James Won-Ki Hong Jong-Seo Kim
Dept. of Computer Science and Engineering
[email protected]
Jong-Tae Park
School of Electronic and Electrical Engineering
Kyungpook National University
[email protected]
This paper presents a CORBA-based Quality-of-Service (QoS) management framework fordistributed multimedia services and applications. The QoS MIB has been defined for the QoSmanagement of various multimedia services, and consists of information objects thatrepresent a set of layered QoS parameters. These managed objects are organized into fourlogical groups: service, application, system, and network. We have also defined a set of QoSmanagement services for monitoring and controlling QoS-related resources. Using thedesigned QoS MIB and management services, we have developed a QoS management systemfor managing and controlling QoS in a distributed multimedia system called MAESTRO. Theprototype management system is Web-based and uses OrbixWeb to interface with the QoSmanagement server which is implemented as a CORBA object, and monitors QoS parametersof distributed multimedia services, which are also implemented as CORBA objects. Keywords: Quality of Service, QoS MIB, QoS Management, CORBA, Multimedia Services
1. Introduction
The widespread use of distributed multimedia applications has created new challenges innetworking, including managing network resources for guaranteeing Quality-of-Service (QoS)[Cam93, Laz91, Woo94, Yam95]. As users become more familiar with multimedia services, QoSmust also be approached from the user's point of view, rather than only from the network-orientedview [Alp95, Cro95, Gad95, Hal95]. Users must be given the opportunity to express theirrequirements for the receiving service in terms of familiar QoS parameters. These parameters canbe, in turn, translated into parameters provided by the underlying distributed systems and networks[Jun93].
Distributed multimedia services and applications are time-critical and need managementsupport for ensuring agreed QoS [Geo94, Som95]. An important aspect of distributed multimediaservices is that they require QoS guarantees for the transfer and processing of continuous media data (such as video and audio). Emerging networks such as ATM [Min89] and the proposedintegrated services with reservations [She94] can provide guarantees on bandwidth and delay fordata transfer. Modern computer systems now have sufficient computing power and I/O bandwidthfor handling continuous media. Management in computing environments that support multimediaservices must promote QoS guarantees for each level of the system [Put95] because the overall QoSdepends on the combined QoS of the underlying distributed systems and networks [Hut94]. End-to-end management must include management capabilities for each layer participating in the service.
Researchers have recently proposed new communication architectures which are broader inscope and cover both network and end-system domains. These architectures differ in several ways,probably because different communities developed them. Among differences, QoS specification andQoS parameters are considered fundamental to capture user-level QoS requirements. Currently, nointernational standard or dominant specification for QoS parameters exists from the service layer tothe network layer. This need has motivated us to design and provide a set of standard QoSparameters, as well as QoS management services, which cover the layers from the service layer tothe network layer.
In this paper, we propose a CORBA-based QoS management framework for managing QoS ofdistributed multimedia services and applications. We have developed a set of QoS managementservices to monitor and control QoS parameters in distributed multimedia applications and theirsupporting services. These QoS management services have been defined using CORBA IDL andcan be used for quick and easy development of management applications. A generic QoS MIB hasbeen defined for the QoS management of various multimedia services. This generic QoS MIB canbe extended easily to develop MIBs for specific multimedia services, and used to integrate QoSmanagement into standard network management frameworks (such as SNMP and CMIP [Bla94]).
The QoS MIB also provides a set of layered QoS parameters and can be used as standard QoSparameters for managing QoS in multimedia services.
Earlier, we developed a CORBA-based distributed multimedia system called MAESTRO[Yun97] which supports the development and operation of distributed multimedia applications. Forvalidation of our QoS management framework, we have attempted to manage QoS in themultimedia services, which are part of MAESTRO, as well as applications running on MAESTRO.
In this paper, we also describe our effort on the prototype implementation of a Web-based QoSmanagement system for MAESTRO. The prototype QoS management system uses OrbixWeb[Ion96] to interface with the management server, which is implemented as a CORBA object andmonitors QoS parameters in distributed multimedia services which are also implemented asCORBA objects. Results show that the administrator can manage QoS of the MAESTRO systemwith a popular Web browser such as Netscape or Internet Explorer.
Section 2 of this paper describes QoS concepts and compares QoS parameters among severalpreviously proposed QoS architectures. Section 3 presents the QoS MIB design. Section 4 defines aset of QoS management services required for the QoS management of distributed multimediaservices and applications. Section 5 describes our prototype implementation of QoS management.
Section 6 summarizes our work and discusses possible future work.
2. QoS Concepts and Comparison
Work on QoS models and concepts has been done in various scenarios and within a number ofprojects. Most of them differ according to their assumptions, targets, and solution approaches.
QoS modeling includes: specifying the structures and types of QoS parameters, negotiation of QoS,mapping of QoS, resource reservation methods, methods for updating QoS parameters, and methodsfor monitoring and maintaining QoS.
