R. Bettati and J. W.S. Liu, "End-to-End Scheduling to Meet Deadlines in Distributed Systems". In The 12th International Conference on Distributed Computing Systems, pages 452-459, Yokohama, Japan, June 1992.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....all tasks can make their deadlines. Note that while designing and implementing a system that most changes made will affect the real time properties of the system. This makes real time scheduling a global cross cutting dependency. While many different schedulability analysis techniques exist [1, 2, 4, 9, 11, 16, 26], they differ in their assumptions on the task set and none of the existing analysis is applicable to all real time embedded systems. The compatibility between schedulability analyses and the characteristics of the designed system is a typical crosscutting dependency that is hidden from the ....

....subtask T ij meets its intermediate deadline, the whole task meets its deadline D i . Consequently, the distributed schedulability analysis is reduced to the analysis of each node independently with phased offset. This phase offset policy is similar to the Modified Phase Modification Protocol [2] and a protocol described in [26] For the schedulability analysis on each node, we employ Audsley s priority assignment and analysis algorithm found in [1] The Audsley algorithm provides an optimal priority assignment and feasibility test algorithm for static priority tasks with arbitrary start ....

Bettati, R., (1994) End-to-End Scheduling to Meet Deadlines in Distributed Systems, PhD Thesis, Department of Computer Science, University of Illinois at Urbana-Champaign.

....and (c) we use a very di#erent performance metric that measures how well prepared our static mapping is for absorbing run time variations in the system load. Our metric is aimed for evaluating a statically derived mapping for systems that operate in dynamic environments. Research e#orts in (e.g. [42], 43] 44] 45] 19] 46] do consider task assignment as well as task scheduling. The research in [42] and [43] used a very simple partition of processors for allocation of tasks to each partition. In the research e#orts of [45] and [19] a branch and bound search algorithm was used for the ....

....absorbing run time variations in the system load. Our metric is aimed for evaluating a statically derived mapping for systems that operate in dynamic environments. Research e#orts in (e.g. 42] 43] 44] 45] 19] 46] do consider task assignment as well as task scheduling. The research in [42] and [43] used a very simple partition of processors for allocation of tasks to each partition. In the research e#orts of [45] and [19] a branch and bound search algorithm was used for the allocation phase. Even though these e#orts had allocation schemes, these methods were not directly suited ....

R. Bettati and J. W.-S. Liu, "End-to-end scheduling to meet deadlines in distributed systems," in International Conference on Distributed Computing Systems, pp. 452--459, June 1992.

....Yang Yu # , Shriram B. Gundala # , Sethavidh Gertphol # , Howard Jay Siegel # , Anthony A. Maciejewski # , and Viktor Prasanna # Purdue University # University of Southern California School of Electrical and Computer Engineering Department of Electrical Engineering West Lafayette, IN 47907 1285 USA Los Angeles, CA 90089 2560 USA alis, kim42 ecn.purdue.edu yangyu, gundala, gertphol, prasanna halcyon.usc.edu Colorado State University # Department of Electrical and Computer Engineering Department of Computer Science Fort Collins, CO 80523 1373 USA hj, aam colostate.edu ....

....# , Sethavidh Gertphol # , Howard Jay Siegel # , Anthony A. Maciejewski # , and Viktor Prasanna # Purdue University # University of Southern California School of Electrical and Computer Engineering Department of Electrical Engineering West Lafayette, IN 47907 1285 USA Los Angeles, CA 90089 2560 USA alis, kim42 ecn.purdue.edu yangyu, gundala, gertphol, prasanna halcyon.usc.edu Colorado State University # Department of Electrical and Computer Engineering Department of Computer Science Fort Collins, CO 80523 1373 USA hj, aam colostate.edu Abstract Real time ....

[Article contains additional citation context not shown here]

R. Bettati and J. W.-S. Liu. End-to-end scheduling to meet deadlines in distributed systems. In International Conference on Distributed Computing Systems, pages 452--459, June 1992.

