D. Ferrari, J. Ramaekers and G. Ventre, `Client-Network Interactions in Quality of Service Communication Environments', IFIP Transactions C ( Communication Systems ),Vol. C-14, pp 221-234, 1993.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....is now commerciallyavailable in some network interface equipment. Even for static connection requests, the call setup process mayinvolve a negotiation between the user and the networkinwhich the user modifies their request to conform to the current level of resource utilization. For example [21] outlines a real time channel setup procedure in which detailed information is sent to the user if a connection request is refused, allowing the user to take this feedbackinto account in a revised request. Some work is underway on modifying user traffic inputs via pricing in the Internet [22] and ....

D. Ferrari, J. Ramaekers and G. Ventre, `Client-Network Interactions in Quality of Service Communication Environments', IFIP Transactions C ( Communication Systems ),Vol. C-14, pp 221-234, 1993. 18

....protocol which runs on top of RTIP and provides sequenced and periodic delivery of continuous media samples with QoS control over throughput, delays and error bounds. The Tenet Group makes a distinction between deterministic and statistical guarantees for hard real time and continuous media flows [50], respectively. In the deterministic case, guarantees provide a hard bound on the performance of all cells within a session. Statistical guarantees promise that no more than x of packets would experience a delay greater than specified, or no more that x of cells might in a session might be ....

Ferrari D, RamaekersJ,Ventre G (1992) Client-Network Interactions in Quality of Service Communication Environments. In: Proc. 4th IFIP Conference on High Performance Networking, University of Li ege Belgium

....values define an interval that should be enforced. Especially, threshold defines a hard bound that is not allowed to be exceeded. Moreover, QoS parameters are of different importance to applications. This can be expressed by several types of service (ToS) such as the three types suggested by [3]: a deterministic, a statistical, and a best effort service. The ToS can be selected separately for each QoS parameter [2] Internal resource managers are responsible for reservation and assignment of resources according to the individual service requirements. Typical tasks include admission ....

Ferrari, D. et al.: Client-Network Interactions in Quality-of-Service Communication Environments; in A. Danthine, O. Spaniol: High Performance Networking IV; IFIP, North-Holland, Amsterdam, 1993, pp. 221-234.

....of Service policy, and synchronisation (respect of the time constraints) This explains why these protocols are not analyzed in the previous survey. Examples of these new development are : The TENET group, in ICSI Berkeley, designs and experiments with new communication systems [Ferrari 92a] Ferrari 92b] for multimedia environments and high performance network. ST II [Topol 90] is an IP layer protocol that provides end to end guaranteed service across an internet (but is not originally an Internet work) ST II is intended to support and facilitate the implementation of a wider range of ....

....Serge Dassault, Rseaux Locaux Temps Rel Militaires. Protocoles Couche Transfert Recital. GAM T 103 C, March 1989. Ferrari 92a] D. Ferrari, A. Banerjea, H. Zhang, Network Support for Multimedia, a Discussion of the TENET Approach. Technical Report TR 92 001. ICSI, UC Berkeley. November 1992. Ferrari 92b] D. Ferrari, J. Ramaekers, G. Ventre. Client Network Interactions in Quality of Service Communication Environments. Proceedings of the fourth IFIP International Conference on High Performance Networking (HPN 92) Lige. December 1992. Frederick 92] R.Frederick, NV X11 video conferencing tool, ....

[Article contains additional citation context not shown here]

D. Ferrari, J. Ramaekers, G. Ventre. Client-Network Interactions in Quality of Service Communication Environments. Proceedings of the fourth IFIP International Conference on High Performance Networking (HPN '92). Lige. December 1992.

....parameters in a quantitative manner, the commitment clause allows these requirements to be refined in a qualitative way so as to allow a distinction to be made between hard and soft network performance guarantees. There are broadly three classes of service commitment the network can support [13]: i) deterministic, which is typically used for hard real time performance applications; ii) statistical, which allows for a certain percentage of violations in the requested flow spec, and is particularly suitable for continuous media applications; and iii) best effort, the lowest priority ....

....the forward reservation service. However, it is clearly difficult to determine the duration of interactive communication sessions, and therefore we remain somewhat sceptical about enforcing such a regime upon the user. We include the duration time in the service as a marker for further study. In [13] an advance reservation mechanism is described whereby the network may allow the user an extension after the specified duration has expired. This is achieved without disruption to other users who have pre reserved resources. 3.6 cost t The QoS A project has not yet addressed the issues of cost and ....

