Task assignment in loosely‐coupled multiprocessor systems

Abstract

In loosely‐coupled multiprocessor systems, a parallel program has its modules distributedly assigned among the processors. The assignment policy is to minimize interprocessor communication cost and to balance the workload of processors. However, there exists conflict between these two criteria and a compromise must be made to obtain an optimal solution. In this paper, we propose a new task assignment model for distributed computing systems, and for solving the assignment problem based on partitioning graphs. The problem of finding an optimal solution had been shown to be in the class of NP‐complete. For the sake of computation efficiency, we propose some heuristics for obtaining suboptimal solutions.     

Heuristics for the generalised assignment problem: simulated annealing and tabu search approaches

The generalised assignment problem (GAP) is the problem of finding a minimum cost assignment of a set of jobs to a set of agents. Each job is assigned to exactly one agent. The total demands of all jobs assigned to any agent can not exceed the total resources available to that agent. A review of exact and heuristic methods is presented. A-generation mechanism is introduced. Different search strategies and parameter settings are investigated for the-generation descent, hybrid simulated annealing/tabu search and tabu search heuristic methods. The developed methods incorporate a number of features that have proven useful for obtaining optimal and near optimal solutions. The effectiveness of our approaches is established by comparing their performance in terms of solution quality and computional requirement to other specialized branch-and-bound tree search, simulated annealing and set partitioning heuristics on a set of standard problems from the literature.     

A two-stage three-machine assembly scheduling problem with a position-based learning effect

The two-stage assembly scheduling problem has attracted increasing research attention. In many such problems, job processing times are commonly assumed to be fixed. However, this assumption does not hold in many real production situations. In fact, processing times usually decrease steadily when the same task is performed repeatedly. Therefore, in this study, we investigated a two-stage assembly position-based learning scheduling problem with two machines in the first stage and an assembly machine in the second stage. The objective was to complete all jobs as soon as possible (or to minimise the makespan, implying that the system can perform better and efficient task planning with limited resources). Because this problem is NP-hard, we derived some dominance relations and a lower bound for the branch-and-bound method for finding the optimal solution. We also propose three heuristics, three versions of the simulated annealing (SA) algorithm, and three versions of cloud theory-based simulated annealing algorithm for determining near-optimal solutions. Finally, we report the performance levels of the proposed algorithms.     

Mathematical model and metaheuristics for simultaneous balancing and sequencing of a robotic mixed-model assembly line

This article presents the first method to simultaneously balance and sequence robotic mixed-model assembly lines (RMALB/S), which involves three sub-problems: task assignment, model sequencing and robot allocation. A new mixed-integer programming model is developed to minimize makespan and, using CPLEX solver, small-size problems are solved for optimality. Two metaheuristics, the restarted simulated annealing algorithm and co-evolutionary algorithm, are developed and improved to address this NP-hard problem. The restarted simulated annealing method replaces the current temperature with a new temperature to restart the search process. The co-evolutionary method uses a restart mechanism to generate a new population by modifying several vectors simultaneously. The proposed algorithms are tested on a set of benchmark problems and compared with five other high-performing metaheuristics. The proposed algorithms outperform their original editions and the benchmarked methods. The proposed algorithms are able to solve the balancing and sequencing problem of a robotic mixed-model assembly line effectively and efficiently.     

Two-level modified simulated annealing based approach for solving facility layout problem

In this paper, we are considering the quadratic assignment model (QAP) of the facility layout problem (FLP) which is known to be NP-hard. We relax the integer constraints of the QAP and solve it on a commercially available package called LINGO 8. In the optimal solution so obtained, X_ij s take real values between zero and one. We identify promising X_ij s having a value strictly greater than 0.5 in the optimal solution and set them to one. We add the constraints (X_ij ?=?1) associated with promising X_ij s into the QAP (with integer restrictions) and resolve using LINGO 8. In all the cases attempted we obtained a superior feasible solution to the QAP which was further improved by the proposed modified simulated annealing (MSA) procedure. An encouraging comparative performance of this procedure is thus reported.     

