Distributed multiprocessor task scheduling for a real-time system such as a cyber-physical system is challenging because of the varying communication delay to communicate to them. The task scheduling algorithm needs to take care of the communication delay factor into consideration while scheduling the tasks.so there is a need for a processor location-based (which decides the communication delay) task scheduling algorithm. There are few literary works that deal with such multiprocessor scheduling problems, which are summarized below.
Thus a sort of multi-thread, a multi-core-based speedy algorithm was designed and for the medium Delphi, language coding is done[1]. Massive no. of Traveling Salesman Problem (TSP) illustrations are taken from TSPLIB, through which quick searching is achieved without any quality losses or quality drop. The implemented algorithm can efficiently stabilize the dispute between the size of the increasing problem and the computational efficiency with hardware restriction; these are shown and verified through experimental results. Therefore suitable solutions are achieved.
In this article[2], the problem of developing plan making is assigned to an uneven TSP. The mathematical model of developing plan making is solved in the opinion of product restraints of iron and steel enterprises and the hot developing production method. Through the experimental results, it’s shown that the implemented algorithm and the model are effective and viable.
In paper[3], in order to find m closed trips, the Min-Max online m-Steiner Traveling Salesman Problem (MinMax-mSTSPonline) visits every single client at least once so that them salesmen cost becomes minimum. Thus they provided an online algorithm and its competitive ratio becomes maximum at
denotes the no. of blockages. The ratio seems to be asymptotically tight due to lower bound
and it is mentioned in [16].
The main theme of thepaper[4] is to create a hybrid algorithm to solve the TSP vigorously and efficiently. In order to achieve this, the traveling distance of salesman throughout the city minimization is proposed. An optimal solution is established based on this hybrid optimization algorithm, which concerns to minimize the traveling distance of the salesman or resolving the TSP. The aim of this implemented strategy is to integrate two optimization algorithms called Rider Optimization Algorithm (ROA) and Spotted Hyena Optimizer algorithm (SHO) in order to form a new algorithm or new variant of it called Spotted Hyena-based Rider Optimization (S-ROA). In the end, the obtained experimental results are compared in vice-versa with actually obtained results of hybrid algorithm, which implemented to solve these types of TPS cases and also verifies the competitive performance of the implemented model.
In paper[5], task scheduling problems are examined for Cloud Data-Centers (CDCs), and a mathematical model is introduced to schedule the two-stage tasks. A new genetic algorithm is formed by combining a genetic algorithm with Johnson's rule; this new generic algorithm is called as Johnson'sRule-Based Genetic Algorithm (JRGA). JRGA has the multi-processing scheduling feature in CDCs. To make the algorithm connection quicker, new crossover and mutation processes were developed. To improve the makespan of every single machine, Johnson’s rule is utilized in the decoding process. The JRGA’s performance is compared with a list of scheduling algorithms and the list of the enhanced algorithm. The comparison results validate JRGA.
A study[6] examined a wide range of resource-type cloud services and their computational system. To solve the present tasks, cooperative optimization scheduling is proposed. At first, a New Adaptive Genetic Algorithm (NAGA) was implemented. Due to the enhancement of the crossover mutation genetic operative, the tremendous entities were able to save by the algorithm as much as possible, improve the algorithm's optimization ability, and the chance of the algorithm falling for local ideal solution ratio was significantly decreased. Then the primary aspects which affect the service quality, i.e., task completion time, system load, and network bandwidth. To overcome this issue, an advanced fitness operator method was proposed to take forth the cloud resource collaborative optimization scheduling problem. Lastly, on the basis of an improved genetic algorithm (OSIG) an algorithm is implemented for cloud service resources. The implemented OSIG algorithm was able to efficiently optimize the resource scheduling plan, reduction of the task completion time, facilitate the system load stabilizing, and enhancement of the system's service quality. These are demonstrated in a cloud computing simulation platform called CloudSim. The theoretical analysis seems reliable on concerning the obtained experimental results.
In article[7], the authors consider side by side and non-parallel or sequential unloading of tasks to a multiple mobile edge computing servers. In order to minimize the unloading latency and also to reduce the failing chances, the task which contains a fine set of inter-dependent sub-tasks are scheduled to the servers. To resolve the scheduling problems, two algorithms were implemented and they were framed on the basis of genetic algorithm and conflict graph models. It’s found out that the performance of the given algorithm is quite near to the optimal solution’s performance is through simulation results and obtained in an extensive search. Moreover, the orthogonal channels were utilized by side-by-side unloading; results shows that the non-parallel or sequential unloading gets a lower failing chance in comparison with side-by-side unloading and also a demonstration is done to verify this. In spite of that side by side, unloading gives less latency. Nevertheless, the gap between side-by-side and sequential or non-parallel schemes decreases when the reliability of sub-tasks increases.