[Article contains additional citation context not shown here]

Ferrari, D., Ramaekers J. , and G. Ventre, "Client-Network Interactions in Quality of Service Communication Environments", Proc. 4th IFIP Conference on High Performance Networking, University of Liege, Liege, Belgium, December 1992.

....the network and what resources to reserve for it. The connection request is refused by the network if accepting it would lead to QOS degradation for one or more currently active connections which are complying with their contracts. Some CAC schemes provide feedback on the reason for the refusal [3], which guides the user in submitting a revised request. Some proposed CAC schemes reserve the peak bandwidth needed by an accepted connection deterministic multiplexing as required for constant bit rate ( CBR ) sources. The gain in network utilization possible by taking advantage of the ....

D. Ferrari, J. Ramaekers and G. Ventre, `Client-Network Interactions in Quality of Service Communication Environments', IFIP Transactions C ( Communication Systems ), Vol. C-14, pp 221-234, 1993.

....not between the consecutive layers. The specification of the negotiation capabilities in the transport layer is described in the framework of the OSI 95 Transport Service [8] In [17] the CCITT I Series perspective is presented. A signaling scheme from user to the ATM network is described in [17] [26]. The signaling protocol EXPANSE [23] supports the establishment and modification of complex multimedia connections for B ISDN. There is no renegotiation in CCITT ATM Recommendations ( 17] Renegotiation is considered in signaling protocol EXPANSE [23] Lancaster transport protocol [28] 18] 2. ....

D.Ferrari, J.Ramaekers, G.Ventre, "Client-Network Interactions in Quality of Service Communication Environments", Proc. 4th IFIP Conf. on High Performance Networking, pp. E11E1 -14, Liege, Belgium, December 1992

....applications has to consider these characteristics in order to provide an efficient communication facility. The two best known protocols stacks, Internet and OSI, do not address these issues[Bla92, Tou95] and many proposals have been developed to try to solve this problem as presented in [Cam93, Cou95, Fer92, Haf95, Liu93]. All these proposals use the concept of Quality of Service. Quality of Service (QoS) is a general way to express a set of characteristics for a multimedia application. In the highest level, QoS can be interpreted as an image quality (color or black and white) or a sound quality (telephone or CD ....

D. Ferrari, J. Rmaekoers, G. Ventre, Client Network Interaction in Quality of Service Communication Environment,

....and end system resources. 1.2 Related Work Despite the fact that the necessity of mechanisms for ReRA seems to be accepted in the research community, at least to our knowledge, only few work has been done so far in the field of ReRA. Ferrari, Ramaekers and Ventre are among the first who notice in [6] that ReRA is a useful concept. They describe the parameters for the start and duration of a reservation but go not further into the requirements of ReRA nor do they discuss design aspects. Campbell, Coulson and Hutchinson describe in [1] their Quality of Service Architecture and specify start and ....

D. Ferrari, J. Ramaekers, G. Ventre: "Client--Network Interactions in Quality of Service Communication Environments", Proceedings of the Fourth IFIP Conference on High Performance Networking, University Liege, Belgium, December 1992, pp. E1-1 -- E1-14.

.... in addition to the current best effort service: controlled delay [Shenker,95b] which tries to provide several levels of delay which the application can choose between; predicted delay [Shenker,95b] which provides a statistical delay bound similar to the Tenet Group s statistical service [Ferrari,92] and the Comet Groups guaranteed service [Lazar,90] guaranteed delay [Shenker,95b] which provides an absolute guaranteed delay bound. Each service definition also specifies QoS parameters used to invoke the service. The current set of defined services are appropriate for a wide range of ....

....parameters in a quantitative manner, the commitment clause allows these requirements to be refined in a qualitative way so as to allow a distinction to be made between hard and soft network performance guarantees. There are broadly three classes of service commitment the network 75can support [Ferrari,92] i) guaranteed, which is typically used for hard real time performance applications; ii) statistical, which allows for a certain percentage of violations in the requested flow spec, and is particularly suitable for continuous media applications; and iii) best effort, the lowest priority ....

[Article contains additional citation context not shown here]

Ferrari, D., Ramaekers J. , and G. Ventre, "Client-Network Interactions in Quality of Service Communication Environments", Proc. 4th IFIP Conference on High Performance Networking, University of Liege, Liege, Belgium, December 1992.

....selection policy and relative priority. Three transport quality of service semantics in addition to best effort are proposed for this service: compulsory, threshold and maximal QoS. The Tenet Group at the University of California at Berkeley have developed CMTP[46] which operates on top of RTIP [83] and provides sequenced and periodic delivery of continuous media samples with QoS control over throughput, delays and error bounds. Notification of all undelivered and or corrupted data can be provided if the client selects this option. The HeiTS project [47] at IBM Heidelberg has developed a ....

