L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized rate-monotonic scheduling theory: A framework for developing real-time systems. Proceedings of the IEEE, 82(1):68--82, Jan. 1994.  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... arbitrary periods, task executions may get out of phase resulting in large latencies in communications [21] Harmonicity constraints are used to simplify the reading writing logic and reduce the communication latencies [20] Also harmonic periods increase the feasible processor utilization bound [25]. To this end, the period of the consumer is often made to be a multiple of the period of the producer: Figure 5 gives an example of such a task graph. We use this task graph for illustration since it is more general than the tasks of Figure 2. 8 C. Our Goal and Overview of Our ....

L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized rate monotonic scheduling theory: A framework for developing real-time systems. Proceedings of the IEEE, 82(1):68--82, January 1994.

.... fresh inputs. It is not at all obvious that the software architecture meets its deadline requirements. In this section, we perform a formal analysis to check whether this is the case. We use results from real time scheduling theory, in particular, xed priority scheduling theory (e.g. see [8, 7, 5, 14, 1, 2, 15]) and the so called HKL model and analysis [3, 4] First, we cast the platoon application into the formal model. Then, we estimate the execution times and other latencies involved in the system, and compute the total CPU utilization. This is found to be about 74 , that is, less than 1, which is a ....

L. Sha, R. Rajkumar, and S.S. Sathaye. Generalized rate-monotonic scheduling theory: A framework for developing real-time systems. IEEE Proceedings, January 1994.

....premise. At any given time, the highest priority runnable tasks are being executed. This assumes that tasks can be preempted if a higher priority task is ready to be executed. Priority scheduling algorithms differ in how priorities are allocated to tasks. Generalized Rate Monotonic Scheduling [1, 14] (GRM) utilizes static priorities to schedule periodic tasks. The key idea is to assign priorities based on periodic arrival rates where the highest priority is assigned to the task with the shortest period. Dynamic Priority Scheduling is also known as Earliest Deadline First [6] EDF) scheduling. ....

L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized rate-monotonic scheduling theory: A framework for developing real-time systems. Proceedings of the IEEE, 82(1):68--81, 1994.

.... run with arbitrary periods, task executions may get out of phase, which results in large latencies in communication [21] Harmonicity constraints can simplify the reading writing logic and reduce those latencies [20] Harmonic periods may also increase the feasible processor utilization bound [25]. To this end, we assume the period of the consumer is a multiple of that of the related producer. 2.2 Our Goal Given a set of sites and a set of functionally equivalent strategies, our goal is to find a feasible strategy. A strategy is feasible if and only if: within the LCM (Least Common ....

L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized rate monotonic scheduling theory: A framework for developing real-time systems. Proceedings of the IEEE, 82(1):68--82, January 1994.

....communication scheduling in distributed systems and presented an analysis. TTP implements the TDMA (time division multiple access) scheduling strategy. Both contributions assume a periodic activation of processes. Recent extensions of the mentioned work allow periodic activation with jitter, e.g. [15], and arbitrary deadlines and burst [8] for static priority scheduling. Sprunt et al. 16] analyze the influence of sporadic process activation. Unfortunately, all mentioned approaches assume a single coherent scheduling strategy for a given system, whether single or multiprocessor. Very few ....

....scheduled according to the rate monotonic priority assignment [10] while a round robin scheduler alternately assigns CPU 2 to either of its processes. The time slots for P 3 and P 4 are t P 3 ,slot = 5ms and t P 4 ,slot = 3ms, respectively. The core execution time intervals of the processes are [15,17]ms for P 1 , and [8, 11]ms, 10,11]ms, and [3, 5]ms for the processes P 2 , P 3 , and P 4 , respectively. We are looking for conservative (upper and lower) bounds on the response times for each event that is input to the system until a corresponding event is output to the environment. For ....

L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized rate-monotonic scheduling theory: A framework for developing real-time systems. Proceedings of the IEEE, 82(1):68--82, January 1994.

....0.3]t 0.3t t 0.8t Pu [0.6, 1] tl tl 16 tl 16 tl P3 [1.1, 1.2] tl 1. 2tl 8tl Table h Latency Time Intervals and GRMS Param eters To give a simple example how the performance of a system modeled in the SPI model can be validated, we apply a generalized rate monotonic scheduling analysis method [3] to verify if the system is schedulable under the given timing constraints. This method can be applied and the processes can be seen as independent since their communication is nonblocking. The latency path constraint over process P1 can be accounted for as deadline D1 while for the other ....

