The grid environment is a dynamic, heterogeneous, and changeable computing system that distributes various services amongst different clients. To attain the benefits of collaborative resource sharing in Grid computing, a novel and proficient grid resource management system (RMS) is essential. Therefore, detection of an appropriate resource for the presented task is a difficult task. Several scientists have presented algorithms for mapping tasks to the resource. Few of them focus on fault tolerance, user fulfillment, and load balancing. With this motivation, this study designs an intelligent grid scheduling scheme using deer hunting optimization algorithm (DHOA), called IGSS-DHOA which schedules in such a way that the makespan gets minimized in the grid platform. The IGSS-DHOA technique is mainly based on the hunting nature of humans toward deer. It also derives an objective function with candidate solution (schedule) as input and the outcome is the makespan value denoting the quality of the candidate solution. The simulation results highlighted the supremacy of the IGSS-DHOA technique over the recent state of art techniques with the minimal average processing cost of 31717.9.

Grid computing is a set of heterogeneous and dynamic resources from various administrative domains and provides access to the resource [

Once the Resource Management System (RMS) obtains the service request from the user, it splits the presented service tasks into a subtask that could be performed simultaneously. The RMS assigns 3 subtasks to the accessible resource for concurrent performance [

Grid schedulers work in 3 stages: job execution, and resource discovery, and allocation [

This study designs an intelligent grid scheduling scheme using deer hunting optimization algorithm (DHOA), called IGSS-DHOA. The goal of the IGSS-DHOA technique aims to determine a solution which produces optimum schedules in such a way that the makespan gets minimized in the grid platform. The IGSS-DHOA algorithm has derived an objective function with the minimization of makespan in the grid environment with the candidate solution as input and makespan value as output. The design of DHOA technique for scheduling process shows the novelty of the work. In order to assess the optimal scheduling performance of the IGSS-DHOA technique, an extensive simulation analysis is carried out.

The rest of the paper is organized as follows. Section 2 offers a literature review and Section 3 proposes the IGSS-DHOA technique. Then, Section 4 offers the experimental validation and Section 5 draws the conclusion.

This section offers a comprehensive review of existing grid scheduling approaches in the literature. Sousa et al. [

Dan et al. [

In Nazir et al. [

Kołodziej et al. [

Keerthika et al. [

The resource in computational grid is needed for performing the function, i.e., the processor utilized to process data. An investigation organization of computational grid has been responsible to study finding and distribution of tasks to certain resources. In general, it can be simple in getting the data on the capability for processing data in accessible resources. During this work, the scheduling issue in which

During these cases, the Expected Time to Compute (ETC) all jobs on all machines are calculated from task set determined as user and CPU time was resultant from grid systems. The resolve of ETC values are distinct investigation issues, and the statement of ETC data was utilized as benchmark for scheduling issues. ETC matrices were utilized for estimating the needed time to applying for all the jobs from all machines. The ETC matrix is

For formulating the issue, determine

An objective is for finding a permutation matrix

The scheduling constraint guarantee which all the tasks are allocated to exactly one resource.

In this case, the vector-based representation was utilized for encoding the schedule or solution to Grid scheduling issue. The size of vector was equivalent to the number of tasks. An index number of the element in vector represents the ID of tasks. Each element is integer in the range of 1 and m, where

An essential model of enhancement heuristic was iteratively enhancing incumbent solutions from all searches step by changing the present solution with neighbouring solution interms of quality/fitness value of neighbouring solutions. For describing the quality/fitness of all the solutions and guides the search model on the solution space, an estimation function was utilized for associating a real value for all solutions. The makespan has most general metric utilized to represent the amount of scheduling in Grid computing. Therefore, the estimation function was determined as function that estimates the makespan value of some provided candidate schedules [

where

An objective function in the IGSS-DHOA algorithm has the function which requires that exists optimized for achieving the aim. During this case, it can be regarded as one of the widely studied optimization conditions, for instance, the minimization of makespan, and it can be expressed as determining an objective function as estimation function. The input parameter for an objective function has been candidate solution/schedule. An output of objective function was makespan value demonstrating the estimation/amount of candidate solution or schedule. Assume that

The major goal of the presented DHOA method is to detect an optimum location for an individual to hunt the deer, it is essential to explore the deer's nature. They have special features that make complex hunting for the predator. A separate feature characterizes visual power i.e., 5 times bigger compared to humans. But they had problems seeing red and green colours. This section discusses the mathematical modeling of DHOA.