....QoS management aspects of Clark s QM. application RTP, etc UDP IP data packets control packets RSVP, etc dispatcher SA SA SA SA: service agent quality management interface: 4. 7 The Tenet Architecture The Tenet Group at the University of California at Berkeley have developed a family of protocols [83] [84] which run over an experimental wide area ATM network. The protocol family (as illustrated in Figure 4 10) includes a Real Time Channel Administration Protocol (RCAP) 61] in addition to Real Time Internet Protocol (RTIP) Continuous Media Transport Protocol (CMTP) 46] The former provides ....

[Article contains additional citation context not shown here]

Ferrari, D., Ramaekers J. , and G. Ventre, "Client-Network Interactions in Quality of Service Communication Environments", Proc. 4th IFIP Conference on High Performance Networking, University of Liege, Liege, Belgium, December 1992.

....requirements change. In this case, hard guarantees require dedication of sufficient resources to meet the maximum need of an application. Often such a strong commitment is not needed and wasteful [15] Other approaches use statistical guarantees or specify a range of acceptable quality parameters [6, 11]. This flexibility enables the system to reduce quality temporarily in transient overload situations resulting in more tasks being admitted and in a better resource utilization. 2.2.2 Resource Management Tasks Several categories for resource management tasks and QoS processing steps have been ....

D. Ferrari, J. Ramaekers, and G. Ventre. Client-network interactions in quality of service communication environments. In Proc. of the 4th IFIP Conference on High Performance Networking, Laige, Belgium, Dec. 1992, pp. E1-1 - E1-14.

....attractions of ATM is that out of band control and the connection oriented approach allow a Quality of Service (QoS) to be associated with each connection. Certain guarantees can be made about the performance of individual circuits because the aggregate load from all circuits can be calculated [FRV92] Ideally the wireless network should provide the same QoS interface as the wired network, but in practice some details of the semantics are likely to be different. In a mobile environment, more flexibility may well be desirable such as providing a range of acceptable performance values which a ....

D. Ferrari, J. Ramaekers, and G. Ventre. Client-network interactions in quality of service communication environments. Technical report, Tenet Group, University of California, Berkeley, 1992.

....available to it will change. Other systems guarantee that, over an indefinite period of time, the average latency or bandwidth will be a given value, or within a given bound [8] The types of guarantees that can be given are characterized by Kurose in [11, section 3] and Ferrari et al. in [7]. It seems generally accepted by authors in this field that guarantees are necessary for high speed multimedia applications. It also seems accepted that those guarantees cannot always be met. In the Tenet group s Graceful Adaption of Guaranteed Performance Connections [13] the user program ....

Domenico Ferrari, Jean Ramaekers, and Giorgio Ventre. Client-network interactions in quality of service communication environments. In Proceedings of HPN92, 4th IFIP Conference on High Performance Networking, Liege, Belgium, pages E1--1--E1--14, December 1992. Available at ftp://tenet.berkeley.edu/pub/tenet/Papers/FeRaVe92.ps.

....parameters in a quantitative manner, the commitment clause allows these requirements to be refined in a qualitative way so as to allow a distinction to be made between hard and soft network performance guarantees. There are broadly three classes of service commitment the network can support [Ferrari,92] i) deterministic, which is typically used for hard real time performance applications; ii) statistical, which allows for a certain percentage of violations in the requested flow spec, and is particularly suitable for continuous media applications; and iii) best effort, the lowest priority ....

....the forward reservation service. However, it is clearly difficult to determine the duration of interactive communication sessions, and therefore we remain somewhat sceptical about enforcing such a regime upon the user. We include the duration time in the service as a marker for further study. In [Ferrari,92] an advance reservation mechanism is described whereby the network may allow the user an extension after the specified duration has expired. This is achieved without disruption to other users who have pre reserved resources. 3.6 cost t The QoS A project has not yet addressed the issues of cost ....

[Article contains additional citation context not shown here]

Ferrari, D., Ramaekers J. , and G. Ventre, "Client-Network Interactions in Quality of Service Communication Environments", Proc. 4th IFIP Conference on High Performance Networking, University of Liege, Liege, Belgium, December 1992.

....simple throughput specification. Most multimedia applications will not be sending data at a constant rate. Compressed audio and video will normally be sent at a variable rate. Several ways of specifying a variable level of traffic throughput have been presented in the literature. The Tenet scheme [FRV92] uses the following four parameters to specify traffic throughput. ffl x min : The minimum amount of time between sending frames. ffl x ave : The average amount of time between sending frames. ffl I: The length of the interval over which x ave is determined. ffl s max : The maximum frame size. ....