L. Sha, R. Rajkumar, and S.S. Sathaye. Generalized rate monotonic scheduling theory: A framework for developing real-time systems. Proceedings of the IEEE, 82(1):68-82, January 1994.

....protocol) for communication scheduling in distributed systems and presented an analysis. TTP implements the TDMA (time division multiple access) scheduling strategy. Both contributions assume a periodic activation of processes. Recent extensions allow periodic activation with jitter, e.g. [19], and arbitrary deadlines and burst [11] for static priority scheduling. Sprunt et al. 20] analyze the influence of sporadic process activation. Unfortunately, all mentioned approaches assume a single coherent scheduling strategy for a given system, whether single or multi processor. Very few ....

L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized rate-monotonic scheduling theory: A framework for developing real-time systems. Proceedings of the IEEE, 82(1):68--82, January 1994.

....models, e.g. periodic or burst. From the IP perspective, these can be used to specify operation conditions for which the response times are guaranteed. For single component systems, the resource level already is the system level. There also exist many techniques for homogeneous multi processors [20]. However, this does not hold for heterogeneous multi component systems. The system level analysis for heterogeneous HW SW platforms has been neglected for a long time. In a recent publication, Pop et al. 17] extended the existing scheduling analysis to specialized classes of multi component ....

....for communication scheduling in distributed systems and presented an analysis. TTP implements the TDMA (time division multiple access) scheduling strategy. Both contributions assume a periodic activation of tasks. Recent extensions of the mentioned work allow periodic activation with jitter, e.g. [20], and arbitrary deadlines [12] Sprunt et al. 21] analyze the influence of sporadic task preemption. Tindell presents an approach for task bursts [24] Gresser [7] and Thiele et al. 23] use more general activation models. They introduce a vector of sequential time intervals rather than a few ....

[Article contains additional citation context not shown here]

L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized rate-monotonic scheduling theory: A framework for developing real-time systems. Proceedings of the IEEE, 82(1):68--82, January 1994.

....in aerospace avionics, transportation, multimedia, digital TV, etc, have to serve a mixture of time critical and conventional, interactive or batch computing functions. It is very expensive to build a real time OS starting from the device driver level up to enforce real time scheduling theories [12, 18]. A more cost effective alternative is extending commercial, off the shelf Operating Systems (COTS) for real time service, which could allow the coexistence of time constraint and conventional applications on the same paltform. Middleware is a (mostly) user level software layer, which fills in the ....

Sha, L., Ragunathan, R. and Sathaye, S., "Generalized Rate-Monotonic Scheduling Theory: A Framework for Developing Real-Time Systems," in Proceedings of the IEEE, Vol. 20, No. 1, pp. 68--82, Jan. 1994.

....the system we use the run time scheduler of [16, 18, 19] Previous approaches to real time analysis have focused on software [2] since the performance analysis of ASICs is considered a well studied problem already. Rate Mono tonic Analysis (RMA) 3] and Generalized Rate Monotonic Analysis (GRMA) [4] both assume that tasks are independent and that each task has its own period and deadline. RMA has been extended to account for re lease jitter and resource contention [5, 6] RMA has also been extended to allow precedence among tasks by formulating the problem as a big task with the length of ....

L. Sha, R. Rajkumar and S. Sathaye, "Generalized rate monotonic scheduling theory: a framework for developing real-time systems," Proceedings of the IEEE, 82(1):68-82, January 1994.

....while we can handle any LTL formula. Moreover, only infinite paths can be selected in these works. A Distributed Real TimeSystem: To demonstrate the usefulness of our method, we have applied it to a distributed real time system of realistic complexity, derived from the example described in [27]. Real time systems are used in many critical applications such as aircraft control or medical monitoring systems. Because of the consequences of failures in such systems, determining their correctness is a vital task. Several features of this example make it an interesting target for our ....

....control processor. However, verifying that these deadlines are met using standard techniques is made more difficult because of the distributed nature of the problem. Analytical methods such as the rate monotonic scheduling must impose restrictions on the system, for example, intermediate deadlines [27]. The complex interaction between the various components of the system also makes its analysis using continuous time models unmanageable. Our tools, on the other hand, were able to analyze the system and verify that the deadlines are met by the design. Moreover, we have been able to identify ....

[Article contains additional citation context not shown here]

L. Sha, R. Rajkumar, and S. Sathaye. Generalized rate-monotonic scheduling theory: a framework for developing real-time systems. In Proceedings of the IEEE, Jan 1994.