....we use a very di#erent performance metric that measures how well prepared our static mapping is for absorbing run time variations in the system load. The purpose of our metric is to evaluate a statically derived mapping for systems that operate in dynamic environments. Research e#orts in (e.g. [9], 12] 17] 26] 46] 47] do consider task assignment as well as task scheduling. The research in [9] and [12] used a very simple partition of processors for allocation of tasks to each partition. In the research e#orts of [26] and [46] a branch and bound search algorithm was used for the ....

....run time variations in the system load. The purpose of our metric is to evaluate a statically derived mapping for systems that operate in dynamic environments. Research e#orts in (e.g. 9] 12] 17] 26] 46] 47] do consider task assignment as well as task scheduling. The research in [9] and [12] used a very simple partition of processors for allocation of tasks to each partition. In the research e#orts of [26] and [46] a branch and bound search algorithm was used for the allocation phase. Even though these e#orts had allocation schemes, these methods were not directly suited ....

R. Bettati and J. W.-S. Liu. End-to-end scheduling to meet deadlines in distributed systems. In International Conference on Distributed Computing Systems, pages 452--459, June 1992.

....and deadlines from end to end constraints. In [1] and [14] similar problems are addressed, but the focus was more on schedulability analysis, and less on the derivation of task parameters. There have also been some studies on the decomposition of end to end dead lines into local task deadlines [16, 2, 5, 11]. Recently, 13] reports a study on deriving task periods based on control performance metrics. 2 Problem Formulation In this section, we first present the system and network model. Then, we formulate a system design model in terms of a task graph and end to end timing constraints. We also ....

R. Bettati and J. W.-S. Liu. End-to-end scheduling to meet deadlines in distributed systems. In Proceedings, IEEE Conference on Distributed Computing Systems, pages 452 459, 1992.

....tinct control loops. Control loop Ci consists of mi tasks ;i, i2, For example, Ci may include a sampling task, an input task which generates a reference input, a computation task which executes the control algorithm, and an output task which issues output commands to the controlled system [ 1]. They form a chain of tasks, as shown in Figure 2. Each control loop Ci has two kinds of timing constraints: 1) maximum loop processing period MacPi, and (2) maximum input to output latency 3 ]axLi. Given a multi rate control system, our problem is to optimize loop processing period T and ....

R. Bettati and J. W. S. Liu. End-to-end scheduling to meet deadlines in distributed systems. In In Proceedings of the 12th International Conference on Distributed Computing Systems, pages 452-459, June 1992.

....algorithms, e.g. heuristic methods, branch and bound, simulated annealing, genetic algorithms, tabu search, constraint programming, see [Redell, 1998] for an overview from an aspect of distributed control. End to end scheduling in distributed systems is treated from different perspectives in [Bettati and Liu, 1992]. A typical flow shop job or a task chain has five subtasks as in figure 2: sensor, the link to the controller, the controller, the link to the actuator and finally the actuator. The task model is as follows: a set of task chains is to be scheduled on m number of processors. Every task in the ....

R. Bettati and Jane W.-S. Liu, "End-to-End Scheduling to Meet Deadlines in Distributed Control", IEEE, 1992.

....in terms of the way it consumes resources (amount, preemtiveness, priority, etc. see [56] All tasks are considered periodic and interruptions are handled using some kind of sporadic attendance scheme [71] They also support certain degree of analysis in certain multiprocessing environment model [67, 15]. Their major strength is that they are of polynomial complexity on the number of tasks [44] We can group these tools according to w hich kind of results are used. Based on Analytical Results. In this subgroup we find tools that use or suggest known results to statically asses the ....

....bounded or bursty events are treated as they were periodic. Verification. Verification Technique Analytical results for periodic task model are used [60, 67, 71, 58, 69, 11, 34] The analysis of multiprocessor systems is based on the multiprocessor model [67] and end to end scheduling model [15]. Queries. Deadline satisfaction, worst case execution time, blocking time, Jitter satisfaction, End to End analysis. Complexity. The formulas are of polynomial complexity on the number of tasks. 2.3.2 Approaches Based On State Space Analysis ffl RTD [37] It is a formal notation based on ....

R. Bettati, J.W.S. Liu. End-to-End Scheduling to Meet Deadlines in Distributed Systems. Proceedindgs of the 12th International Conference on Distributed Computing Systems, Yokohama, Japan, June 1992.