In article[8], on the basis of Ant Colony System (ACS) an event-driven active workshop scheduling model is introduced. In order to deal with the active events, two scheduling techniques were implemented. These techniques are called as parallel scheduling and parallel priority scheduling. Parallel scheduling aims to minimalize the overall makespan, whereas parallel priority scheduling aims to minimalize the active events delivery time. In addition to optimal scheduling,a strategy has been established with respect to the active events urgency degree, which is known as a selective scheduling strategy. At last, in order to solve the Dual-Objective Dynamic Job Shop Scheduling Problem (DJSP), the validity of the selective scheduling approach is verified in a large-scale problem test set and also experimentally.
In article[9], in order to simultaneously optimize both the execution time and the cost of it, a cloud workflow scheduling model is proposed. The proposed model can also optimize the multi-objective problems. A multi-objective framework is implemented on the basis of co- metamorphic multi-population of an innovative multi-objective ant colony system. To solve two objectives, this framework acquires two colonies. So to overcome the multi-objective challenges effectively, the implemented method integrates with three innovations. Those three innovations are: (i) a new pheromone update rule is implemented to make every single colony to search for its optimization objective adequately on the basis of the group of non-influential solutions, which are chosen/ taken from a global archive; (ii) in order to avoid the colony to focus only on its sole optimization task a balancing strategy is proposed so that it can collaborate with the pheromone update rule to stable the search of both objectives; (iii) the global Pareto front is attained with the help of the implemented best study strategy, which enhances the global archive’s solution quality to achieve this. On five different types of real-life scientific systems Experimental simulations were done and also these are carried out in consideration of the Amazon EC2 cloud platform’s characteristics. The implemented algorithm outperforms previous algorithms like some state-of-the-art multi-objective optimization approaches and the constrained optimization approaches, which are shown in experimental results.
The paper[10] adopts a vigorous optimization method in order to overcome the processing time’s uncertainty. The benefit of this approach is there is no need to assume the random data distribution the schedule, which is obtained, will stay strongly viable if the variation doesn’t exceed the pre-determined uncertainty limit. To develop the vigorous scheduling problem, a programming model is given, which is called as mixed-integer linear programming. To offer an efficient solution for the problem as soon as possible, a hybrid meta-experimental algorithm is introduced thereby syndicates the Ant Colony System (ACS) and advances the local search. At last, both the unsystematically generated and real-life cases were involved in conducting an experiment, a widespread experiment to verify the efficiency of the implemented algorithm. It is shown that the algorithm can solve the smaller cases optimally and also it can outperform well than two state-of-the-art meta-experimental algorithms in solving the large cases
The central theme of the paper[11] is to propose a new method to decrease the Cyber-Physical Systems (CPS) ’ thermally-prompted damages in its processors caused due to Dynamic Voltage and Frequency Scaling (DVFS) of high-level objective phase’s activity. Thereby dropping down the high-level objective task’s phase specifically can save more energy and with a certain amount of computational drop down the thermal stress is achieved. This method is shown to be the best because it can solve the issue without concerning the activity levels with conformist approaches, which use DVFS. The implemented reassignment of the task all over the core is determined by the present estimation of the core dependency, which is superior to traditional methods that either use present temperature or history of the temperature. This method increases the dependency majorly by about 20% in standard DVFS methods.
In order to overcome this problem also to increase the seclusion level and also to balance the performance of the system, an intruder location estimation is implemented for the seclusion of source location in CPS on the basis of Fake Source Scheduling (FSSE)[12]. The implemented FSSE has two main stages they are: in the first phase, it focuses on building a backbone to create a baseline in considering the privacy of the source location and delay of transmission, which is reliant on the communication information capturing probability of self and neighboring nodes. Fake message scheduling is the second stage. It’s implemented to compromise the privacy, transmission delay, and communication overhead according to a hypothesized position of the intruder. The implemented method was able to stabilize the seclusion level and also stabled most effectively the delay of transmission and energy consumption in comparison with the following three algorithms: phantom routing, tree-based diversionary routing, and dynamic fake source selection. A demonstration is done via investigation and simulation
Two algorithm’s necessary features are compared in thepaper[13]. Then they are recalculated based on synthesized data; these data are synthesized via bus path micro-simulation to produce a close ideal scheduling strategy. Based on their precision and effectiveness in producing a low-cost solution, the algorithms were compared. Even though both the algorithms provide low-cost solutions, the Ant Colony algorithm exhibits great effect, i.e., it achieves a decent solution with less schedule strategy estimation. The advantages of the improved algorithm in the bus path scheduling were discussed.
Thus based on the literature review, it is observed that the genetic algorithm can solve many optimization problems [14–16]. So this article intended to solve the distributive processor scheduling using a genetic algorithm with the contribution of
remain part of the article is organized as below: section 2 gives the system model, section 3 provides results and discussion,and section 4 concludes the article with a summary.