....by the rates at which the system interacts with the environment through a relatively small number of sensors and actuators. Having each processor run tasks of the same or harmonically related periods has the potential to increase its feasible utilization bound (under fixed priority scheduling) [4]. Moreover, such a task assignment is likely to reduce the least common multiple (LCM) of task periods on each machine (compared to an assignment that does not group tasks by harmonic periods) Hence, it simplifies schedulability analysis and requires a smaller amount of memory space for storing a ....

.... cluster modules with similar periods if they are harmonic multiples (in which case LCM(P i ; P j ) P j ) Unlike ffi 1 , it penalizes module pairs with non harmonic periods (where LCM(P i ; P j ) P j ) Clustering modules of harmonic periods increases the feasible processor utilization bound [4]. ffl ffi 3 = P j =LCM(P i ; P j ) This expression attempts to cluster modules with periods which are harmonic multiples. Unlike ffi 1 and ffi 2 , it does not penalize modules for having significantly different periods as long as they are harmonic. ffl ffi 4 = V ij =P i . This expression ....

[Article contains additional citation context not shown here]

L. Sha, R. Rajkumar, and S. S. Sathaye, "Generalized rate monotonic scheduling theory: A framework for developing real-time systems," Proceeding of IEEE, vol. 82, no. 1, pp. 68--82, Jan 1994.

....Traditional approaches for providing these performance assurances, such as resource reservation [23] The work reported in this paper was supported in part by the National Science Foundation under grants CCR 0093144 and CCR 0098269. and a priori knowledge of worst case execution conditions [27], are no longer applicable. To achieve predictable behavior in distributed, poorly modeled, uncertain environments of today s open performance assured applications, several recent research efforts have suggested the use of control theory [16, 30, 21, 18] This theory offers a new types of ....

L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized rate monotonic scheduling theory: A framework for developing real-time systems. Proceedings of the IEEE, 82(1):68--82, January 1994.

....with a summary of contributions and discussion of future work. 2 Related Work The traditional approach to providing guarantees in performance critical systems has been to rely on static allocation and scheduling algorithms that assume full a priori knowledge of the resource requirements of tasks [5, 31, 32, 40, 41]. The concept of dynamic real time systems [34] pioneered by the Spring kernel project [35, 36] was introduced later to describe applications where runtime workload is unknown until admission control time. It resulted in innovative planning based scheduling algorithms that provide online ....

L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized rate monotonic scheduling theory: A framework for developing real-time systems. Proceedings of the IEEE, 82(1):68--82, January 1994.

....of the application. A widely used way to make schedulability analysis is to use an o ine worst case execution time (WCET) calculation. However, when several processes (or threads) interact via shared data, this calculation often becomes extremely complicated. This problem has been addressed in [9] using a framework where the o ine analysis is extended with some application dependent knowledge and use of priority Basic Research in Computer Science, Centre of the Danish National Research Foundation. 1 inheritants protocol. The work described in this paper is directed towards automatic ....

L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized rate-monotonic scheduling theory: A framework for developing real-time systems. Proceedings of the IEEE, 82(1), January 1994.

....Hence, it s difficult to achieve service differentiation and allocation guarantee using a best effort scheduler. In order to meet the timeliness requirements presented by multimedia applications, much work in recent years has been focusing on applying real time techniques to multimedia systems [50, 36, 70, 44, 79, 22, 10, 39, 80, 73, 72, 56]. However, most of real time schedulers require the global knowledge of the workload and the precise CPU bandwidth requirement of each application. As described in Section 1.2, such information is often hard to obtain in a dynamic environment featuring multimedia applications. Furthermore, ....

....the timeliness requirements presented by multimedia applications, much work in recent years has been focusing on applying real time techniques to multimedia systems. Classical real time scheduling algorithms such as the Earliest Deadline First (EDF) 94 Rate Monotonic (RM) and their variations [50, 36, 70, 44] are suitable for hard real time applications. However, when applying to multimedia applications, several difficulties arise. First, most conventional real time schedulers assume global knowledge of resource requirements (e.g. period, deadline, computational requirements) of a workload to ....

L. Sha, R. Rajkumar, and S.S. Sathaye. Generalized Rate-Monotonic Scheduling Theory: A Framework for Developing Real-Time Systems. In Proceedings of the IEEE, number 1, pages 68--82, January 1994. 117

.... Perhaps the earliest approach to providing guarantees in performance critical systems has been to rely on static allocation and scheduling algorithms that assume full a priori knowledge of the resource requirements of tasks and their arrival times [20, 12, 19, 1] Rate monotonic scheduling theory [11] introduced a second paradigm in which knowledge of task arrival times is not required. As a result, sporadic tasks could be accommodated as long as their minimum inter arrival time is known. The concept of dynamic real time systems [14] pioneered by the Spring kernel project [15, 16] introduced ....

