Ralf Steinmetz. Analyzing the multimedia operating system. IEEE Multimedia, 2(1):68--84, 1995.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....does not immediately extend to general computation. Hierarchical schedulers have been advanced to support heterogeneity of applications [4, 6] They employ leaf schedulers of diverse types. Classical real time schedulers like rate monotonic or earliest deadline first lack the firewall property [11, 15]. To adapt to dynamic application behavior, certain scheduling algorithms require close application participation, and sophisticated schedulability tests [14] Other systems have appealed to policing mechanisms external to the scheduling algorithm, such as priority depression [10, 13, 17] We ....

Ralf Steinmetz. Analyzing the multimedia operating system. IEEE Multimedia Magazine, 2(1):68--84, 1995.

....load this delay can become significant. However, it is independent of the Internet model (whether best effort or any other enhanced version) and any reductions in this delay would imply software enhancements to Operating System (OS) kernel such OS s are called Multimedia Operating system s [38] that provide enhanced process , resource , file and memory management techniques, with real time scheduling and the application. 3.1.2: Packet Transmission Delay 20 This is the time taken by the physical layer at the source to transmit the packets over the link. This delay depends on ....

R. Steinmetz, "Analyzing the multimedia operating system", IEEE Multimedia, vol: 2(1), Spring 1995, pp: 68-84.

....or mutual exclusion as two or more finite state machines try to take control of the same resource. Problems also occur when multimedia systems are distributed over a network [6] Issues relating to compression, latency, jitter, bandwidth, and operating system characteristics become important [3, 5]. It is clear that even content creators need to have a precise understanding of the technology they will use for the delivery of multimedia data. Conclusion Multimedia is a diverse and rapidly evolving discipline. Much of the activity central to multimedia does not takes place under the ....

Steinmetz, R. Analyzing the multimedia operating system. IEEE Multimedia, Spring 1995, 68--84.

....in software, requires not only the support of the respective lower layers but also operating system support. The main points here are CPU scheduling, memory management for bu ering, and ecient le storage on mass media. Real time CPU scheduling is discussed intensively in the literature [22, 23, 27]. The aim of bu er management is to avoid jitter and to minimize copy operations [27] This may be achieved by a large number of bu ers, which, however, may in turn lead to a waste of resources. Ecient le storage is achieved by storage hierarchies [21] and disk striping. Important QoS parameters ....

....operating system support. The main points here are CPU scheduling, memory management for bu ering, and ecient le storage on mass media. Real time CPU scheduling is discussed intensively in the literature [22, 23, 27] The aim of bu er management is to avoid jitter and to minimize copy operations [27]. This may be achieved by a large number of bu ers, which, however, may in turn lead to a waste of resources. Ecient le storage is achieved by storage hierarchies [21] and disk striping. Important QoS parameters for the operating system are size and number of bu ers , scheduling class (e.g. ....

R. Steinmetz. Analyzing the Multimedia Operating System. IEEE MultiMedia, 2(1):68-84, Spring 1995.

....necessary for real time multimedia operating systems [3,46,53,60,87,102,127] Herrtwich [42] gives an argument for resource reservation and careful scheduling in these systems. Others prefer a best effort approach to OS design for multimedia applications [19,21,90] Recent survey papers [116,132] and books [12,115] discuss work in this area. One of the simplest ways to do resource allocation for multimedia applications is to dedicate an entire machine to a single multimedia application. Multimedia applications for single user personal computers typically assume that only a single ....

R. Steinmetz. Analyzing the Multimedia Operating System. IEEE Multimedia, 2(1):63--84, Spring 1995.

....and dependability (fault tolerance) Multimedia applications in space exploration, plan control, robotics control, and aircraft avionics are traditional dependable multimedia applications. A number of approaches have investigated to cope with realtime scheduling for multimedia tasks. Paper [1] has studied rate monotonic and earliest deadline first scheduling algorithms for multimedia research systems and test beds. The drawback of rate monotonic algorithm is that it only deals with periodic tasks and could not schedule dynamically. When system is overloaded, earliest deadline first ....

Ralf Steinmetz, Analyzing the Multimedia Operating System , IEEE Multimedia, Spring pp.68-84, 1995.

....the device, responsible for moving data. In a busattached system, the mover may be the device driver. For a NAP, however, part or all of the mover s functionality might be implemented at the NAP itself. 7 as graphics data. While research is being conducted on such topics within non NAP systems [57] and for networked systems, I know of no work specifically related to using NAPs for real time services. The Fibre Channel community has considered isochronous classes of service but the work is low priority. The ATM standard supports isochronous transfers. This issue is tightly bound to the ....

....disks distributed arbitrarily around the cluster, attached to other host nodes or to one or more Hierarchical Storage Controllers. There is a large body of work on network and operating systems support for multimedia, notably the workshops on digital audio and video [30] Also see papers such as [57] . These have not focussed specifically on NAPs, but the principles are important. 36 http: www.pegasus.esprit.ec.org papers pegpapers.html 10 7 Existing Network Attached Peripherals Numerous network attached peripherals using IPI 3 over HiPPI already exist and are in production use, ....