QoS is defined as, “A set of qualities related to the collective behavior of one or more objects”in [Iso95]. The International Organization for Standardization (ISO) definition assumes an abstractmodel of objects, while the International Telecommunications Union – TelecommunicationsStandardization Sector (ITU-T), includes user's personal degree of satisfaction. Detailed QoSparameters specified in different documents of protocols vary as well.
Table 1 compares approaches according to their detailed QoS parameters specified and used.
Although numerous research work focuses on QoS, we have compared only three works (QoS-A[Cam93], Int-serv [Bra94, She95, Int95], and OSI95 [Iso95]). Only important QoS parametershave been listed to avoid a potentially large number of empty entries. Entries with “O” denote adefined parameter and entries with “X” denote an undefined parameter. “(X)” denotes that althoughan exactly matching parameter is not defined, the requested semantic can be satisfied by anotherparameter. In this paper, the QoS parameters required for managing QoS in multimedia services aredefined in the form of MIB. This can be used to integrate QoS management into standard networkmanagement frameworks, and used as the basic QoS parameters in other QoS managementarchitectures.
QoS Parameters
Packet Size
Transmission Rate
Response Time
Error Control
Data Corruption
Data Loss
Data Replication
Ordered Delivery
Four main functions for QoS are: negotiation, mapping, resource reservation, and delivery.
QoS negotiation involves settling down the differences between what has been requested and whatcan be provided between the parties involved in a unicast or multicast connection. In general, anincrease or reduction of a single QoS parameter value is possible. QoS mapping has to be done invarious separate steps. The main goal is for the end-system and the network to provide the QoSrequested by applications. Therefore, application QoS (A-QoS) has to be mapped onto a systemQoS (S-QoS) or network QoS (N-QoS), and onto protocol-relevant parameters. S-QoS themselveshave to be mapped onto resources and parameters of the applied operating system and the networkresources as well. Resource reservation mainly applies to two areas: the local end-system (normallythe operating system), and the intermediate system (normally the network). In general, tasks encompass the reservation of resources, such as memory (buffer), bandwidth, processing time(CPU), and scheduling mechanisms for certain application requirements. Finally, every mentionedmethod above is irrelevant if a communication protocol is not available. The primary tasks ofprotocol processing delivers the requested QoS to the application.
3. QoS MIB
In this section, we define a generic Quality-of-Service (QoS) management information base(MIB) for multimedia services. We call it a generic QoS MIB because it contains QoS relatedmanagement information common to most (if not all) multimedia services. Thus, the QoS MIB canbe used to monitor and control QoS values in multimedia services. However, if specific QoSmanagement information must be obtained in order to manage a particular multimedia service, theQoS MIB can easily be extended by adding the service-specific QoS management information.
In designing our QoS MIB, we have used much work that was previously done by severalIETF working groups. Recent efforts by IETF on defining MIBs for integrated services are similarto ours. This includes “intSrv MIB” [Bak97a] and “intSrvGuaranteed MIB” [Bak97b]. Additionally,a number of RFCs (from RFC 2210 to RFC 2216 inclusively) are defined for integrated services.
Traditional QoS (as in the ISO standards) mostly referred to measures at the network layer of acommunications system. QoS enhancement was achieved by introducing QoS into transportservices. For multimedia services, however, the notion of QoS must be extended, as many otherservices contribute to the end-to-end service quality. To discuss QoS then, we need a layered modelof the multimedia service with respect to QoS [Nah95]. Figure 1 provides one model.
The internal architecture of a multimedia service consists of three layers: application, system(including operating system services), and network. Above the application at the client side residesa human user. This layered structure implies the introduction of service QoS, application QoS,system QoS, and network QoS. The QoS MIB mainly includes objects which represent values ofeach layered QoS parameters. These managed objects are organized into four logical groups (asillustrated in Figure 2): service, application, system, and network. In each group, we define objectsmaintaining QoS related information. Following is a brief description of the design of each group.
Full QoS MIB definitions can be found in [Jsk98].
3.1 Service Group
The service layer QoS parameters specify the service quality which the user wishes to see orhear (e.g., TV quality of video, telephone quality of audio). The service quality, however, is hard toquantify and its evaluation is subjective and user-dependent. Service QoS parameters include thespeed which the user listens or views the playback of audio/visual media.
We used a five-level scale to define the quality of a multimedia service. Thus, users canpotentially specify one of these levels to express their requirements. The service layer QoSsubsumes the widely accepted user requirements, but also allows the specification of a wide rangeof options. The service group of QoS MIB is defined according to service layer QoS parameters.
3.2 Application Group
The application layer QoS parameters describe requirements for application services specifiedin terms of media quality, which includes media and transmission characteristics, and mediarelations, which specify the relations among media. The media quality consists of a stream andcomponent specifications. The stream specification gives the media characteristics of ahomogeneous media stream such as sample size, sample rate and priority/importance. If theindividual samples in the stream differ in quality, component specification must occur. Eachsubsample must be specified by the user/application in the stream structure using componentspecification. The parameterization also includes an application-oriented specification of therequired transmission characteristics for end-to-end delivery (e.g., end-to-end delay bounds).