L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized rate monotonic scheduling theory: A framework for developing real-time systems. Proceedings of the IEEE, 82(1):68--82, January 1994.

....errors. Simplex handles timing faults and programming system faults by temporal and spatial encapsulations. Software semantic faults are handled by the use of analytic redundancy. For real time scheduling, the Simplex architecture is based on the generalized rate monotonic scheduling theory [Sha94] Furthermore, the Simplex architecture assumes that the underlying operating system supports either the priority inheritance protocol or the priority ceiling protocol to circumvent priority inversion problems during the management of shared resources. A number of demonstration applications have ....

.... Integration of CORBA Services 9 Composite objects establish timing firewalls [Polze97] between real time and non realtime (CORBA) computing, so that the non real time part cannot violate the real time scheduling rules that are needed by the real time part [Sha94] Implementation details on Composite Objects and timing firewalls using a scheduling server can be found in [Polze97] and [Polze98] Figure 2: A composite system architecture for Simplex In our approach towards a CORBA based version of Simplex shown in Figure 2, we implement decision unit and ....

L.Sha, R.Rajkumar, S.S.Sathaye; "Generalized Rate-Monotonic Scheduling Theory: A Framework for Developing Real-Time Systems"; Proceedings of the IEEE, Vol. 82, No. 1, January 1994.

....with the priority assignment policy. An on line preemptive scheduler produces the execution schedule using the priorities previously assigned. Theory of fixed priority scheduling went through some important developments in recent years. It now supports task models that are much more complex ( 2] [25]) than those supported by early works ( 17] 18] It is common the appearance of precedence relationships among tasks in distributed systems. A precedence relationship is created by a need of synchronization and or transmission of data between two tasks of the application. A message creates ....

....applied to this equivalent system. Most published work use fixed priorities and static task release to implement precedence relations. Arbitrary precedence relations in monoprocessors are considered in [1] and [12] Paper [31] is about arbitrary precedence relations in distributed systems. Works [25] and [27] deal with linear precedence, where each task has at most one predecessor and one successor. It is also possible to implement precedence relations by an explicit message from the predecessor task to the successor task. This message informs that the predecessor task is finished and the ....

L. Sha, R. Rajkumar, S. S. Sathaye. Generalized Rate-Monotonic Scheduling Theory: A Framework for Developing Real-Time Systems. Proceedings of the IEEE, Vol. 82, No. 1, pp. 68-82, january 1994.

.... algorithm will always meet its deadlines, for all processes phasings, if and only if where C j and T j are the execution time and period of task t j respectively and Detailed explanation of the critical zone theorem and illustrations of its application are given elsewhere [Lehoczky et al. 1989] [Sha and Rajkumar 1994]. We use this theorem in the second stage of our task level domain to estimate the likelihood of generating a feasible schedule at the task level. Again, it provides a bound for non preemptive schedules by removing nonpreemption constraints. While it is computationally more expensive to apply the ....

.... time of each task, culminated in a classic rate monotonic scheduling algorithm [Liu and Layland 1973] Consequently, rate monotonic scheduling has been extensively analyzed and generalized, mainly by researchers at Carnegie Mellon university [Lehoczky et al. 1989] Sha and Goodenough 1990] [Sha et al. 1994]. The most notable practical application of real time scheduling approaches, and in particular rate monotonic and generalized rate monotonic scheduling algorithms, include the inclusion of rate monotonic scheduling as the scheduling policy for the IEEE POSIX real time operating system standard and ....

[Article contains additional citation context not shown here]

Sha, L., Rajkumar, R., and Sathaye, S.S. 1994. "Generalized Rate-Monotonic Scheduling Theory: A Framework for Developing Real-Time Systems", Proc. of the IEEE, 82, 1, 68-82.

....where an assignment and schedule are found for replicated tasks such that all deadlines are met in the presence of up to k processor failures. In [57, 105] replicated tasks with precedence constraints are 152 considered. For periodic task sets, the generalized rate monotonic scheduling theory [116] and holistic schedulability analysis [136] have proven to be particularly useful for pre run time analyses. These algorithms were coupled with concurrency control methods such as priority ceiling [52] and dynamic priority ceiling protocols [32] The pre run time resource allocation and ....