....the variation in the amount of delay (jit ter) experienced by frames. For example, voice data should be delivered with little jitter to sound good. Specifying a maximum jitter value is equivalent to specifying minimum and maximum delays. Most of the work that has been done (including [DS94] FRV92] Flo94] allows the application layer to specify maximum values for both delay and jitter. Transmission Reliability Transmission reliability gives the application layer a way of specifying the probability that frames will be lost due to buffer overflow or transmission errors. Ap CHAPTER 2. ....

[Article contains additional citation context not shown here]

Domenico Ferrari, Jean Ramaekers, and Giorgio Ventre. Client-network interactions in quality of service communication environments. In Proceedings of the 4th IFIP Conference on High Performance Networking, pages E1--1 -- E1--14, Liege, Belgium, December 1992.

....the time of that article) upcoming ISDN networks. Hence, its applicability to computer based multimedia systems is somewhat limited due to the different application and usage characteristics and the restricted system environment. Ferrari, Ramaekers and Ventre are among the pioneers who notice in [9] that ReRA is a useful concept. They describe the parameters for the start and duration of a reservation. They also discuss that the resource management system (only considering the network but not any other resources such as crucial resources involved in the processing within a video on demand ....

D. Ferrari, J. Ramaekers, G. Ventre: "Client--Network Interactions in Quality of Service Communication Environments", Proceedings of the Fourth IFIP Conference on High Performance Networking, University Liege, Belgium, December 1992, pp. E1-1 - E1-14.

....together specify the relationship between the start of the presentation of the object o and the start of the delivery of the object o. Delay Jitter Constraints: Delay jitter constraints are specified as the maximum tolerable variations in the delay suffered by an object or a network packet [4, 5]. In order to specify the delay jitter constraints, we need to introduce the following variables: ffl D(o) Denotes the average delay suffered by an object on its transfer over the computer network. ffl D max (o) Denotes the maximum delay that is permissible for the object transfer. The ....

....may now be specified as follows: D max (o) Gamma D(o) ffi which is easily seen to be a difference constraint. Cell Loss Probability Constraints: The cell loss probability constraints describe the maximum percentage of cells (or the network packets) that can be lost during an object transfer [4, 5]. In order to specify the cell loss probability constraints, we need to introduce the following variables : ffl C(o) Denotes the number of network cells that an object o is composed of. ffl L(o) Denotes the number of network cells that were transferred to the client. The cell loss probability ....

D. Ferrari, J. Ramaekers and G. Ventre, `Client-Network Interactions in Quality of Service Communication Environments', Proc. of High Performance Networking, 4th IFIP Conf. on High Performance Networking, Liege, Belgium, 14-18, December '92.

No context found.

D. Ferrari, J. Ramaekers and G. Ventre, `Client-Network Interactions in Quality of Service Communication Environments', IFIP Transactions C ( Communication Systems ),Vol. C-14, pp 221-234, 1993.

No context found.

D. Ferrari, J. Ramaekers, and G. Vente, \Client-network interactions in quality of service communication environments," in Proceedings of the 4th IFIP Conference on High Performance Networking, (Liege, Belgium), Dec. 1992.

No context found.

FERRARI, D., RAMAEKERS, J., AND VENTE, G. Clientnetwork interactions in quality of service communication environments. In Proceedings of the 4th IFIP Conference on High Performance Networking (Liege, Belgium, Dec. 1992).

No context found.

D. Ferrari and J. Ramaekers. Client-Network Interactions in Quality of Service Communication Environments. In Proc. 4th IFIP Conference on High Performance Networking, 1992.

No context found.

Ferrari, D., J. Ramaekers, and G. Ventre, "Client-Network Interactions in - Quality of Service Communication Environments", Proc. 4th IFIP Conference on High Performance Networking, University of Liege, Liege, Belgium, December 1992.

No context found.

D. Ferrari, J. Ramaekers and G. Ventre (1992) Client-Network Interactions in Quality of Service Communication Environments, Proc. of High Performance Networking, 4th IFIP Conf. on High Performance Networking, Liege, Belgium, 14-18, December '92.

No context found.

Ferrari, D., Ramaekers J. , and G. Ventre, "Client-Network Interactions in Quality of Service Communication Environments", Proc. 4th IFIP Conference on HighPerformance# Networking, University of Liege, Liege, Belgium,December,# 1992.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC