997), etc., which are able to bind network resources for specificapplications/services and thus are able to provide QoS guarantees, offer a straightforwardsolution. However, this approach does not concern the totality of special healthcarenetworks design requirements. Requirements such as graceful degradation of QoS inoverloaded conditions, rapid design and deployment/modification of new services,support of multimedia/multiparty applications, robustness to failures and changingenvironment, high performance, scalability and security. The usage of management andcontrol architectural frameworks supporting QoS as well as simulation tools that canpredict the performance and behaviour of healthcare information networks operating inrealistic environments come to the fore.

In this emerging context, the paper discusses a modelling and simulation frameworkfor the design of regional healthcare information networks running compound medicalQoS-sensitive applications. Traffic characterisation of such applications constitutes thecore concept of the proposed framework for estimating the appropriate capacity of thelinks of a regional healthcare information network. Compound medical confidentialapplications are decomposed into elementary applications modelled through well-knowntraffic profiles. Using these elementary profiles and applying various simulation scenariosin an advanced platform (e.g. OPNET), a healthcare network designer can estimate therequired resources for the demanded QoS. The proposed framework has been validated

111234567891011123456789201112345678930123456789401234567118

A modelling and simulation framework for compound medical applications 429


through its application for the design of a large scale regional healthcare informationnetwork, for remote islands in the Aegean Sea, running actual compound medical dataapplications in the context of a national research project funded by the North AegeanRegional Healthcare Organisation.

The rest of the paper is organised as follows: Section 2 discusses the proposedmodelling and simulation framework. Section 3 gives an example application of theproposed framework for designing a Regional Area Network (RAN) serving seven actualconfidential applications. Finally, Section 4 concludes the paper discussing some openissues and directions for future work.

2 The proposed modelling and simulation framework

The proposed framework can be used for the design of regional healthcare informationnetworks running compound QoS-sensitive and security demanding applications. Itincludes the following generic steps:

� From the health provider user side: Rough description of the requested compoundconfidential applications.

� From the network designer side: Decomposition of the requested compoundconfidential applications into elementary traffic profiles.

� Assessment of the appropriate traffic parameters for the profiles being the outcomeof the previous step.

� Definition and application of suitable simulation scenarios in an advanced platform.

The outcome of the previous steps is the bandwidth estimation of the links of the regionalhealthcare network under design.

The decomposition of the compound applications into elementary traffic profiles isbased on an iterative question-based mechanism. The more information is available abouta compound application (from the user side), the more questions can be answered and thusthe more precise the definition of the underlying elementary traffic profiles can be. Theframework’s decomposition questions are:

� What is the type of the information used by a compound application? The most familiar types of information are voice, video and medical data. If theinformation of an application is more complicated, this application has to bemodelled by a composition of elementary traffic profiles producing single types of information.

� Is there a central storage point? Can the information produced from the parts that use the application be distributed among them or should it be stored in a central base point (i.e. a central hospital) and reused when needed? In the first case, the applications should be modelled by a client–client traffic model. In thesecond case, there must be an association with a client–server model.

111234567891011123456789201112345678930123456789401234567118



� Is the application synchronous or asynchronous? Synchronous applications requirean open session and the simultaneous participation of both parts. Asynchronousapplications require an intermediate base station where the information producedfrom one part can be temporarily stored in order to be delivered to the other part.

� Is the application characterised as Constant Bit Rate (CBR) or Variable Bit Rate (VBR)? For CBR applications, the binding of a standard amount of capacity is sufficient for satisfying their QoS needs. VBR applications are characterised by ON periods where there is transmission of information and OFF periods where there is no activity. The duration of ON and OFF periods can be constant or random. In most cases the duration varies in time and the network designer must consider it as a random variable and figure out its distribution.