L. Sha, R. Rajkumar, and S. S. Sathaye, "Generalized rate monotonic scheduling theory: A framework for developing real-time systems," Proceedings of the IEEE, vol. 82, no. 1, pp. 68--82, January 1994.

....while we can handle any LTL formula. Moreover, only infinite paths can be selected in these works. A Distributed Real Time System: To demonstrate the usefulness of our method, we have applied it to a distributed real time system of realistic complexity, derived from the example described in [28]. Realtime systems are used in many critical applications such as aircraft control or medical monitoring systems. Because of the consequences of failures in such systems, determining their correctness is a vital task. Several features of this example make it an interesting target for our ....

....realistic application, its components are existing systems and protocols that are actually used in many real situations. The example consists of three main components, a FDDI network, a multiprocessor connected to this network and one of the processors in the multiprocessor, the control processor [28]. The FDDI network is a 100Mb s local metropolitan area network that uses a token ring topology [3] It has gained popularity recently, particularly in real time applications, since it allows communication time to be bounded. There are several stations connected to the network in the system. They ....

[Article contains additional citation context not shown here]

L. Sha, R. Rajkumar, and S. Sathaye. Generalized rate-monotonic scheduling theory: a framework for developing real-time systems. In Proceedings of the IEEE, Jan 1994.

....every job releases the semaphores it uses on termination, and that the total cpu utilization is less than 1. 6.2. Feasibility Criterion From the above assumptions and the blocking bounds, we obtain a PVS proof of a well known feasibility test based on computing worst case response times (e.g. [2, 11]) The proof is built in two stages. Let A(p) denote the set of tasks of priority at least p. First, we showed that the sequence u defined as u 0 = B(p) u n 1 = B(p) X i2A(p) C(i) u n T(i) reaches a fixed point M(p) This fixed point is the smallest solution of the equation ....

L. Sha, R. Rajkumar, and S. Sathaye. Generalized RateMonotonic Scheduling Theory: A Framework for Developing Real-Time Systems. Proceedings of the IEEE, 82(1):68-- 82, January 1994.

....meet its deadlines. The approach sketched in Figure 1 (as well as other approaches) generalize to many other task models, including models where tasks can communicate with each other and where deadlines 1 The tasks with lower period are given higher priority. are different from the task periods [5,6]. With such analysis techniques, the theory of real time scheduling, especially in the fixed priority context, is now mature and applicable to real world systems. i H j j k j j i k i i i T W C C W C W 1 0 Figure 1. Computing the Worst Case Response Time of a Task However, ....

....are valid for arbitrary collections of jobs and are the basis for further scheduling analysis. The proofs used the notion of schedules and the various results developed in the support library. We then applied the general results to derive a well known schedulability criterion for sporadic tasks [5,6]. A sporadic task can be thought of as a sequence of jobs of the same priority, separated by a minimal interarrival delay. Sporadic tasks are useful for modeling both strictly periodic computations and interrupt driven activities with a 6 limited interrupt frequency. Knowing bounds on the length ....

Sha, L., R. Rajkumar, and S. Sathaye, 1994, Generalized Rate-Monotonic Scheduling Theory: A Framework for Developing Real-Time Systems, Proceedings of the IEEE, Vol. 82, No. 1, pp. 68-82.

....ring transfer time encompasses solutions to them as well. In this paper, we propose an efficient and low cost real time extension of SCI, called the Job Packing Scheme, for real time message transmission. The algorithm is based on generalized ratemonotonic scheduling theory (GRMS) [17]. However, unlike GRMS, our scheme is a distributed one 2 It is the time between the transmission of the first bit of a packet from the source to the ring and the reception of the last bit of the packet by the target from the ring. 3 Since the distance between the source and the target is ....

....and workstation applications, but it does not have sufficient priority levels to support more involved real time applications. The 8 bit priority protocol implementation is more complex, but provides a relatively complete set of protocols for implementing hardware based ratemonotonic scheduling [17] with limited priority levels per node. 3 The Job Packing Algorithm In this section we outline our algorithm for real time job scheduling in a SCI ring. The algorithm is based on generalized rate monotonic scheduling theory (GRMS) 17] However, unlike GRMS, our scheme is a distributed one that ....

[Article contains additional citation context not shown here]

L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized Rate-Monotonic Scheduling Theory: A Framework for Developing Real-Time Systems. Proceedings of the IEEE, 82(1):68-82, Januany 1994.