....mostly concentrated on the single processor case. The exploration of parallelism in real time systems has focussed on scheduling tasks simultaneously on a pool of identical processors such that the scheduler makes no distinction between processors. Notable work in the multiprocessor case includes [22, 11, 23, 4]. 11] investigated different ways of deadline assignment to parts of a task on different hosts and their effects on overall deadline miss statistics. 23] looked at the end to end response time of a task under different synchronization protocols. Most of these results are validated by ....

....case, and give a sufficient condition together with a cheaper computational method for schedulability testing. This paper does not deal with the assignment of relative deadlines which was discussed in [19] in the context of the processor allocation problem and more recently by other researchers [22, 11, 23, 4]. A number of heuristics have been proposed for the deadline assignment problem, and some performance bounds have been obtained. A simple heuristic for relative deadline assignment is to divide the end to end deadline evenly to yield the same relative deadline for every frame in a task. Our ....

R. Bettati. End-to-End Scheduling to Meet Deadlines in Distributed Systems. PhD thesis, the University of Illinios at Urbana-Champaign, 1994.

....to individual tasks using the derived periods 1 . While several other techniques exist for deriving task level deadlines and release times (see, for 1 In a recent paper, Nilsson et al. 12] presented an application of PCM on an avionics control software using constraint programming. 5 example, [13, 14, 15, 16, 17, 18]) few techniques actually exist for deriving other types of task level constraints. However, Ramamritham [19] and Abdelzaher Shin [20] propose techniques that generate task clustering recommendations based on heuristic analyses of inter task communication needs and task periods, respectively. ....

R. Bettati and J. W.-S. Liu, "End-to-End Scheduling to Meet Deadlines in Distributed Systems," Proc. of the IEEE Int'l Conf. on Distributed Computing Systems, Yokohama, Japan, June 9--12, 1992, pp. 452--459.

....are forced to modularize software for maintainability and reusability reasons or exploit parallelism for performance reasons. As a consequence of this decomposition, the E T E deadline must be distributed over the component tasks. Many researchers have addressed the deadline distribution problem [5, 6, 7, 8, 9, 10], all under a common assumption that task assignments are known a priori, that is, strict locality constraints. In many real time systems, however, only a small number of task assignments are governed by strict locality constraints, typically those tasks constrained by demands of resources in ....

....Gonz alez Harbour proposed a heuristic iterative approach that, given an initial local deadline assignment, finds an improved solution in reasonable time. For each iteration a new deadline assignment is calculated based on a metric that measures by how much schedulability failed. Bettati and Liu [7] presented a technique for scheduling a system of flow shop tasks. Local deadlines are assigned by distributing E T E deadlines evenly over tasks. For this method, the simplifying assumption is made that execution times are either identical for all tasks or identical for all tasks assigned to the ....

[Article contains additional citation context not shown here]

R. Bettati and J. W.-S. Liu, "End-to-End Scheduling to Meet Deadlines in Distributed Systems," Proc. of the IEEE Int'l Conf. on Distributed Computing Systems, Yokohama, Japan, June 9--12, 1992, pp. 452--459.

....et al. 7] proposed the method to guarantee a system s end to end requirements of real time systems. In [30] Tindell et al. attempted to compute upper bounds on the end to end response time. They also proposed priority assignment in distributed system where jobs have end to end deadlines. In [1], Bettati studied the problem of scheduling a set of jobs with arbitrary release times and end to end deadlines. Our Approach. We propose to address real time scheduling problems by means of analysis based on ACSR VP. In this approach, a specific instance of a problem is specified as an ACSR VP ....

R. Bettati. End-to-end Scheduling to Meet Deadlines in Distributed Systems. PhD thesis, University of Illinois at UrbanaChampaign, 1994.

....distribution is to adopt a divide and conquer approach that first divides the overall problem into smaller problems that are solved locally and then combined to obtain a global solution, thus reducing the complexity of the problem. Many researchers have addressed the deadline distribution problem [1, 2, 3, 4, 5, 6, 7], all under a common assumption that task assignments are entirely known, i.e. strict locality constraints. In many real time systems, however, only a small number of task assignments are governed by strict locality constraints, e.g. those tasks constrained by demands of resources in their ....