Media relations specify relations among the media streams. Synchronization skew representsan upper bound on time offset between two streams in a single direction. This information can beused for a finer granularity scheduling decision of multimedia streams than a sample rateinformation of periodic streams provides. If no skew is specified, the system uses the sample rate ofeach stream for scheduling decisions. Communication relation defines the communication topology,such as unicast (peer-to-peer), multicast (peer-to-group), or broadcast (peer-to-all). The applicationlayer QoS parameters can be mapped from the service layer or directly characterized by users. Theycan be parameterized differently according to media or service type.
3.3 System Group and Network Group
System QoS parameters describe requirements on the communication services and operating system (OS) services resulting from the application QoS. The OS services require followingresources: processing times required for task, secondary storage, and memory buffer. OS resourcesare needed by the application-layer and network-layer tasks to handle input/output of media andsending/receiving connections. The definition of QoS parameters at the system layer requiresmainly CPU scheduling, memory management for buffering, and efficient storage on mass media.
Network layer QoS parameters describe network services requirements. They may be specifiedin three domains. The first domain includes basic parameters (called throughput specification) suchas packet size, packet rate, and burstiness. The second domain includes environment-sensitiveparameters (called flow specification) such as inter-arrival delay, round-trip delay and packet lossrate. The last domain includes the specification of the overall communication requirements (calledperformance specification) such as packet ordering and priorities. We have defined the networkQoS parameters on the basis of intSrv MIB [Bak97a] defined by IETF. Additionally, we have used anumber of other RFCs and referred to RSVP [Wro97], QoS-A [Cam93] and QoS Broker [Kla95] forour definition.
4. QoS Management Services for Distributed Multimedia Services
The QoS management services for distributed multimedia services are classified into four setsof management-layer services: service, application, system and network. These services are used bythe multimedia applications for supporting QoS guarantees. Figure 3 illustrates the servicesincluded as part of the QoS Management Service Object (QMSO), which is composed of a numberof layers and planes.
Figure 3. The Structure of QoS Management Services The service layer offers the user the capability to specify quality requirements and priorities forthe employed media services. This service layer also provides the necessary communicationprimitives for establishing and disconnecting multimedia sessions, translating the underlying QoSparameters to user-perceptible parameters. The application layer translates user requirements formultimedia services to a set of QoS, application, system, and network requirements. It executesvarious control mechanisms to compare and control the user-specified quality with the currentlyavailable quality. The controlling mechanism operates on different layers (service, application,system and network layer). Examples of control mechanisms are media prioritization or automaticactions to prevent resource saturation in case of QoS violation. The system layer functions as atranslator from the application layer QoS parameters to underlying layer QoS parameters andreverse translation. It also has a QoS control mechanism and is responsible for end-system resource management: CPU scheduling, memory and I/O management. The network layer translates upperlayer QoS parameters to network QoS parameters and QoS control.
The connection management plane manages multimedia sessions from a user supplied profile.
This plane encompasses layer-specific connection managers that bind multimedia processing units(MPUs) at each layer in order to meet end-to-end connectivity. The resource management planemanages and monitors resources. It performs both admission testing and and resource reservation atevery level in the end-system. It actively measures the CPU usage, and periodically informs theCPU scheduler of its use. The QoS control plane is the central arbitrator of end-to-end QoS. It iscomprised of layer-specific QoS managers that negotiate resources with peer QoS managers, andmaintains an internal state associated with application specific QoS. This translates user requests formultimedia flows to a set of QoS parameters. The QoS mapping is based on the definition of QoSMIB, described in the previous section. An important function of the QoS control plane is tomonitor layer-specific QoS and report any QoS violations of the contracted profile directly to themultimedia applications. Other QoS violations fielded by QoS control plane include indicationsfrom the resource management plane during the negotiation phase.
The QMSO is composed of two management service objects: the application QMSO(appQMSO) and the network QMSO (netQMSO). The appQMSO object provides functions for theservice, application, and system-layer QoS management services. The netQMSO object providesfunctions for the network-layer QoS management service. A subset of appQMSO and netQMSOdefined in CORBA IDL are given below, showing the QoS translator procedures involved in theQoS control plane, and the admission control procedures involved in the resource managementplane: interface appQMSO {
short queryQMSO ( // QMSO's entry procedure, called by multimedia application
in QosParam srvParam, // service QoS parameter
in QosParam addParam, // additional QoS parameter
out NoticeType notice); // accept, reject, and modify