....and a new real time service will be created at runtime. Otherwise, the request will be rejected. Thus, composite objects establish timing firewalls [Polze 97] between real time and non real time (CORBA) computing, so that the non real time part cannot violate the real time scheduling rules [Sha 94] that are needed by the real time part. Currently, composite objects is an untested design concept. Although some aspects of the inverted pendulum model problem reflect composite object techniques (e.g. thread priority separation) other aspects of composite objects remain to be demonstrated ....

Sha, L.; Rajkumar, R.; & Sathaye, S. S. "Generalized Rate-Monotonic Scheduling Theory: A Framework for Developing Real-Time Systems." Proceedings of the IEEE 82,1 (January 1994): 68-82.

....the system we use the run time scheduler of [19] Previous approaches to real time analysis have focused on software [1] since the performance analysis of ASICs is considered a well studied problem already. Rate Monotonic Analysis (RMA) 3] and Generalized Rate Monotonic Analysis (GRMA) [4] both assume that tasks are independent and that each task has its own period and deadline. RMA has been extended to account for release jitter and resource contention [5, 6] RMA has also been extended to allow precedence among tasks by formulating the problem as a big task with the length of the ....

L. Sha, R. Rajkumar and S. Sathaye, "Generalized rate monotonic scheduling theory: a framework for developing real-time systems," Proceedings of the IEEE, 82(1):68-82, January 1994.

....policy and its schedulability analysis under hard real time constraints. The static scheduling policy represented by rate monotonic(RM) and the dynamic scheduling by earliest deadline first(EDF) have been extended to accommodate more realistic assumptions than those given by Liu and Layland[6] [8, 13]) and the analysis of algorithms for periodic tasks has been performed to incorporate sporadic or aperiodic tasks[10, 2] The real time tasks in these works have hard deadlines of which the miss can be catastrophic. In contrast, in a multimedia application which is a new area of real time ....

L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized ratemonotonic scheduling theory: A framework for developing real-time systems. Proc. of the IEEE, 82(1):68--82, 1994.

....of the application. A widely used way to make schedulability analysis is to use an offline worst case execution time (WCET) calculation. However, when several processes (or threads) interact via shared data, this calculation often becomes extremely complicated. This problem has been addressed in [9] using a framework where the offline analysis is extended with some application dependent knowledge and use of priority inheritants protocol. The work described in this paper is directed towards automatic collection of application dependent knowledge, resulting in less manual (and error prone) ....

L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized ratemonotonic scheduling theory: A framework for developing real-time systems. Proceedings of the IEEE, 82(1), January 1994.

.... Perhaps the earliest approach to providing guarantees in performance critical systems has been to rely on static allocation and scheduling algorithms that assume full a priori knowledge of the resource requirements of tasks and their arrival times [4, 32, 43, 44] Rate monotonic scheduling theory [31] introduced a second paradigm in which knowledge of task arrival times is not required. As a result, sporadic tasks could be accommodated as long as their minimum inter arrival time is known. The concept of dynamic real time systems [35] pioneered by the Spring kernel project [36, 37] introduced ....

L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized rate monotonic scheduling theory: A framework for developing real-time systems. Proceedings of the IEEE, 82(1):68--82, January 1994. 22

....the system we use the run time scheduler of [16, 18, 19] Previous approaches to real time analysis have focused on software [2] since the performance analysis of ASICs is considered a well studied problem already. Rate Monotonic Analysis (RMA) 3] and Generalized Rate Monotonic Analysis (GRMA) [4] both assume that tasks are independent and that each task has its own period and deadline. RMA has been extended to account for release jitter and resource contention [5, 6] RMA has also been extended to allow precedence among tasks by formulating the problem as a big task with the length of the ....

L. Sha, R. Rajkumar and S. Sathaye, \Generalized rate monotonic scheduling theory: a framework for developing real-time systems," Proceedings of the IEEE, 82(1):68-82, January 1994.

....fixed period, whereas for the aperiodic task a minimum interarrival time may be specified. Each release of a task is considered to request the worst case amount of computation capacity from the processor. This model of periodic tasks is the basis for rate monotonic analysis (Liu and Layland, 1973; Sha et al. 1994) and its extensions to accommodate aperiodic tasks e.g. Sprunt et al. 1989) However, there have been several extensions of this coarse classification in order to capture more detailed a priori knowledge about a certain task: The Hyperperiodic Task Model (Aggarwall and Chraibi, 1993) allows a ....

