Showing papers on "Scheduling (computing) published in 1980"

Linear Multiple Objective Problems with Interval Coefficients

Abstract: In this paper we consider linear multiple objective programs with coefficients of the criteria given by intervals. This class of problems is of practical interest since in many instances it is difficult to determine precisely the coefficients of the objective functions. A subproblem to test if a feasible extreme point is efficient in the problem considered is obtained. A branch and bound algorithm to solve the subproblem as well as computational results are provided. Extensions are discussed.

A Characterization of Waiting Time Performance Realizable by Single-Server Queues

Abstract: In this paper we study the problem of designing scheduling strategies when the demand on the system is known and waiting time requirements are pre-specified. This important synthesis problem has received little attention in the literature, and contrasts with the common analytical approach to the study of service systems. This latter approach contributes only indirectly to the problem of finding satisfactory scheduling rules when the desired (or required) response-time performance is known in advance. Briefly, the model studied assumes a Markov queueing system with M (priority) classes of jobs. For each class, a desired mean waiting time is given in advance. Making use of a well known conservation law, we prove a necessary and sufficient condition for the existence of a scheduling strategy that achieves the desired performance. We also give a constructive procedure for checking the condition and, if a solution exists, a procedure for finding one such strategy. Our assumptions are discussed and the possibil...

Scheduling Tasks with Exponential Service Times on Nonidentical Processors to Minimize Various Cost Functions.

Abstract: : We consider preemptive scheduling of N tasks on m processors; processors have different speeds, tasks require amounts of work which are exponentially distributed, with different parameters. The policies of assigning at every moment the task with shortest (longest) expected processing time among those not yet completed to the fastest processor available, 2nd shortest (longest) to the 2nd fastest etc., are examined, and shown to minimize expected values of various cost functions. As special cases we obtain policies which minimize expected flowtime, expected makespan and expected lifetime of a series system with m component locations and N spares. (Author)

Scheduling Subject To Resource Constraints: Classification And Complexity

Abstract: In deterministic sequencing and scheduling problems, jobs are to be processed on machines of limited capacity. We consider an extension of this class of problems, in which the jobs require the use of additional scarce resources during their execution. A classification scheme for resource constraints is proposed and the computational complexity of the extended problem class is investigated in terms of this classification. Models involving parallel machines, unit-time jobs and the maximum completion time criterion are studied in detail; other models are briefly discussed.

Scheduling of Network Television Programs

Abstract: Maximizing average audience size is a major objective of national television networks. This paper examines the effect of scheduling programs on audience ratings. We determine the predictability of ratings and the major factors which influence ratings. We measure the extent to which ratings can be changed by rescheduling programs, and then test commonly discussed scheduling strategies and propose new strategies. Finally, we ascertain the extent to which the audience's benefit is served by a system of three major networks, each competing to maximize individual ratings. A forecasting model based on five years of historical data estimates ratings for hypothetical schedules. A heuristic fitting procedure yields a model which has consistent estimated parameters and explains about 70% of the variance. The previous year's ratings of a program and competing programs in its time slot are the best predictors of its rating. Also important are the day, time, and lead-in audience. The resulting estimates are incorporated in a combinatorial model, representing the decision of a program scheduler seeking to maximize ratings. A typical sized problem of scheduling 23 programs of various lengths into a week of 40 half-hour time slots can be formulated as a 65 × 800 assignment-type integer program. The optimizations show that rescheduling programs can, on the average, substantially increase a network's audience size 11.6% and advertising revenue $61 million per year. The optimal program scheduling for one network generally decreases the ratings of competing networks. The forecasting and optimization models are then used to evaluate commonly discussed scheduling strategies: Protecting Newcomers, Starting Fast, Homogeneity, Counterprogramming, and Bridging. Of these, Counterprogramming is the only strategy that is presently and optimally used and justified by the model. An additional strategy, termed "Avoidance," is discovered and theoretically justified by subadditivity in ratings. This strategy increases total audience size and benefits the viewing audience.

Scheduler for a multiple computer system

Abstract: A scheduler for selecting and scheduling the tasks to be executed by a computer in a multiple computer system is disclosed. One scheduler is associated with each computer, and the schedulers coordinate their operation by sending and receiving messages. Each scheduler comprises a status table (604) storing the status of each task assigned to its computer, and a scheduling status table (608) storing the tasks recently selected for execution by the computer. The scheduler further includes a record data ready module (600) which records in the status table (604) the reception of the data variables required for the execution of each task. A completed task recorder (612) records which tasks have been executed by itself or any other computer in the system. An unselected/selected task recorder records the selection and unselection of tasks by other computers. A task unselector records the tasks which have been unselected by itself. A task selector (610) selects from the status table and records in the scheduling status table the highest priority task ready for execution which has not been selected by any computer in the system. A task releaser (618) forwards the selected task for execution each time the computer signals the completion of the preceding scheduled task.

Deterministic Scheduling with Pipelined Processors

Abstract: In this paper several results are proven showing a correspondence between problems involving task systems with single-stage processors and similar ones using pipelined processors. For example, an optimal schedule for a task system with arbitrary precedence constraints using a single pipelined processor with two stages is found by the familiar Coffman–Graham algorithm for a system with two single-stage processors. Precedence constraints in the form of inforests, out-forests, and directed acyclic graphs are examined. Task systems with release times and deadlines for each task are also considered.

Scheduling to minimize makespan on unequal parallel processors

Abstract: The problem of scheduling n independent jobs on m unequal parallel processors to minimize makespan is known to be very difficult for the nonpreemptive case. Until recently, research on this problem has consisted primarily of investigating the “worst case” behavior for simple approximation algorithms for the special cases of equal or uniform processors. Recently, Ibarra and Kim [10] have completed a similar analysis on five simple heuristics for general unequal processors. The approach in the present paper is to investigate the “average” behavior via large-scale computational testing. A new heuristic is devised. It is tested on a large number of problems (about 9,000) for both uniform and general processors, and the results are compared with the best solutions obtained from a truncated branch-and-bound procedure. It is observed that, although the heuristic takes negligible time compared to the branch-and-bound procedure, on the average it is within 5 percent of the latter in the uniform case and within 1.3 percent in the general case. It also performs better than a composite heuristic whose output, for any given problem, is the best of all previously suggested heuristics.

Scheduling Jobs Subject to Nonhomogeneous Poisson Shocks

Abstract: Consider n tasks which must be attempted in some order. To successfully complete any given task requires a random amount of time. However, we suppose that there are external events, called shocks, which occur according to a nonhomogeneous Poisson process. If no shocks occur while a task is being performed, then that task is considered a success. If a shock does occur, then work on that task ends and work on the next one begins. We consider such problems as maximizing The expected number of successful tasks; The length of time until no tasks remain; The expected total reward earned; where a reward R1 is obtained upon successful completion of task i.

An Interactive Stock Control System with a Strategic Management Role

Abstract: This note describes the design, implementation and consequences of a stock control system employing interactive mini computers for a number of builders and plumbers merchants.

On the interactive solution to a multicriteria scheduling problem

Abstract: A mixed integer multiple criteria model is formulated for scheduling problems. Its solution is obtained by an interactive method based on the Tchebycheff-approximation. For solving greater problems some modifications are discussed at the end of the paper, which are essentially based on the notion of “heuristic efficiency”.

Bounds on the Scheduling of Typed Task Systems

Abstract: We study the scheduling of different types of tasks on different types of processors. If there are k types of tasks and $m_i $, identical processors for tasks of type i, the finishing time of any list schedule is at most $k + 1 - (1/ \max (m_1 , \cdots ,m_k ))$ times worse than the optimal schedule. This bound is best possible. If the processors execute at different speeds then the performance of any list schedule (relative to the optimal schedule) is bounded by k plus the maximum ratio between the speeds of any two processors of the same type.

Subsystems of Processes with Deadlock Avoidance

Abstract: A graph-oriented approach to deadlock avoidance, which treates both shared and exclusive locking, has been described [6]. The method is particularly suited for database systems. With enhancements introduced here, the problem of indefimite delay, i.e., the possibility that a process will not run to completion (will be delayed indefinitely) can be eliminated. The approach taken is to partition the resource system into subsystems, each of which can be scheduled independently. Indefinite delay is avoided by the construction of subsystems that guarantee the completion of a process or the granting of a resource request. Further, we show how the subsystem approach can be applied systematically so as to approximate FIFO scheduling of resource requests, while avoiding deadlock and indefmite delay. Other scheduling disciplines can also be realized. A lock manager program utilizing the FIFO method has been implemented.

Prevention of task overruns in real-time non-preemptive multiprogramming systems

Abstract: Real-time multiprogramming systems, in which a hardware processor is dynamically assigned to run multiple software processes each designed to control an important device (user), are considered. Each software process executes a task in response to a service request repeatedly coming from the corresponding user. Each service task is associated with a strict deadline, and thus the design problem that we are concerned with is to ensure that the service tasks requested can always be executed within the associated deadlines, i.e., no task overrun occurs. This problem was studied by several investigators for the cases where preemptive scheduling strategies are used. In contrast, very few studies have been conducted for cases of non-preemptive scheduling. In this paper we show that a non-preemptive strategy, called relative urgency non-preemptive (RUNP) strategy, is optimal in the sense that if a system runs without a task overrun under any non-preemptive strategy, it will also run without a task overrun under the RUNP strategy. Then an efficient procedure used at the design time for detecting the possibility of a task overrun in a system using the RUNP strategy is presented. The procedure is useful in designing overrun-free real-time multiprogramming systems that yield high processor utilizations. Some special types of systems using the RUNP strategy for which even simpler detection procedures are available are also discussed.

A general review of the use of computers in scheduling buses and their crews

Abstract: The paper examines a variety of methods that have been applied to bus and crew scheduling. Most implementations described have been achieved using heuristic methods, although there are indications that mathematical programming will play an increasing role in such applications. It is also likely that in future such schedules will be compiled through an interactive process involving the scheduler and a computer. From the examples of implemented scheduling systems described, it is thought that more use should be made of such systems within bus operating bodies. There appears to be a number of examples of their use in the USA with only isolated examples in Europe although in the UK several methods have been used successfully at a number of locations but only on an on-off basis. For the covering abstract see UMTRIS 377151.

An Analysis of Preemptive Multiprocessor Job Scheduling

Abstract: The preemptive scheduling of a partially ordered set of tasks is studied. A class of scheduling heuristics is introduced, and the performance of schedules in this class is analyzed with respect to the least finishing time optimality criterion. If there are m processors, then the finishing time of any schedule in the class is at most $\sqrt{m} + 1/2$ times worse than optimal, independent of the speeds of the processors. Examples are given which indicate that there are schedules which may be as bad as $\sqrt{m-1}$ times worse than optimal even for machines with one fast processor.

A Model for Advanced Reservations for Large Scale Conferencing Services

Abstract: This note describes a model for advanced reservations extended to systems which can serve many customers at the same time. The new model has been developed for a communication network which provides intercity visual conferencing services via a satellite capable of accommodating many conferences simultaneously. Each customer calls in advance to reserve time on a future date. A scheduler either satisfies the request, or offers available alternatives which may or may not be acceptable to the customer. The model is used to examine different measures of quality-of-service and system utilization.

A Heuristic Scheduling Algorithm

Abstract: This paper describes a heuristic algorithm developed to schedule a group of individuals such that every person performs each of the different activities they desire at some point during the time-frame of the schedule and the difference between the exogenously given number of people desired at each available location-activity-period position and those allocated to these positions is minimized. The contribution of the present work is in the formulation of the problem, and the resulting ease with which good solutions to large-scale problems can be generated, rather than in the mechanics of the algorithm itself. The mathematic formulation of the scheduling problem is presented first, and subsequently, the solution strategy is elaborated. Experimental results on some reasonably large problems are also presented.

Fundamental Scheme for Train Scheduling (Application of Range-Constriction Search).

Abstract: : Traditionally, the compilation of long-term timetables for high density rail service with multiple classes of trains of the same track is a job for expert people, not computers. We propose an algorithm that uses the range-constriction search technique to schedule the timing and pass-through relations of trains smoothly and efficiently. The program determines how the timing of certain trains constrains and timing of others, finds possible time regions and pass-through relations, and then evaluates the efficiency of train movement for each pass-through relation. (Author)

Scheduling Jobs on Two Facilities to Minimize Makespan

Abstract: This paper is concerned with the problem of scheduling tasks on a system consisting of two parallel processors in order that makespan be minimized. In particular, we treat a variety of modifications to this basic theme, including the cases of identical, proportional, and unrelated processors. In addition, a heuristic scheme is suggested when precedence constraints exist where such constraints may be assignment dependent. Substantial computational experience is reported in all cases where it would appear that relatively large problems can be handled routinely.

Control-theoretic formulation of operating systems resource management policies

Abstract: : In this thesis we propose the following general control-theoretic approach to the formulation of resource management policies for operating systems. (1) In order to develop a resource management policy, model the corresponding program behavior as a stochastic process. (2) Using identification techniques and empirical data, identify a suitable model structure for the process and estimate typical values of model parameters. (3) Based on the model, formulate a prediction strategy for the stochastic process, and hence a resource management policy. The policy so obtained is dynamic in the sense that it varies the allocation of the system resource to a user job depending upon the recent past behavior of the job. It, thus, provides the run time optimization not possible with the queueing theory approach. Also, notice that the individuality of the job is fully exploited. The key step in the approach is the formulation of the stochastic process model in such a way that the allocation problem reduces to a prediction problem. We exemplify this approach by formulating control-theoretic policies for CPU scheduling and page replacement. Policies for allocation of other shared resources (e.g., disks) can be, similarly, formulated. (Author)

Job Scheduling in a Single-Node Hierarchical Network for Process Control

Abstract: The operating characteristics of a typical single-node hierarchical element in a multiprocessor network such as might be used for process control are described, and the class of scheduling problems arising from the use of such a network is discussed and related to similar problems which have been reported in the literature. Two measures of system performance are then defined: the "settling time," defined as the maximum of the finishing times of all jobs when a set of jobs is initiated once in the network, and the "steady-state service," a related performance index for the case where jobs are initiated on a periodic basis. Finally, a simulation model is used to evaluate the operating characteristics of the hierarchical network using the two metrics in conjunction with several synthetic workloads; a general optimization procedure is shown to lead to values of scheduling parameters which are close to optimal even for network workloads with many degrees of freedom (i.e., when many computers are initiating jobs).