short Service2App ( // QoS translation from service layer
// parameter to application layer parameter
in QosParam srvParam, // service QoS parameter
out QosParam appParam); // application QoS parameter

short App2System ( // QoS translation from application layer parameter
// to system layer parameter

in QoaParam appParam, // application QoS parameter
out QosParam sysParam); // system QoS parameter
short GetAppQoS (// Get QoS parameter from application QoS profile (QoS MIB)
out QosParam appParam); // application QoS parameter
short SetAppQoS ( // Set QoS parameter in application QoS profile (QoS MIB)
in QosParam appParam); // application QoS parameter
short GetSysQoS (// Get QoS parameter from system QoS profile (QoS MIB)
out QosParam sysParam); // system QoS parameter
short SetSysQoS ( // Set QoS parameter in system QoS profile (QoS MIB)
in QosParam sysParam); // system QoS parameter
short AdmitSysQoS ( //Admission test procedure
in QosParam sysParam,
out NoticeType notice); // accept, reject, and modify
short NegotiateAppQoS ( // Negotiates system QoS between service sender and receiver
in QosParam appParam,
out NoticeType notice);
TrapType MonitorSysQoS ( // Monitor local system resource
in ServiceType service, // Best or Guaranteed
in TrapType trap, // signal or no_signal
in AdaptType adapt); // Adapt, signal, or no_action