Sha, Lui, Ragunathan Rajkumar and Shirish S. Sathaye (1994). Generalized rate-monotonic scheduling theory: A framework for developing real-time systems. Proceedings of the IEEE 82(1), 68--82.

.... timing constraints on periodicity, arrival and required time of each task, were culminated in a classic rate monotonic scheduling algorithm [25] Consequently, rate monotonic scheduling has been extensively analyzed and generalized, mainly by researchers at Carnegie Mellon University [24] 43] [45]. The most notable practical application of real time scheduling approaches, in particular rate monotonic and generalized rate monotonic scheduling algorithms, include the inclusion of rate monotonic scheduling as the scheduling policy for the IEEE POSIX real time operating system standard and ....

.... the IEEE POSIX real time operating system standard and IEEE Futurebus standards [19] 20] 44] and use of the generalized rate monotonic scheduling techniques in several major advanced technology projects such as the Space 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 ....

L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized rate-monotonic scheduling theory: A framework for developing real-time systems. Proc. of the IEEE, 82(1):68--82, January 1994.

....used. The goal of this separation is to allow use of a number of admission control policies to ensure timeliness while leaving consistency to the multicast algorithm. For example, instead of the schedulability condition stated in Theorem 1, one may use the generalized rate monotonic analysis [12], FDDI synchronous bandwidth allocation analysis [13] or delay analysis for Controller Area Networks [14] depending on the application at hand. ACSA Retransmission ClockSync lower protocol layers lower protocol layers Application Level Minimal Configuration. b) Full Configuration (a) ....

L. Sha, R. Rajkumar, and S. S. Sathaye, "Generalized rate monotonic scheduling theory: A framework for developing real-time systems," Proceedings of the IEEE, vol. 82, no. 1, pp. 68--82, January 1994.

.... The task is feasible if D(i) 6 M(i) This general approach to determining feasibility by computing response times was initiated by Joseph and Pandya [17] and by Harter [16] Algorithms for computing M(i) have been proposed for di erent classes of jobs and with di erent assumptions about deadlines [3,5,6,15,35,37]. In case, D(i) is no more than T(i) and all the tasks have di erent priorities, M(i) is the smallest solution of the following equation: C(i) B(p) X l2H(i) C(l) M(i) T(l) M(i) 5.1) where p is the priority of task i and H(i) is the set of tasks of higher priority than i. 3 ....

L. Sha, R. Rajkumar, and S. Sathaye. Generalized Rate-Monotonic Scheduling Theory: A Framework for Developing Real-Time Systems. Proceedings of the IEEE, 82(1):68-82, January 1994.

....when sufficient CPU resources are available; in overload conditions, it can ensure that all processes fairly miss their deadlines. In the case of real time processes it is clear that sharing processes fairly is inappropriate. The real time community have looked at realtime scheduling in depth [1,2,5,7,11]. A variety of real time scheduling algorithms and their associated schedulability tests have been developed to handle real time scheduling under different process constraints. Many of these algorithms can be successfully applied for scheduling real time processes within conventional operating ....

....of the more important scheduling design issues. 4.1 Scheduling Algorithm The Dreams system was designed to allow various scheduling algorithms and corresponding schedulability tests to be trialed. Most traditional real time systems use Rate Monotonic Scheduling (RMS) 2] and its derivatives [7]. The basic RMS approach is to always schedule the runnable process with the shortest period. Usually this is implemented by assigning a fixed priority to each periodic process based on the period of the process; the shorter the period, the higher the priority of the process. A conventional ....

L. Sha, R. Rajkumar, and S.S. Sathaye. Generalized Rate-Monotonic Scheduling Theory: A Framework for Developing RealTime Systems. Proceedings of the IEEE 82(1): 68-82, 1994.

....alternative program units. Because it does not require rollback, analytic redundancy is an example of a forward recovery method. 3.3 Rate Monotonic Scheduling Theory For real time systems, we also have to be able to make performance guarantees. Generalized Rate Monotonic Scheduling (GRMS) Theory [Sha 94] guarantees that, as long as certain conditions are met (e.g. task execution time, strict adherence to priority, etc. a given task set can be guaranteed to meet its deadlines. Rate Monotonic Analysis (RMA) techniques coupled with the appropriate real time operating system scheduling support ....

Sha, L.; Rajkumar, R.; & Sathaye, S. "Generalized Rate Monotonic Scheduling Theory: A Framework for Developing Real-time Systems, " Proceedings of the IEEE, January 1994.