Effective hybrid genetic algorithm for minimizing makespan on a single-batch-processing machine with non-identical job sizes

The paper addresses minimizing makespan by a genetic algorithm (GA) for scheduling jobs with non-identical sizes on a single-batch-processing machine. A batch-processing machine can process up to B jobs simultaneously. The processing time of a batch is equal to the longest processing time among all jobs in the batch. Two different GAs are proposed based on different encoding schemes. The first is a sequence-based GA (SGA) that generates random sequences of jobs using GA operators and applies the batch first fit heuristic to group the jobs. The second is a batch-based hybrid GA (BHGA) that generates random batches of jobs using GA operators and ensures feasibility by using knowledge of the problem based on a heuristic procedure. A greedy local search heuristic based on the problem characteristics is hybridized with a BHGA that has the ability of steering efficiently the search toward the optimal or near-optimal schedules. The performance of proposed GAs is compared with a simulated annealing (SA) approach proposed by Melouk et al. (Melouk, S., Damodaran, P. and Chang, P.Y., Minimizing makespan for single machine batch processing with non-identical job sizes using simulated annealing. Int. J. Prod. Econ., 2004, 87, 141–147) and also against a modified lower bound proposed for the problem. Computational results show that BHGA performs considerably well compared with the modified lower bound and significantly outperforms the SGA and SA in terms of both quality of solutions and required runtimes.     

A duplication heuristic for static scheduling of tasks on distributed memory multiprocessors

Abstract

A task duplication heuristic, DSH, was proposed in [11]. The underlying concept of the task duplication heuristic is duplicating some tasks on processors such that the earliest start time of tasks on processors can be reduced, that is, tasks on processors can be executed sooner. This leads to a better scheduling length. In this paper, we propose a more general task duplication heuristic, bottom‐up top‐down duplication heuristic (BTDH), for static scheduling of directed‐acyclic graphs (DAGs) on distributed memory multiprocessors. The key difference between BTDH and DSH is the method used for duplicating tasks. BTDH allows tasks to be duplicated on processors even though the duplication of tasks will temporarily increase the earliest start time of some tasks. DSH only allows those duplications which will reduce the earliest start time of tasks. Simulation results show that, for coarse‐grain DAGs, the scheduling length of BTDH is almost the same as the scheduling length of DSH. However, for medium‐grain and fine‐grain DAGs, BTDH produces better scheduling length than DSH.     

Controlling asynchronous team design environments by simulated annealing

An organizational strategy for design environments,asynchronous teams, is reviewed. Simulated annealing is used to implement the necessarycontracting search behavior of asynchronous teams. An example of an asynchronous team design environment controlled by simulated annealing is given from the building design domain. The simulated annealing algorithm used, which has been modified for distributed use and multi-criteria, non-preference objectives, is described.     

Multicriteria airport gate assignment and Pareto simulated annealing

This paper addresses an airport gate assignment problem with multiple objectives. The objectives are to minimize the number of ungated flights and the total passenger walking distances or connection times as well as to maximize the total gate assignment preferences. The problem examined is an integer program with multiple objectives (one of them being quadratic) and quadratic constraints. Of course, such a problem is inherently difficult to solve. We tackle the problem by Pareto simulated annealing in order to get a representative approximation for the Pareto front. Results of computational experiments are presented. To the best of our knowledge, this is the first attempt to consider the airport gate assignment problem with multiple objectives.     

PPT: A parallel programming tool for distributed memory multiprocessors

Abstract

Traditionally, to program a distributed memory multiprocessor, a programmer is responsible for partitioning an application program into modules or tasks, scheduling tasks on processors, inserting communication primitives, and generating parallel codes for each processor manually. As both the number of processors and the complexity of problems to be solved increases, programming distributed memory multiprocessors becomes difficult and error‐prone. In a distributed memory multiprocessor, the program partitioning and scheduling play an important role in the performance of a parallel program. However, how to find the best program partitioning and scheduling so that the best performance of a parallel program on a distributed memory multiprocessor can be achieved, is not an easy task. In this paper, we present a parallel programming tool, PPT, to aid programmers to find the best program partitioning and scheduling and automatically generate the parallel code for the single program multiple data (SPMD) model on a distributed memory multiprocessor. An example of designing a parallel FFT program by using PPT on an NCUBE‐2 is also presented.     

Performance of high‐resolution TVD schemes for 1D dam‐break simulations

Abstract

The performance of high‐resolution total variation diminishing (TVD) schemes for simulating dam‐break problems are presented and evaluated. Three robust and reliable first‐order upwind schemes, namely FVS, Roe and HLLE schemes, are extended to six second‐order TVD schemes using two different approaches, the Sweby flux limiter approach and the direct MUSCL‐Hancock slope limiter. For idealized dam‐break flows, comparisons of the simulated results with the exact solutions show that the flux vector splitting (FVS) scheme coupled with the direct MUSCL‐Hancock (DMH) slope limiter approach has the best numerical performance among the presented schemes. Application of the FVS‐DMH scheme to a dam‐break experiment with sloping dry bed shows that the simulated water depths agree well with the measured.     

An efficient coarse‐grained parallel algorithm for global–local multiscale computations on massively parallel systems

The existing global–local multiscale computational methods, using finite element discretization at both the macro‐scale and micro‐scale, are intensive both in terms of computational time and memory requirements and their parallelization using domain decomposition methods incur substantial communication overhead, limiting their application. We are interested in a class of explicit global–local multiscale methods whose architecture significantly reduces this communication overhead on massively parallel machines. However, a naïve task decomposition based on distributing individual macro‐scale integration points to a single group of processors is not optimal and leads to communication overheads and idling of processors. To overcome this problem, we have developed a novel coarse‐grained parallel algorithm in which groups of macro‐scale integration points are distributed to a layer of processors. Each processor in this layer communicates locally with a group of processors that are responsible for the micro‐scale computations. The overlapping groups of processors are shown to achieve optimal concurrency at significantly reduced communication overhead. Several example problems are presented to demonstrate the efficiency of the proposed algorithm. Copyright © 2009 John Wiley & Sons, Ltd.     

A hybrid Tabu sample-sort simulated annealing approach for solving distributed scheduling problem

The distributed scheduling problem has been considered as the allocation of a task to various machines in such a way that these machines are situated in different factories and these factories are geographically distributed. Therefore distributed scheduling has fulfilled various objectives, such as allocation of task to the factories and machines in such a manner that it can utilise the maximum resources. The objective of this paper is to minimise the makespan in each factory by considering the transportation time between the factories. In this paper, to address such a problem of scheduling in distributed manufacturing environment, a novel algorithm has been developed. The proposed algorithm gleans the ideas both from Tabu search and sample sort simulated annealing. A new algorithm known as hybrid Tabu sample-sort simulated annealing (HTSSA) has been developed and it has been tested on the numerical example. To reveal the supremacy of the proposed algorithm over simple SSA and Tabu search, more computational experiments have also been performed on 10 randomly generated datasets.     

Balancing and sequencing of stochastic mixed-model assembly U-lines to minimise the expectation of work overload time

A mixed-model assembly U-line is a flexible production system capable of manufacturing a variety of similar models, and it has become popular as an important component of the just-in-time production system. However, it poses new challenges for the optimal design of assembly lines because both the task assignment and the production sequence affect the workload variance among workstations. As a consequence, this paper addresses the line balancing problem and the model sequencing problem jointly and proposes a 0–1 stochastic programming model. In this model, task times are assumed to be stochastic variables independently distributed with normal distributions and the objective is to minimise the expectation of work overload time for a given combination of cycle time and number of workstations. To solve the problem, a simulated annealing-based algorithm is developed, which can also be used to minimise the absolute deviation of workloads in a deterministic environment. The experimental results for a set of benchmark problems show that the proposed algorithm outperforms the existing algorithms in terms of solution quality and running time.     

A two-stage three-machine assembly scheduling problem with deterioration effect

The two-stage assembly scheduling problem has received growing attention in the research community. Furthermore, in many two-stage assembly scheduling problems, the job processing times are commonly assumed as a constant over time. However, it is at odds with real production situations some times. In fact, the dynamic nature of processing time may occur when machines lose their performance during their execution times. In this case, the job that is processed later consumes more time than another one processed earlier. In view of these observations, we address the two-stage assembly linear deterioration scheduling problem in which there are two machines at the first stage and an assembly machine at the second stage. The objective is to complete all jobs as soon as possible (or to minimise the makespan, implies that the system can yield a better and efficient task planning to limited resources). Given the fact that this problem is NP-hard, we then derive some dominance relations and a lower bound used in the branch-and-bound method for finding the optimal solution. We also propose three metaheuristics, including dynamic differential evolution (DDE), simulated annealing (SA) algorithm, and cloud theory-based simulated annealing (CSA) algorithm for find near-optimal solutions. The performances of the proposed algorithms are reported as well.     

Cost‐effective algorithms for processor arrays with reconfigurable bus systems

Abstract

A processor array with a reconfigurable bus system (abbreviated to PARBS) is a computation model which consists of a processor array and a reconfigurable bus system. It is a very powerful computation model in that it possesses the ability to solve many problems efficiently. However, most existing efficient algorithms on PARBS's use a large number of processors to solve problems. For example, to determine the maximum (minimum) of n data items in O(l) time, O(n ²) processors are required [12]. To solve the all‐pairs shortest paths and the minimum spanning tree problems in O(log n) time, O(n ⁴) processors are required [20]. These networks will therefore become very expensive for large n. In this paper, we introduce the concept of iterative‐PARBS, which is similar to the FOR‐loop construct in sequential programming languages. The iterative‐PARBS is a building block through which the processing data can be routed several times. We can think of it as a “hardware subroutine.’’ Based on this scheme, it is possible to explore more cost‐effective, time‐efficient parallel algorithms for use in a PARBS. The following new results are derived in this study: 1. The minimum (maximum) of n data items can be determined in O(l) time on a PARBS with O(n ^1+? ) processors for any fixed 8 > 0.

2. The all‐pairs shortest paths and the minimum spanning tree problems can be solved in O (log n) time on a PARBS with O(n ^3+? ) processors for any fixed 8 > 0.

     

Methodology for the optimal component selection of electronic devices under reliability and cost constraints

The objective of this paper is to present an efficient computational methodology for the reliability optimization of electronic devices under cost constraints. The system modeling for calculating the reliability indices of the electronic devices is based on Bayesian networks using the fault tree approach, in order to overcome the limitations of the series–parallel topology of the reliability block diagrams. Furthermore, the Bayesian network modeling for the reliability analysis provides greater flexibility for representing multiple failure modes and dependent failure events, and simplifies fault diagnosis and reliability allocation. The optimal selection of components is obtained using the simulated annealing algorithm, which has proved to be highly efficient in complex optimization problems where gradient‐based methods can not be applied. The reliability modeling and optimization methodology was implemented into a computer program in Matlab using a Bayesian network toolbox. The methodology was applied for the optimal selection of components for an electrical switch of power installations under reliability and cost constraints. The full enumeration of the solution space was calculated in order to demonstrate the efficiency of the proposed optimization algorithm. The results obtained are excellent since a near optimum solution was found in a small fraction of the time needed for the complete enumeration (3%). All the optimum solutions found during consecutive runs of the optimization algorithm lay in the top 0.3% of the solutions that satisfy the reliability and cost constraints. Copyright © 2007 John Wiley & Sons, Ltd.     

A genetic algorithm with domain knowledge for weapon‐target assignment problems

Abstract

In this paper, a novel genetic algorithm, including domain specific knowledge into the crossover operator and the local search mechanism for solving weapon‐target assignment (WTA) problems is proposed. The WTA problem is a full assignment of weapons to hostile targets with the objective of minimizing the expected damage value to own‐force assets. It is an NP‐complete problem. In our study, a greedy reformation and a new crossover operator are proposed to improve the search efficiency. The proposed algorithm outperforms its competitors on all test problems.     

流体饱和孔隙介质参数反演的模拟退火算法

本文研究了模拟退火算法在流体饱和孔隙介质参数反演中的应用.通过计算响应数据与实测响应数据的拟合将参数反问题归结为最优化问题.由于流体饱和孔隙介质运动方程的复杂性,动力响应与材料参数之间呈复杂的非线性关系,优化目标函数是非凸多峰函数.传统的梯度类优化方法一方面受局部极值的困扰难以搜索到全局最优解; 另一方面确定搜索方向须进行复杂的参数敏度分析.为克服这些困难,本文应用模拟退火算法进行了多参数反演数值模拟,模拟结果表明了模拟退火算法的可行性和稳健性.     

Single and multiobjective structural optimization in discrete-continuous variables using simulated annealing

A multivariable optimization technique based on the Monte-Carlo method used in statistical mechanics studies of condensed systems is adapted for solving single and multiobjective structural optimization problems. This procedure, known as simulated annealing, draws an analogy between energy minimization in physical systems and objective function minimization in structural systems. The search for a minimum is simulated by a relaxation of the statistical mechanical system where a probabilistic acceptance criterion is used to accept or reject candidate designs. To model the multiple objective functions in the problem formulation, a cooperative game theoretic approach is used. Numerical results obtained using three different annealing strategies for the single and multiobjective design of structures with discrete-continuous variables are presented. The influence of cooling schedule parameters on the optimum solutions obtained is discussed. Simulation results indicate that, in several instances, the optimum solutions obtained using simulated annealing outperform the optimum solutions obtained using some gradient-based and discrete optimization techniques. The results also indicate that simulated annealing has substantial potential for additional applications in optimization, especially for problems with mixed discrete-continuous variables.