interface netQMSO {
short Sys2Net ( // QoS translation from system layer parameter
// to network layer parameter
in QosParam sysParam, // system QoS parameter
out QosParam netParam); // network QoS parameter

short App2Net ( // QoS translation from application layer parameter
// to network layer parameter

in QoaParam appParam, // application QoS parameter
out QosParam netParam); // network QoS parameter

short GetNetQoS (// Get QoS parameter from network QoS profile (QoS MIB)
out QosParam netParam); // network QoS parameter
short SetNetQoS ( // Set QoS parameter in network QoS profile (QoS MIB)
in QosParam netParam); // network QoS parameter
short AdmitNetQoS ( //Admission test procedure
in QosParam netParam,
out NoticeType notice); // accept, reject, and modify
short NegotiateNetQoS ( // Negotiates network QoS between service sender and receiver
in QosParam netParam,
out NoticeType notice);
TrapType MonitorNetQoS ( // Monitor network resource
in ServiceType service, // Best or Guaranteed
in TrapType trap, // signal or no_signal
in AdaptType adapt); // Adapt, signal, or no_action

The CORBA IDL definition for QMIO (QoS Management Interface Object) is given below. QMIOis an instrumentation object created for each object that needs to be managed related to QoS. Itprovides QoS-related information to QMSO from the service object it is instrumented into. It alsoallows QMSO to modify various values as part of QoS control actions.
interface QMIO {
readonly attribute short attr_count;
boolena get ( // Get value from a service object in MAESTRO

in OID oid, // Object ID
out Attr attr); // Attribute value

boolena set ( // Set value into a service object in MAESTRO
in OID oid, // Object ID
in Attr attr); // Attribute value