....reservation model is suitable for a wide range of applications, but applications that require different reservation semantics might need a model that takes advantage of other scheduling algorithms and analyses. For example, an application might require generalized rate monotonic analysis [39,64,107,109], aperiodic servers with different replenishment policies [111,112,120] earliest deadline first scheduling [27,49,67] sporadic task scheduling [51] or deadline monotonic scheduling [7] The reservation system also depends of priority inheritance protocols for fixed priority scheduling ....

L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized Rate-Monotonic Scheduling Theory: A Framework for Developing Real-Time Systems. Proceedings of the IEEE, 82(1):68--82, January 1994.

....used to decouple real time and nonreal time processing. A composite object establishes a timing firewall [Polze97] between the real time and nonreal time (CORBA) computing part, so that the non real time part cannot violate the real time scheduling rules that are needed by the real time part [Sha94]. Implementation details on Composite Objects and timing firewalls using a scheduling server can be found in [Polze97] and [Polze98] In our approach towards a CORBA based version of Simplex, we implement decision unit and safety controller inside the real time part of composite objects. We use ....

L.Sha, R.Rajkumar, S.S.Sathaye; "Generalized Rate-Monotonic Scheduling Theory: A Framework for Developing Real-Time Systems"; Proceedings of the IEEE, Vol. 82, No. 1, January 1994.

....modification tools, real time process management and communication services, and the fault tolerant middleware that lies between the application code and the real time operating system. Multiple real time processes are scheduled to run simultaneously by the Rate Monotonic Algorithm (RMA) [5]. Each of the processes is a runtime instance of a software module and can be facilitated with a replacement unit, the basic building block of the Simplex architecture. Replacement units are so designed that the unit processes and their connections to the other processes can be added and deleted ....

....modules at runtime, and therefore enables runtime insertion removal of the experimental controller which is implemented as a replacement unit process. Rate monotonic scheduling guarantees a software module can be replaced or modified in real time while other processes still meet their deadlines [5]. The inter process communication is realized by the real time publishers and subscribers facility [3] which enables the processes to dynamically publish and subscribe needed information to each other. The key feature of the architecture is to allow the upgrades, which are implemented as software ....

[Article contains additional citation context not shown here]

L. Sha, R. Rajkumar, and S. Sathaye, "Generalized Rate Monotonic Scheduling Theory: A Framework of Developing Real-Time Systems". In the IEEE Proceedings, January, 1994.

No context found.

L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized rate-monotonic scheduling theory: A framework for developing real-time systems. Proceedings of the IEEE, 82(1):68--82, Jan. 1994.

No context found.

L. Sha, R. Rajkumar, S. S. Sathaye, Generalized Rate Monotonic Scheduling Theory: A Framework for Developing Real-time Systems, Proceedings of the IEEE 82 (1).

No context found.

L.Sha, R.Rajkumar, S.S.Sathaye, \Generalized Rate Monotonic Scheduling Theory: a Framework for Developing Real Time Systems," IEEE Proceedings, Vol.82, No.1, 1994.

No context found.

L.Sha, R.Rajkumar, S.S.Sathaye, \Generalized Rate Monotonic Scheduling Theory: a Framework for Developing Real Time Systems," Proceedings of the IEEE, Vol.82, No.1, Jan.1994.

No context found.

L. Sha, R. Rajkumar and S. Sathaye, "Generalized rate monotonic scheduling theory: a framework for developing real-time systems," Proceedings of the IEEE, 82(1):68-82, January 1994.

No context found.

L. Sha, R. Rajkumar, and S. Sathaye. Generalized rate monotonic scheduling theory, a framework of developing real-time systems. Proceedings of The IEEE, January 1994.

No context found.

L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized RateMonotonic Scheduling Theory: A Framework for Developing Real-Time Systems. Proceedings of the IEEE, 82(1):68-- 82, January 1994.

No context found.

L. Sha, R. Rajkumar, and S. S. Sathaye. Generalized rate monotonic scheduling theory: A framework for developing real-time systems. Proceedings of the IEEE, 82(1):68--82, January 1994.

No context found.

Sha, L., Rajkumar, R. and Sathaye, S.S. "Generalized Rate-Monotonic Scheduling Theory: A Framework for Developing Real-Time Systems" Proceedings of the IEEE 82 (1), January 1994.

No context found.

L. Sha, R. Rajkumar, S. Sathaye, Generalized Rate Monotonic Scheduling Theory: A Framework for Developing Real-Time Systems. Proc. of the IEEE, Vol 82, No. 1, Jan. 1994.

First 50 documents  Next 50

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