R. Steinmetz. Analyzing the multimedia operating system. IEEE Multimedia, 2(1):68--84, 1995.

.... operating system [Sha et al. 1994] The strong endorsement of several research and development groups of the earliestdeadline first and rate monotonic scheduling as most suitable resource allocation policies for continuous media servers [Nagarajan and Vogt 1992] Ramamritham and Stankovic: 1994] [Steinmetz 1995] and ATM switch scheduling further emphasizes importance of this hard real time scheduling approach. The topic of partitioning of single or multiple tasks during compilation for parallel and distributed programmable platforms has been a widely studied topic [Almasi and Gottlieb 1994] Scheduling ....

Steinmetz, R. 1995. "Analyzing the Multimedia Operating Systems", IEEE Multimedia, 2, 1, 68-84.

....uninterrupted media playback over fast networks. Several such disk storage schemes are described by Gemmell et al. 13] The need for augmenting an operating systems and its subsystems in order to support data and computation intensive multimedia applications has been discussed by Steinmetz [37]. Operating systems for streaming media applications need process management functions that deal with soft and hard real time restrictions in addition to resource reservations and guaranteed timing delays for memory access. An overview of these and other issues involved in building a streaming ....

R. Steinmetz. Analyzing the multimedia operating system. In IEEE Multimedia, April 1995. BIBLIOGRAPHY 75

....power is available. The focus of this work is to achieve the highest levels of processing performance and thereby enable sufficient computing power for a real time operating system to provide real time performance. For further reference on real time operating systems for multimedia, Steinmetz [100] and Nieh et al. 101] provide good overviews of the issues and describe potential implementations. 74 Evaluation Method To accurately evaluate the intrinsic characteristics of the multimedia applications, the compiler was set to apply only classical optimizations while compiling the benchmark ....

Ralf Steinmetz, "Analyzing the Multimedia Operating System," IEEE Multimedia, Spring 1995, pp. 68-84.

....times of another periodic task invocation times of a periodic task . Phase Figure 1. Phased scheduling. Scheduling algorithms proposed for multimedia systems include earliest deadline first and its variants, proportional share, and rate monotonic algorithm(Han and Shin 1995, Kim et al. 1996, Steinmetz 1995). The basic technique proposed in this paper applies to any of the scheduling algorithms. However, we limit our attention to Earliest Deadline First or EDF since intensive studies have been conducted on the mixed scheduling of periodic and aperiodic tasks in the context of EDF(Chetto and Chetto ....

....release time or phase r i , i.e. i = P i ; C i ; r i ) A task instance processes each data unit of a continuous media. The xth instance of i , denoted ix , is released at time ix = r i (x Gamma 1)P i . To avoid congestion, the instance must complete execution before the next one arrives(Steinmetz 1995), that is, its deadline is at the end of its period, ix P i . Without loss of generality, the value of each phase r i is assumed to fall in the interval [0; P i ] Since the phases are relative among themselves, we choose r n = 0 as reference point to all the other phases. Let HP denote the ....

[Article contains additional citation context not shown here]

Steinmetz, R. (1995). Analyzing the multimedia operating system. IEEE Multimedia pp. 68--84.

....over the backbone network. 2. 1 Functional parallelism in processing of frames There are three distinct functions in the communication system: i) multiplexing and de multiplexing of data streams, ii) scheduling control on streams, and iii) packetization and de packetization of frames [8]. On a SUN Sparc processor, we estimate that the functions (i) ii) and (iii) consume about 5 6 milliseconds (msec) and 4 msec respectively for a reasonably sized system (excluding other operating system related overheads) 2 . Parallelism in inter frame processing is coarse grained, in that, a ....

R. Steinmetz. Analyzing Multimedia Operating Systems. IEEE Multimedia, Spring 1995.

.... Station Program and the European Space Agency s on board operating system [45] The strong endorsements from several research and development groups of the earliest deadline first and ratemonotonic scheduling as most suitable resource allocation policies for continuous media servers [28] 37] [48] and ATM switch scheduling [42] further stress importance of this hard real time scheduling approach. A search and force directed heuristics are often used as optimization mechanisms for computationally intractable problems [40] It has been proved that it is the best informed search strategy in ....

R. Steinmetz. Analyzing the multimedia operating systems. IEEE Multimedia, 2(1):68--84, 1995.

.... been recognized that to support end to end QoS guarantees not only network aspects must be considered, but the end system and OS resources must also be taken into account [26] This requirement holds especially for continuous media applications as they have the most stringent resource requirements [36], 34] In step with advances in resource guarantee provision in both fields, researchers identified the need for resource orchestration and developed methods that allow for meeting the user s QoS requirements on an end to end basis [25] 6] Most methods involve QoS negotiation procedures mainly ....

R. Steinmetz. Analyzing the multimedia operating system. IEEE Multimedia, 2(1):68--84, Spring 1995.

....to execution time for CPU scheduling) are not deterministic, because they depend on the position of the disk head and the position of data on disk; and (3) disk I O represents the main performance bottleneck in today s systems. Multimedia data can be managed on disk in two different ways [Steinmetz 95] 1) the file organization on disk is not changed and the required real time support is provided by special disk scheduling algorithms and large buffers to avoid jitter; or (2) the data placement is optimized for continuous multimedia data in distributed storage hierarchies like disk arrays ....