5. Prototype Implementation
In Section 3, we presented a MIB definition for the QoS management of distributedmultimedia services. Using this and the QoS management services defined in Section 4, we havedeveloped a prototype QoS management system to validate our concepts. The target system wewish to manage is MAESTRO [Yun97], which is a CORBA-based distributed multimedia system.
MAESTRO is an object-oriented, distributed multimedia system, whose goal is to providemultimedia services needed to easily develop and operate a variety of multimedia applications.
MAESTRO is composed of four essential distributed service objects: Name Service Object (NSO),Communication Service Object (CSO), Session Service Object (SSO), and Storage and RetrievalService Object (SRSO).
The Management Service Object (MSO) manages and controls service objects in MAESTRO[Jyk97a, Jyk97b]. MSO interacts with managed service objects via Management Interface Object(MIO) for management purposes. We have extended MSO and MIO and developed QoSManagement Service Object (QMSO) and QoS Management Interface Object (QMIO) as shown inFigure 4. QMIO is a specialized version of MIO used for accessing management informationrelated to QoS from multimedia service objects. That is, QoS parameters defined in the QoS MIBare instrumented into the managed objects in the form of QMIO.
Figure 4. A Web-based QoS Management System Prototype for MAESTRO In the QoS management system, QMSO performs QoS functions (including QoS control, QoSmapping, resource monitor and control). When users start a multimedia application (such as videoconferencing), they are asked to provide the quality of service desired. In our prototype, this isdone by the user-level QoS manager tool, specifying the quality of video, quality of audio, windowsize, etc. QMSO then translates these user requirements into application-specific parameters, andinto QoS requirements for the underlying resources. Additionally, QMSO monitors and controls theavailable host and network resources through the interaction with CSO and SSO. If QoS violationoccurs, QMSO sends an event message to MSO and performs an appropriate QoS control procedure.
Because MAESTRO was developed on a CORBA platform, IONA Orbix 2.2 [Ion97], our QoSmanagement service object has also been implemented on the same platform. In order to provide auniform interface and multi-platform management, our management system has been implementedusing Web technology (i.e., using Web server and browser). Administrators can easily manage QoSof MAESTRO services from any platform where a Java-enabled Web browser such as Netscape orInternet Explorer is run. OrbixWeb [Ion96] has been used to allow Java-based managementapplications to invoke methods on management service objects that have been implemented usingCORBA.
Figure 4 is the prototype for our work. The QMSO is initiated by the input of userrequirements. The input of user's QoS requirements are performed by the graphical user interface(GUI). This GUI interactively aids the user in selecting QoS characteristics by viewing a samplevideo with image rate, image height, image width, color, etc. specified through manipulation of aslider (audio might use a test sound passage with specified amplitude, etc.). The user selections(e.g., image height, image color) are translated into application QoS parameters such as frame size,frame rate and others.
The application QoS requirements are mapped into resource requirements for the local system.
The QMSO negotiates with the system, using an admission procedure implemented at theapplication QMSO (appQMSO) level. The admission procedure at the appQMSO level performstwo tests against temporal resources: (1) a local schedulability test to see if the tasks can manageI/O from media devices within the required time bounds; (2) an end-to-end delay test to see if taskscan meet the specified end-to-end delay upper bound.
Once local system resources are reserved, then negotiation at the appQMSO level with theremote appQMSO occurs. The negotiation at the appQMSO concerns the sender's ability toaccommodate requested multimedia characteristics. This depends on the sender having appropriateI/O devices, processing capacity and storage space available. Until these are determined,appropriate bandwidth allocation cannot be made. Furthermore, this negotiation can also be used toexchange additional application information. If the answer is "accept", the appQMSO initiates therequest for network QoS and their resource reservation/allocation, which correspond to themultimedia application QoS.