....Gonzalez Harbour proposed a heuristic iterative approach that, given an initial local deadline assignment, finds an improved solution in reasonable time. For each iteration a new deadline assignment is calculated based on a metric that measures by how much schedulability failed. Bettati and Liu [4] presented a technique for scheduling a system of flow shop tasks. Local deadlines are assigned by distributing end toend deadlines evenly over subtasks. For this method, the simplifying assumption is made that execution times are either identical for all subtasks or identical for all subtasks ....

[Article contains additional citation context not shown here]

R. Bettati and J. W.-S. Liu, "End-to-End Scheduling to Meet Deadlines in Distributed Systems," Proc. of the IEEE Int'l Conf. on Distributed Computing Systems, Yokohama, Japan, June 9--12, 1992, pp. 452--459.

....the communication links can be modeled as processors, and consequentlymessage transmissions can be modeled as communicationtasks on the link processors. ence constraints among subtasks. The second protocol is an extension to the one proposed by Bettati, which was designed for flow shop systems [11]. This protocol has certain limitations. The third protocol combines the strengths of the previous two protocols while avoiding their shortcomings. We measure the performance of these protocols according to two performance criteria, the estimated worst case and average EER times of tasks, and ....

....too soon, they attempt to improve the schedulability of the tasks at the expense of their average EER times. 3. 1 The Phase Modification (PM) Protocol The Phase Modification protocol, abbreviated as the PM protocol, was initially proposed by Bettati and used to schedule periodic flow shop tasks [11]. Unlike the DS protocol, the PM protocol insists that instances of all subtasks of each task are released periodically according to the period of the task. To ensure that the precedence constraints among subtasks are satisfied, each subtask is given its own phase. The phases of subtasks are ....

R. Bettati. End-to-EndScheduling to Meet Deadlines in DistributedSystems. PhD thesis, Universityof Illinoisat UrbanaChampaign, 1994.

....problem in soft real time systems, where the possibility of a task missing deadlines is minimized but allowed. Consequently, the schedulability of a system is expressed by a statistical value rather than a true false assertion. Bettati addressed all three problems in his Ph.D. dissertation [13]. He proposed a solution for the execution synchronization problem, together with schedulability analysis algorithms, but full exploration of these problems is left open in his thesis. This thesis addresses these open problems and develops an end to end scheduling framework that gives an ....

....We demonstrate that in the context of end to end scheduling the priority assignment problem is NP hard . We then propose several heuristic assignment methods. We present five synchronization protocols that govern the release and the execution of subtasks, including the one proposed by Bettati [13], as solutions to the execution synchronization problem. Several schedulability analysis algorithms are developed for end to end systems that use different priority assignment methods and different synchronization protocols. With these building blocks, the end to end scheduling framework offers a ....

[Article contains additional citation context not shown here]

R. Bettati. End-to-End Scheduling to Meet Deadlines in Distributed Systems. PhD thesis, University of Illinois at Urbana-Champaign, 1994.

....paper, we describe three synchronization protocols. The first protocol is a straightforward implementation to enforce the precedence constraints among subtasks. The second protocol is an extension to the one proposed by Bettati, which was designed for flow shop systems with certain limitations [11]. We propose the third protocol that combines the strength of the previous two protocols while avoiding their shortcomings. We discuss the implications of these protocols with respect to two timing aspects, the estimated worst case and average EER times of tasks, and compare the performance of ....

....too soon, they attempt to improve the schedulability of the tasks at the expense of their average EER times. 3.1. The Phase Modification (PM) Protocol The Phase Modification protocol, abbreviated as the PM protocol, was initially proposed by Bettati and used to schedule periodic flow shop tasks [11]. Unlike the DS protocol, the PM protocol insists that instances of all subtasks are released periodically according to the periods of their parent tasks. To ensure the precedence constraints among subtask are satisfied, each subtask is given its own phase. The phases of subtasks are properly ....

R. Bettati. End-to-End Scheduling to Meet Deadlines in Distributed Systems. PhD thesis, University of Illinois at Urbana-Champaign, 1994.