� Is the application real time or non-real time? Real time applications require exactprediction of bandwidth because of stringent bounds on packet delay. Non-real timeapplications, such us Store and Forward Telemedicine applications, are moretolerant in delay, thus, the network designer can make a more conservativeprediction of the required bandwidth.

� Are the application’s traffic parameters constant or time dependent? Timedependent parameters are usually represented as random variables, which obey a probability distribution. It is significant for the correct modelling of theapplication to define the most appropriate probability distribution for these traffic parameters.

The above questions are organised in the flow chart depicted in Figure 1. The terminating(or terminal) cycles of the chart represent the elementary traffic profiles in which acompound confidential application can be decomposed. These profiles are IP phone calls,video traffic, image and medical data traffic, database transactions, and e-mail messages(Bertsimas and Freud, 2000). Each one of them can be accurately described by a certainnumber of traffic parameters.

2.1 IP phone calls

An IP phone call service is a real time delay sensitive service. According to Douskalis(2000) the required bandwidth for this service is:

11B=R+— (Type 1),k

where B is the required bandwidth, R is the coding bit rate and k is the number of frames in each IP packet. Knowing that during a phone call there are periods of silence, the information transmission can be characterised by ON periods where active speech is transmitted with a constant bit rate and OFF periods where speech issuspended.

111234567891011123456789201112345678930123456789401234567118



2.2 Video traffic (Heyman and Lakshman, 1996)

The traffic characteristics of a video service are similar to an IP telephony service (i.e. interactive service producing real time traffic with strict bounds in delay). However,as video consists of both voice and image, a video service demands more bandwidth.Various encoding algorithms (such as various types of MPEG (1998)) can reduce therequired bandwidth. Video traffic is produced mostly by live tele-consultation sessions inmedical emergencies.

For the adopted simplified video traffic profile, video is considered to consist only ofmoving images (other simple traffic profiles such as IP phone calls are used for modellingvoice). In this context, the elementary video traffic profile is consisted of images (calledframes) that succeed each other with a specific rate (frame rate). The size of each framedepends on its horizontal and vertical dimensions (in pixels) and on the pixel’s colouranalysis (measured in bits). The bit rate for transmitting video traffic is calculated fromthe type below:

B=S�R (Type2),

where B is the required bandwidth, s is the frame size and r is the frame rate. Differentcombinations of values of the above parameters can produce the same required bit rate.The specific values of parameters depend on the type and the characteristics of themodelled compound confidential application.

111234567891011123456789201112345678930123456789401234567118


Figure 1 Compound application decomposition mechanism


For CBR video traffic, the frames are transmitted continuously with the same bit rate,while for VBR video traffic there are ON periods where video traffic is produced and OFFperiods where no traffic is produced. Both time periods (ON and OFF) can be constant orthey can vary on time. The compound classified application under modelling defines both thetype of video (CBR or VBR) and the pattern of ON and OFF time periods (for VBR video).

2.3 Medical image and biosignals traffic

Such traffic is served by synchronous protocols running over TCP/IP, which can be usedfor sending and receiving files from a remote computer in a client-server model. When asession is started, the client can transfer medical data from the server (download) or tothe server (upload). The session is terminated when all the requested data are downloadedor uploaded. There is an obvious difference in downloading and uploading operationsconcerning the bandwidth in each direction. Downloading requires more bandwidth in the server–client direction while uploading requires more bandwidth in the inverse (i.e. client–server) direction.

2.4 Electronic health record transactions (Siman, 2000)

These are mostly database transactions (Ram et al., 1999; Gray and Reuter, 1993), whereusers can either retrieve data from the database making queries, or can add newinformation to the database making updates. The whole operation of storing and sortingthe information in the database, as well as the establishment of communication among theusers and the database is achieved by a special-purpose software platform, i.e. a DatabaseManagement System (DBMS). Electronic Health Records and DBMS are stored in adatabase server. The communication between database clients (i.e. users) and server (as well as the corresponding traffic produced by this communication) follows theclient–server model. Note the difference between queries and updates. Queries demandmore bandwidth in the server–client direction, while updates require more bandwidth inthe client–server direction.