In network QMSO (netQMSO), the application QoS requirements are translated into networkQoS requirements using the QoS translator. The QoS translator translates the provided parametersinto network QoS parameters using the QoS MIB definition. After translation, admission to thenetwork level is invoked. The admission service at the network level tests both the networkresources such as end-to-end delay and bandwidth, and availability of MAESTRO services (CSO,SSO) through interaction with QMIO. If the admission at the netQMSO is successful, negotiationwith remote netQMSO is initiated by the local netQMSO. Finally, the netQMSO waits for the replyfrom the remote netQMSO. These responses are translated back to the appQMSO, so that the userunderstands which media at what quality will be transmitted. The multimedia application is initiatedwith QoS guarantees and provides multimedia services using MAESTRO's services.
6. Conclusions and Future Work
In this paper, we have proposed a CORBA-based QoS management framework for managingQoS of distributed multimedia services. A generic QoS MIB has been defined for the QoSmanagement of various multimedia services. The definition and adoption of the QoS MIB allowsfor a uniform and standardized QoS management information in multimedia services that may beeasily extended to develop MIBs for specific multimedia services.
The QoS MIB can be used easily to integrate QoS management into standard networkmanagement frameworks (such as SNMP and CMIP [Bla94]). The QoS MIB provides a set oflayered QoS parameters for managing QoS in multimedia services. We have described the QoSmanagement services for distributed multimedia services. They are classified into four layers ofQoS management services: service, application, system and network. Multimedia applications usethese services to support QoS guarantees. These services are included as part of the QoSManagement Service Object (QMSO), which is composed of a number of layers and planes. Wealso have described our efforts on the prototype implementation of a Web-based QoS managementsystem for MAESTRO. The prototype QoS management system uses OrbixWeb [Ion96] to interfacewith the management server, which is implemented as a CORBA object and monitors QoSparameters in distributed multimedia services, which are also implemented as CORBA objects.
Obviously, the complete standardization of the QoS MIB is a complex and time-consuming process.
Therefore, our work is not completed but progressing.
Many open issues still require further investigation. A better understanding of userrequirements is necessary and system layer QoS parameters need to be refined. More work is alsoneeded in mapping user requirements into application layer parameters and resource (includingsystem resource and network resource) parameters. Additional parameters for each layer should betaken into account. More work is also needed to define accurate QoS control mechanisms.
Our work, moving towards the vision of a real multimedia environment, makes the best use ofnetwork protocols and operating systems that offer QoS guarantees. Unfortunately, today’sdominant network, the Internet, ensures only a best-effort approach considering data delivery. Thedeployment of network protocols that offers QoS guarantees [Ban94] has been rather disappointingdue to required transition to new network technologies like ATM.
From the given scenario emerges the need for a set of interrelated protocols that offer QoSguarantees and that easily integrate with the Internet protocol suite. The IETF has developed itsResource Reservation Protocol (RSVP) [Bra97] that permits the reservation of network bandwidthand assignment of priorities to various traffic types. The Real Time Protocol (RTP) [Sch96] worksalongside TCP, providing end-to-end delivery of such data as video broadcasting and multi-participant audio and video. With the deployment of these protocols, multimedia services that runon the Internet may offer QoS guarantees. With such guarantees, standardized QoS managementinformation like QoS MIB will be essential for QoS management in distributed multimedia services.
[Alp95] B. Alpers, "Applying domain and policy concepts to customer network management", Proc.
of ISS'95, Berlin, April 1995, Vol. 2, pp. 356-360.
[Ban94] A. Banerjea et al, "The tenet Real-time Protocol Suite:Design, Implementation, and Experiences", University of California at Berkeley, TR-94-059, November 1994.
[Bak97a] F. Baker, J. Krawczyk, and A. Sastry, "Integrated Services Management Information Base using SMIv2", RFC 1213, September 1997.
[Bak97b] F. Baker, J. Krawczyk, and A. Sastry, "Integrated Services Management Information Base Guaranteed Service Extensions using SMIv2", RFC 1214, September 1997.
[Bla94] U. Black, Network Management Standards: SNMP, CMIP, TMN, MIBS, and Object Libraries, McGraw Hill, November 1994.
[Bra94] R. Braden, D. Clark, and S. Shenker, "Integrated Services in the Internet Architecture: an [Bra97] R. Braden, D. Clark, and S. Shenker, "Resource ReServation Protocol (RSVP) – Version 1 Functional Specification", RFC 2205, 1997.
[Cam93] A. Campbell, G. Coulson, F. Garc, D. Hutchison, and H. Leopold, "Integrated Quality of Service for Multimedia Communications'', Proc. IEEE INFOCOM'93, San Francisco,USA, April 1993, pp. 732-739.
[Cro95] S. Cronjaeger and P. Vindeby, "Translation of multimedia user requirements into ATM switching requirements by using appropriate transport protocols", Proc. of ISS'95. Berlin,April 1995, Vol. 1, pp. 180-184.
[Gad95] B. Gadher et al, "A distributed broadband metropolitan network for residential multimedia applications", Proc. of ISS'95, Berlin, April 1995, Vol. 2, pp. 190-194.
[Geo94] F. Georges, "Performance evaluation & management of interconnected networks", Proc. of SBT/IEEE International Telecommunications Symposium, Rio de Janeiro, August 1994,pp. 322-326.
[Hal95] J. Hall et al, "Customer requirements on teleservice management”, Integrated Network Management IV", Ed. A. S. Sethi et al., Chapman & Hall 1995, pp. 143-155.
[Hut94] D. Hutchison et al, "Quality of service management in distributed systems. Network and Distributed Systems Management", Ed. M. Sloman, Addison-Wesley 1994, pp. 273-302.
[Int95] IETF, Slides from IETF meeting 31, Integrated Service Working Group,, 1995.
[Ion96] IONA, OrbixWeb, IONA Technologies Ltd, Release 2.0, December 1996.
[Ion97] IONA, Orbix 2. IONA Technologies Ltd, Release 2.2, March 1997.
[Iso95] ISO, "Quality of Service Framework'', ISO/IEC JTC1/SC21/WG1 N9680, International [Jsk98] J. S. Kim, J. W. Hong and H. W. Park, "Design and Implementation of Quality of Service (QoS) MIB for Distributed Multimedia Services", Proc. of IEEE GLOBECOM'98, Sydney,Australia, Nov. 8 - 12 1998.
[Jun93] J. Jung and D. Seret, "Translation of QoS parameters into ATM performance parameters in B-ISDN", Proc. of IEEE INFOCOM'93, San Francisco, March 1993, pp. 748-755.
[Jyk97a] J. Y. Kong, J. W. Hong, J. T. Park and D. J. Kim, "A CORBA-Based Management Framework for Distributed Multimedia Services and Applications", Proc. of DSOM,Sydney, Australia, October 1997, pp. 132-144.
[Jyk97b] J. Y. Kong, T. H. Yun, S. W. Park, S. H. Kim, Y. M. Shin and J. W. Hong, "Design and Implementation of a Management System for CORBA-Based Distributed Systems", Proc.
of Asia-Pacific Network Operations and Management Symposium, Seoul, Korea, October1997, pp. 473-485.
[Kla95] K. Nahrstedt and J. M. Smith, "The QoS Broker", University of Pennsylvania, 1995.
[Laz91] A. A. Lazar and G. Pacifici. "Control of resources in broadband networks with Quality of Service guarantees", IEEE Communications Magazine, October 1991, pp. 66-73.
[Min89] Minzer, S.E, "Broadband ISDN and asynchronous transfer mode (ATM)", IEEE Communications Magazine, September 1989, pp. 17-24.
[Nah95] K. Nahrstedt and R. Steinmetz, "Resource management in networked multimedia systems", IEEE COMPUTER, pages 52--63, May 1995.
[Put95] K. J. Putz, "Field trials of multimedia applications in ATM networks: Quality-of-Service, functional and implementation aspects", Proc. of ISS'95, Berlin, April 1995, Vol. 2, pp.
[Sch96] H. Schulzrinne et al, “RTP: A Transport Protocol for Real-time Applications”, RFC 1889, [She94] S. Shenker, "An architecture of future Internet", Proc. of Hot Interconnects Symposium, [She95] S Shenker and J. Wroclawski, "Network Element Specification Template", Internet Draft, [Som95] F. Somers and M. Edholm, "Intelligent resource dimensioning in ATM networks", Proc. of ISS'95, Berlin, April 1995, Vol. 2, pp. 62-66.
[Woo94] M. Woo et al, "A synchronization framework for communication of pre-orchestrated multimedia information", IEEE Network, January 1994, pp. 52-61.
[Wro97] J. Wroclawski, "The Use of RSVP with Integrated Services", RFC 2210, September 1997.
[Yam95] N. Yamanaka et al, "Full-Net: A flexible multi-QoS ATM network based on a logically con- figured VC-network", Proc. of ISS'95 Berlin, April 1995, Vol. 2, pp. 195-199.
[Yun97] T. H. Yun, J. Y. Kong and J. W. Hong, "A CORBA-based Distributed Multimedia System", Proc. of the Fourth Workshop on Distributed Multimedia Systems, Vancouver,Canada, July 1997, pp. 1-8.


NATIVES FOR PLANTING IN MENOMINEE COUNTY, WISCONSIN *see publication "Wisconsin Native Plant Sources" also**for more information and photos, go to SCIENTIFIC NAME MATURE SIZE CHARACTERISTICS decent ash for wetter areas, shade tolerantbeautiful, long lived ash that has hard woodpretty tree, but attacked by many insectspretty tree, but at

Microsoft word - 2-19-13 msma continues to be on the market

Drexel Chemical Company MSMA CONTINUES TO BE ON THE MARKET The Organic Arsenical Products Task Force (OAPTF) would like to clarify to its users and distributors that under the terms of the 2009 agreement with the Environmental Protection Agency (EPA) the sale, distribution, and use of MSMA products labeled for golf course, sod farms, and highway rights of way will continue. With the re

Copyright © 2010-2018 Pharmacy Drugs Pdf