.... Lee and Messerschmitt do) Our latency analysis is related to the work of Gerber et al. in guaranteeing end to end latency requirements on single processor [10] and to the work of Bettati and Liu in which they analyze the distributed end to end scheduling problem known as the flow shop problem [5]. Our work differs from the former in that we cannot assume a periodic task model, and that our node execution rates are derived from the input data rate and the graph. Moreover, unlike [10] we don t introduce new (additional) tasks for the purpose of synchronization. The dataflow model of PGM ....

....model of PGM does not fit the flow shop problem. We do not assume each node must run on a different processor this is the opposite of how PGM graphs should be distributed. Hence, our end to end latency analysis begins with a different set of assumptions than were used by Bettati and Liu in [5]. 5 Research Plan This section outlines our proposed research by breaking it into four steps. To date, we have concentrated on simple chains of nodes that execute on a single processor. Next we plan to extend the results we have created for simple chains to general graphs executing on a single ....

Bettati, R., Liu, J., "End-to-End Scheduling to Meet Deadlines in Distributed Systems", Proceedings of the 12th International Conference on Distributed Computing Systems, Yokohama, Japan, June 1992, pp. 452-459.

....such suspensions break a task into a sequence of subtasks, each of which may experience a priority inversion. Researchers at the University of Illinois have proposed an end to end approach for sharing resources in multiprocessors, in which tasks are converted into a sequence of periodic subtasks [4, 11]. Each subtask can perform either local computation or a single processor transaction. Subtasks are scheduled on a per processor basis, using uniprocessor scheduling schemes. Resources are accessed through the use of the PCP or a similar scheme. Like the DPCP, the end to end approach requires a ....

R. Bettati, End-to-End Scheduling to Meet Deadlines in Distributed Systems, Ph.D. Thesis, Computer Science Dept., Univ. of Illinois, 1994.

No context found.

R. Bettati and J. W.S. Liu, "End-to-End Scheduling to Meet Deadlines in Distributed Systems". In The 12th International Conference on Distributed Computing Systems, pages 452-459, Yokohama, Japan, June 1992.

....2; 3; n. The last subtask T i;n must complete by the deadline of T i . If task T i is a periodic task, this precedence relation holds for every instance of T i . The precedence relation among the subtasks of each task can be easily satisfied by using the phasemodification method proposed in [5]. Let c i;j be the worst case response time of T i;j . According to the phase modification method, once we know c i;k for k = 1; 2; j Gamma 1, we postpone the phase of the subtask T i;j by P j Gamma1 k=1 c i;k . This modification allows us to enforce the precedence relation between ....

....allows us to enforce the precedence relation between subtasks while treating the subtasks in each task as if there is no precedence relation between them. We will return to discuss how to bound the worst case response times of subtasks on each processor using the schedulability condition in [5], provided that the subtasks are assigned fixed priorities and some single processor synchronization protocol is used to control priority inversion. By summing up the worst case response times of all its subtasks, we can determine the worst case response time of each task, and therefore whether ....

[Article contains additional citation context not shown here]

R. Bettati, "End-to-End Scheduling to Meet Deadlines in Distributed Systems". Ph.D. thesis, Department of Computer Science, University of Illinois at Urbana-Champaign, March 1994.

No context found.

R. Bettati. End-To-End Scheduling To Meet Deadlines In Distributed Systems. PhD thesis, Department of Computer Science, University of Illinois at Urbana-Champaign, 1994.

No context found.

R. Bettati and J. W.-S. Liu. End-to-end scheduling to meet deadlines in distributed systems. In 12th International Conference on Distributed Computing Systems, pages 452-459, Washington, D.C., USA, June 1992. IEEE Computer Society Press.

No context found.

R. Bettati, "End-to-end scheduling to meet deadlines in distributed systems", Ph.D. thesis, Department of Computer Science, University of Illinois at UrbanaChampaign, Urbana, Illinois, USA, March 1994. Page 153

No context found.

R. Bettati. End-to-End Scheduling to Meet Deadlines in Distributed Systems. Ph.D. Thesis, University of Illinois, Urbana, IL, 1994.

No context found.

R. Bettati, End-to-End Scheduling to Meet Deadlines in Distributed Systems, Ph.D. Thesis, Computer Science Department, University of Illinois, March 1994.

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