Three are the traffic parameters that can define the amount of traffic produced by thedatabase transactions profile. The size of the database records the number of queries andthe number of updates in a time period.

2.5 E-mail messages (IETF, 1982, 1995)

E-mail communication is an asynchronous service based on the Simple Mail TransferProtocol (SMTP) running over the UDP protocol. The whole operation is synchronised byan e-mail server. Each user has an amount of disk space in the server where his received e-mails are stored. When a user wants to check his e-mails, a session between the client(i.e., the user) and the mail server is established and the information is downloaded fromthe server to the client. When a user wants to send an e-mail, the information is uploadedto the mail server. The mail server is responsible to deliver the e-mail to the mail server ofthe receiver.

The amount of traffic produced in a link from the e-mail messages’ traffic profiledepends on the total size of the e-mails to be sent and/or received, as well as the numberof downloads and uploads in a time period.

111234567891011123456789201112345678930123456789401234567118



According to the proposed modelling and simulation framework, the decompositionof compound applications into combinations of the above elementary traffic profiles and the assessment of the parameters of the defined profiles is complemented by asimulation process. The values of the traffic profiles’ parameters define a set of simulationscenarios. For the most reliable estimation of the bandwidth of the health informationnetworks links, these scenarios have to represent both regular and high traffic loadconditions. The simulation results combined with the requested QoS guarantees can giveconservative or less conservative (but satisfactory) final bandwidth estimations in aheuristic way.

3 An application paradigm of the proposed framework

This section discusses an application of the proposed framework for the design of a RANbranch of a healthcare information network (see Figure 2).

Note that the purpose of this application paradigm is just to highlight the framework’sgeneric steps and not to prove its effectiveness. The framework has been thoroughly testedthrough its application for the design of a large scale healthcare information networkrunning actual compound medical applications for remote islands in the Aegean Sea, inthe context of a national research project. Coming to the presented example, the RANbranch includes seven workstations running different actual medical secure applicationsand a gateway, which concentrates the traffic from all workstations and conveys it to thecore healthcare information network. The basic design goal is the bandwidth estimationof the link connecting the RAN branch with the core network. Specifically, the giveninformation for the applications (numbered by 1 to 7) is the following:

Application 1: A confidential PCM (Pulse Code Modulation) telephony service.

Application 2: A patient telemonitoring application, running 24 hours per day in real timemode. The application both collects biosignals from patients under constant surveillanceand sends them to other locations where medical personnel exists and receives thecorresponding information from peer workstations. The given transmission rate is 9.6 Kb/s.

111234567891011123456789201112345678930123456789401234567118


Figure 2 An example design paradigm based on the proposed framework


Application 3: A compound application with three components:

� Database information (electronic health records of unknown size) retrieval. Thegiven transactions per year are approximately 50,000.

� Tele-conference service with required bit rate of 64 Kb/s and duration between 3 to10 min.

� Heavy-sized information delivery via e-mail (no information about the size and thefrequency of the e-mails is given).

Application 4: Telephone service carrying confidential information sampled at 22.8 Kb/s.

Application 5: Electronic Health Record information retrieval from a central database.The database records have an unknown constant size and the given transactions per yearare approximately 70,000.

Application 6: This compound application differs from Application 3 regarding thesecond component, where the tele-conference service is replaced by an e-mail service.

Application 7: A patient telemonitoring application in batch mode, which monitors threepatients and sends three status reports with their corresponding biosignals in a central basestation every hour. Each status report is a 100 Kb plain text file.

The above description constitutes the first phase of the proposed modelling and simulationframework. Based on this information, we proceed to decompose the confidentialapplications into simple traffic profiles with appropriate parameters. The values of theseparameters can lead to different simulation scenarios. In this paradigm, two simulationscenarios are defined. The first one refers to normal traffic load, while the second concernshigh traffic load.