.... objects, program libraries and executables, text, and accounting records, have to be extended for multimedia file systems with realtime characteristics, coping with larger file sizes (high disk bandwidth) and with multiple continuous and discrete data streams in parallel (real time delivery) Steinmetz 95] The fact that in the last years storage devices have become only marginally faster compared to the exponentially increased performance of processors and networks, makes the effect of this discrepancy in speed for handling multimedia data by file systems even more important. This is documented ....

[Article contains additional citation context not shown here]

Steinmetz, R.: Analyzing the Multimedia Operating System, IEEE Multimedia, Vol. 2, No. 1, Spring 1995, pp. 68-84

....as an important, though limited, aspect in the development of mechanisms for real time QoS assurances. The current state of operating systems support for distributed QoS is that multimedia applications (and other time dependent computations) do not play a large enough role in resource management [3, 20, 27, 30]. Rather, the operating system decides how much CPU time, physical memory, network bandwidth, etc. each application receives. Typically, a multimedia application is written assuming that it will receive a fixed, large portion of resources. If, at the time at which access is granted to a resource, ....

....the use of QoS contracts in which applications are not admitted unless the DQM and the operating system guarantee access to specific resources. Contracts of this kind generally require finer control over resource usage than is possible if the DQM operates as a user process. Other researchers [3, 30] indicate that real time scheduling assistance such as Earliest Deadline First or Rate Monotonic is needed in order to support QoS contracts from the operating system. 4 Experiment To begin to evaluate the proposed software architecture, we performed a series of experiments with a simple DQM and ....

Ralf Steinmetz. Analyzing the multimedia operating system. IEEE Multimedia, Spring 1995.

....to support the stringent timing requirements of those applications including resource and precedence constraints. For example, the rate monotonic (RM) and the earliest deadline first (EDF) scheduling algorithms are most widely used in the on going multimedia research systems and testbeds [9]. However, the RM algorithm can handle only periodic tasks and does not work well in a dynamic environment. EDF is optimal only for a single processor and does not scale to multiprocessor systems. EDF also exhibits non deterministic behavior in an overloaded situation. Accommodating both ....

Ralf Steinmetz, Analyzing the Multimedia Operating System, IEEE Multimedia, Spring 1995, pp. 68-84.

.... such as Space Station Program and the European Space Agency on board operating system [Sha94] The strong endorsement of several research and development groups of the earliest deadline first and rate monotonic scheduling as most suitable resource allocation policies for continuous media servers [Nag92, Ram94, Ste95] and ATM switch scheduling [Swa93] further stress importance of this hard real time scheduling approach. A search and force directed heuristics are often used as optimization mechanisms for computationally intractable problems [Rus95] It has been proved that it is the best informed search ....

R. Steinmetz, "Analyzing the Multimedia Operating Systems", IEEE Multimedia, Vol. 2, No. 1, pp. 68-84, 1995.

....problems addressed in [4, 63] Other important differences limit the suitability of continuous media servers to our requirements. First, there are differences in architectural constraints (for example, see Fig. 2.6) and focus. Research on multimedia extensions for operating systems (see surveys [13, 97, 93]) and for file systems [51, 81, 87, 89, 101] address the support of stored continuous multimedia (e.g. predictive prefetching, batched processing, disk layout, optimal buffering, frame interleaving, etc. On the other hand, this dissertation focuses on best effort playback on general purpose ....

....of c = 33ms, 3) control discontinuities between frames to less than 10ms. Such a priori knowledge about the predictability and periodicity of event playback is an inherent assumption built in on scheduling mechanisms for the playback of continuous media (both Schulzrinne [93] and Steinmetz [97] made similar observations) For example, this is particularly true for deadline based of continuous media [47, 44, 75] On the other hand, a re executable content stream lacks periodicity. In this dissertation, I provide a playback characterization of re executable content streams based on the ....

[Article contains additional citation context not shown here]

R. Steinmetz. Analyzing the multimedia operating system. IEEE MultiMedia, 2(1):68--84, Spring 1995.

No context found.

Ralf Steinmetz. Analyzing the multimedia operating system. IEEE Multimedia, 2(1):68--84, 1995.

No context found.

R. Steinmetz, "Analyzing the multimedia operating system," IEEE Multimedia, Vol. 2, No. 1, pp. 68--84, 1995.

No context found.

R. Steinmetz, "Analyzing the multimedia operating system," IEEE Multimedia, Spring 1995.

No context found.

R. Steinmetz \Analyzing the Multimedia Operating System", IEEE Multimedia, Vol. 2 No. 1, 1995

No context found.

R. Steinmetz, Analyzing the Multimedia Operating System, IEEE Multimedia, Spring 1995, pp. 68-84.

No context found.

R. Steinmetz "Analyzing the Multimedia Operating System", IEEE Multimedia, Vol. 2 No. 1, Spring 1995.

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