No abstract

Purpose Grid computing is an effective environment for the execution of parallel applications that requires great computing power. This paper aims to present, based on the hierarchical architecture, an improved weighted resource discovery (WRD) algorithm to manage allocation of resources and minimize cost of communications between grid nodes. Design/methodology/approach A behavioral modeling method is addressed to prove the proposed method correctness. The behavioral model of the proposed algorithm is implemented by StarUML tool with two different model-checking mechanisms. Then, the resource discovery correctness is analyzed in terms of reachability condition, fairness condition and deadlock-free using NuSMV model checker. Findings The results show that WRD algorithm has better performance in requiring re-discovery process, the number of examined nodes in each request and discovering the free resources with high-bandwidth links. Originality/value To store information of resources, a new data structure called resource information table is proposed which facilitates resource finding of the algorithm. A behavioral modeling method is addressed to prove the proposed method correctness.

In recent years, wireless sensor networks are in great use in applications like disaster management, combat field reconnaissance, border protection and safe care. Although, much research has been done on wireless sensor networks, but in the quality of service (QoS) field there are not enough researches. Since these networks are widely used in many areas, there are different QoS parameters in contrast with traditional networks such as network coverage, optimal number of active nodes, network lifetime and energy consumption. We have proposed an automata-based scheduling method to improve the QoS parameters of the networks. In this method, each node is equipped with a learning automaton to select its correct status (active or passive) at any given time. Simulation results show that the proposed method in comparison with some existing methods such as: CCP, Lacoverage, PEAS and Ottawa reduce energy consumption and increase network's lifetime. As a result, several QoS parameters are considered in sensor networks, simultaneously.

In this paper intelligent search technique of variable structure learning automata (VSLA) has been used to solve single machine total weighted tardiness job scheduling problem. The goal is investigating reduction in delays result in late execution of the jobs after specified deadline as well as reducing the time required to find the best execution order of the jobs. For this reason, fixed structure learning automata and genetic algorithm approaches has been studied and then a new scheduling approach called VSLA-Scheduler has been proposed by employing variable structure learning automata technique. In order to identify strengths and weaknesses of the proposed method, its performance is compared with other intelligent techniques. In this regard, for performance evaluation of the proposed method and comparing it with other methods, computer simulations have been used. Finally, the results produced by the proposed and previous algorithms have been compared with the best solutions in OR library. Experimental results show that the proposed algorithm's performance (VSLA-Scheduler) is more acceptable than other methods. General TermsScheduling, single machine task scheduling, machine learning.