For the rest of the paper, the notation �(�) parameter_name denotes that the parameter parameter_name has an exponential distribution with mean value �. Similarly,the notation �(�) parameter_name denotes that the parameter parameter_name is aPoisson process with mean rate �.

The decomposition of the given applications into elementary well-defined trafficprofiles for the discussed paradigm is as follows:

Application 1: This application falls by its definition into one of the elementary trafficprofiles defined in the previous section (i.e. IP phone calls). Given that the typical codingrate of a PCM telephony service is 64 Kb/s and each packet is consisted of one frame, themaximum required bandwidth (according to Type 1) is 75 Kb/s. For the frequency and theduration of the phone calls, we assume that in the first scenario there are �(4) phone callsper hour and each one lasts �(3) minutes, while in the second scenario there are �(12)phone calls per hour and each one lasts �(5) minutes. Using the ON/OFF model, we assumethat, in both simulation scenarios, there is an �(3) seconds length of silence and talk spurt.

Application 2: The produced information is continuously transmitted, meaning (accordingto the discussed decomposition framework) that the application has to be modelled by aCBR traffic profile. The requested bit rate (i.e. 9.6 Kb/s) is succeeded through theselection of the appropriate parameters values. According to Type 2, a frame size of 960bits and a frame rate of ten frames/s can give the required bit rate.

111234567891011123456789201112345678930123456789401234567118



Application 3: The three application components are modelled separately. The electronic health records transactions profile can obviously model the first component of application 3. As there is no information about the transactions size, we suppose that forthe first simulation scenario a typical transaction size of �(200) Kb, while for the second scenario a typical transaction size of �(500) Kb. The frequency of transactions (singledatabase queries) for the first scenario is �(50,000) transactions per year, while for the second is �(150,000) transactions per year. The second application component ismodelled by a VBR video traffic profile. The rate in ON periods is 64 Kb/s (the video-conference service is active). In the first scenario, we assume that each conferencelasts �(5) min and there are �(4) conferences per hour. In the second scenario, we assumethat there are �(3) conferences per hour that last �(10) min. Finally, the third applicationcomponent is modelled by the e-mail messages traffic profile, where for the first scenario,we suppose that there are �(6) e-mails of �(1) Mb per hour, while in the second, there are�(24) e-mails of �(1) Mb per hour.

Application 4: An IP phone calls traffic profile is suitable for this application. Giving thateach IP packet consists of one frame and the coding rate is 22.8 Kb/s, the total requiredbandwidth (according to Type 1) is 33.8 Kb/s. For the frequency and the duration of thephone calls, we make the same assumptions as in Application 1.

Application 5: The database transactions profile can obviously model this application.Building on the given application description, for the first simulation scenario it issupposed that the frequency of transactions (single database queries) is �(10,000) peryear, while for the second is �(20,000) per year. The transaction size can be either 700 Kb (first scenario), or 2 Mb (second scenario).

Application 6: This compound application substantially includes two components. Thefirst one (i.e. the database transactions) is modelled as the corresponding component ofApplication 3. The second component (i.e. the e-mails delivery) is modelled by the e-mailmessages elementary traffic profile, where for the first scenario, we suppose that there are�(12) e-mails of �(1) Mb every hour, while in the second, there are �(30) e-mails of �(1) Mb every hour.

Application 7: Regarding the patient monitoring information transmitted in batch mode toa central base station, for both simulation scenarios, it is assumed �(3) file uploads (size of 100 Kb) every hour.

Table 1 summarises the elementary traffic profiles, which can model the under studyapplications for the two simulation scenarios. These scenarios are applied in an OPNET simulation platform. For both scenarios, Figures 3–9 depict the traffic produced from Applications 1–7 correspondingly in the intermediate links betweenworkstations (running the corresponding applications) and the gateway. Figures 10 and11 give the total produced traffic in the link connecting the RAN branch with the core healthcare information networks for the normal and high load simulation scenariosrespectively.

111234567891011123456789201112345678930123456789401234567118



111234567891011123456789201112345678930123456789401234567118


Table 1 Elementary traffic profiles of applications 1–7

Application Elementary traffic Parameter Normal traffic High trafficprofile load values load values

1 IP phone calls Rate 64 Kb/s 64 Kb/sCalls per hour p(4) p(12)Call duration �(3) min �(5) minTalk spurt, length/ �(3) sec �(3) secsilence, length

2 Biosignals traffic Frame size 960 bits 960 bitsFrame rate 10 frames/s 10 frames/s

3 EHR transactions Trans/year p(50,000) p(150,000)Trans size �(200) KB �(500) KB

E–mail messages E–mails/h p(6) p(24)E–mail size �(1) MB �(1) MB

VBR video traffic Rate 64 Kb/s 64 Kb/sConfs/h p(4) p(3)Conf duration �(5) min �(10) min

4 IP phone calls Rate 33.8 Kb/s 33.8 Kb/sCalls per hour p(4) p(12)Call duration �(3) min �(5) minTalk spurt, length/ �(3) sec �(3) secsilence, length

5 EHR transactions Trans/year P(10,000) p(20,000)Trans size 700 KB 2 MB

6 EHR transactions Trans/year p(50,000) p(150,000)Trans size �(200) KB �(500) KB

E–mail messages E–mails/h p(12) p(30)E–mail size �(1 MB) �(1 MB)

7 FTP traffic Uploads/h p(3) p(3)File size 100 KB 100 KB

Figure 3 Traffic produced from Application 1 Figure 4 Traffic produced from Application 2


111234567891011123456789201112345678930123456789401234567118




Figure 9 Traffic produced from Application 7 Figure 10 Total traffic (normal load)


Some notes about Figures 3, 5 and 8. The random peaks of Figure 3 represent the phone calls. The high peaks of Figure 5 represent the produced traffic by e-mails, while the lowpeaks correspond to the traffic produced by database transactions. The absence of VBR videotraffic profile in Application 6 causes lower traffic bounds in Figure 8 than in Figure 3.

As already discussed in the previous section, in the last phase of the proposedmodelling and simulation framework, the network designer has to define the bandwidth ofthe links of the health network under design. For this paradigm, the network designer hasto estimate the bandwidth of the link between the RAN gateway and the core network.According to Figures 10 and 11, the total traffic load of this link is unstable with highvariation. The traffic in the second simulation scenario is almost double the traffic in thefirst scenario. It is evident that the bandwidth estimation can vary from a very conservativevalue (i.e. 840 Kb/s, which is the highest peak in the high load scenario) to a lessconservative but QoS acceptable value (i.e. 215 Kb/s, which is the average value of themean traffic load outcomes in both scenarios).

4 Conclusions

The paper discusses an emerging grand challenge concerning the design of regionalhealthcare information networks, which can guarantee QoS for special-purpose medicalapplications. We have presented a modelling and simulation framework for estimating theappropriate capacity of the links of a healthcare network, taking into consideration thespecial technical, social and economical characteristics of such networks. The basicbandwidth estimation goal is to assure the requested QoS, while also making efficient and economical usage of network resources, towards maximisation of network availabilityand minimisation of network operational costs. A simple application paradigm of theframework highlighted its basic concept and operational functionality. A large scaleapplication of the framework has been realised in the context of a national research projectfunded by the North Aegean Regional Healthcare Organization.

111234567891011123456789201112345678930123456789401234567118


Figure 11 Total traffic (high load)


The proposed modelling and simulation framework will be extended towards thefollowing three directions:

1 extra elementary traffic profiles will be added in order to give more flexibility in thecompound medical application decomposition mechanism

2 the incorporation of the proposed framework in a wider management and controlarchitectural framework for providing medical data QoS, will be examined

3 the heuristic approach towards the final bandwidth estimation will be replaced by amore efficient mechanism.

References

Bertsimas, A. and Freud, R. (2000) Data Models and Decisions, Prentice Hall.

Braden, R., Zhang,L., Berson, S., Herzog, S. and Jamin, S. (1997) ‘Resource ReSerVation Protocol(RSVP) – version 1 functional specification’, Internet Engineering Task Force: Request ForComments 2205.

Demiris, G. (2004) ‘Electronic home healthcare: concepts and challenges’, International Journal of Electronic Healthcare, Vol. 1, No. 1 pp.4–16.

Douskalis, B. (2000) IP Telephony – The Integration of Robust VoIP Services, Prentice Hall Inc.

Gray, J. and Reuter, A. (1993) Transaction Processing: Concepts and Techniques, MorganKaufmann.

Heyman, D. and Lakshman, T. (1996) ‘Source models for VBR broadcast–video traffic’,IEEE/ACM Transactions on Networking, Vol. 4, No. 1, pp.40–48.

IETF RFC 1651 (1995) SMTP Service Extensions.

IETF RFC 2250 (1998) ‘Implementations of MPEG video/RTP’.

IETF RFC 821 (1982) SMTP.

Kumar, S. (2004) ‘Mobile communications: global trends in the 21st century’, International Journalof Mobile Communications, Vol. 2, No. 1 pp.67–86.

Li, L. and Rubin, B. (2004) ‘Technology investment in hospital: an empirical study’, InternationalJournal of Management and Enterprise Development, Vol. 1, No. 4, pp.390–410.

Mundy, D. and Chadwick, D.W. (2004) ‘Electronic transmission of prescriptions: towards realisingthe dream’, International Journal of Electronic Healthcare, Vol. 1, No. 1, pp.112–125.

Office of Health and the Information Highway Health Canada (2003) ‘Canada health infoway:paths to better health’, Final Report of the Advisory Council on Health Infostructure(http://www.hc-sc.gc.ca/ohih-bsi/available/pathsvoies/info_e.html).

OPNET Simulation Tool (2005) http://www.opnet.com.

Pouloudi, A. (1999) ‘Information technology for collaborative advantage in healthcare revisited’,Information & Management, Vol. 35, pp.345–356.

Professionals and Citizens Network for Integrated Care (2003) (http://picnic.euspirit.org/).

Ram, P., Do, L. and Drew, P. (1999) ‘Distributed transactions in practice’, ACM SIGMOD, Vol. 28,No 3.

Report and Recommendations from the National Committee on Vital and Health Statistics (2001)‘Information for Health, Report and Recommendations from the National Committee on Vital and Health Statistics’, US Department of Health and Human Services(http://www.ncvhs.hhs.gov/nhiilayo.pdf).

Siman, A.J. (2000) ‘An agenda for the future: a national electronic health record system’, HCIM&C,Vol. XIV, No. 1.

111234567891011123456789201112345678930123456789401234567118



Swallow, G. (1999) ‘MPLS advantages for traffic engineering’, IEEE Communications Magazine,Vol. 37, No. 12, pp.54–62.

Tsiknakis, M., Katehakis, D. and Orphanoudakis, S. (2002) ‘An open, component-basedinformation infrastructure for integrated health information networks’, International Journalof Medical Informatics, Vol. 68, pp.3–26.

Wickramasinghe, N. and Goldberg, S. (2004) ‘How M=EC2 in healthcare’, International Journalof Mobile Communications, Vol. 2, No. 2, pp.140–156.

111234567891011123456789201112345678930123456789401234567118



Documents

George Kormentzas, Ilias Maglogiannis*, Dimitris Vassis ... · proposed framework for designing a Regional Area Network (RAN) serving seven actual conﬁdential applications. Finally,