The major phase of technique is the beginning of hunter population, which is given as follows,

Let

After the initiation population, the deer position and wind angles are the important parameters in determining the optimal hunter position are initialized. Since the search spaces are considered as circles, the wind angles following the circumference of a circle.

In which

Since the location of optimum spaces is initially unidentified, the approach considers the candidate solution near to the optimum i.e., defined according to the FF, as an optimal outcome [

(i) Propagation through leaders’ location: afterward determining the optimum location every individual in the population attempts for attaining an optimum location and therefore, the procedure of upgrading the location starts. Consequently, the encircling nature is demonstrated by,

where

(ii) Propagation through angle location: for improving the search space, the idea is expanded by assuming the angles location in the upgraded rules. The angle estimation is necessary for defining the hunter's location thus prey isn't attentive to the attacks and henceforth, the hunting procedure would be efficient. The visualization angles of the prey/deer are calculated by,

According to the variance among visual and wind angles of the deer, a variable is calculated which is assist to upgrade the angle location.

By considering the angle location, the location is upgraded to implement as,

(iii) Propagation via successor location: In the exploration stage, a similar concept in encircling nature is adjusted by adopting the vector L. As it considers an arbitrary search firstly, the value of vector L is assumed lesser than one. Thus, the upgrade location is depending upon the successor location instead of initial optimum solution attained. This permits a global search which is given by,

From the arbitrary initiation of solution, the method upgrades the search agents’ location at all iterations depending upon attained optimum solution. When

The IGSS-DHOA technique intends to reduce the makespan and scheduling proficiency. Gridsim 5.0 toolkit is employed to evaluate the IGSS-DHOA technique under 16 resources and 512 tasks. The gridlets considered are autonomous and highly computational; it also follows Poisson process. It is considered that every resource can execute an individual gridlet at a time instant. The IGSS-DHOA technique is inspected under 4 cases as listed below.

■ Case 1: High Task Low Machine

■ Case 2: Low Task High Machine

■ Case 3: High Task High Machine

■ Case 4: Low Task Low Machine

Makespan (m) | ||||||
---|---|---|---|---|---|---|

Cases | Min-min | FTMM | BSA | LBFT | MLFT | IGSS-DHOA |

Case 1 | 32734.3 | 30695.6 | 28177.3 | 24819.6 | 22061.4 | 20022.8 |

Case 2 | 12587.8 | 10429.2 | 10429.2 | 9829.6 | 4553.13 | 3353.93 |

Case 3 | 20742.3 | 18943.5 | 16784.9 | 16185.3 | 13667 | 11628.4 |

Case 4 | 9469.84 | 7671.04 | 5872.25 | 5152.73 | 3593.77 | 2034.81 |

In line with, with case 4, the IGSS-DHOA approach has resulted in a lesser makespan of 2034.81 m whereas the Min-min, FTMM, BSA, LBFT, and MLFT algorithms have reached a superior makespan of 9469.84, 7671.04, 5872.25, 5152.73, and 3593.77 m correspondingly.

Hit count | ||||||
---|---|---|---|---|---|---|

Cases | Min-min | FTMM | BSA | LBFT | MLFT | IGSS-DHOA |

Case 1 | 311.393 | 332.801 | 380.787 | 389.646 | 390.384 | 396.29 |

Case 2 | 233.139 | 278.172 | 297.366 | 309.178 | 321.728 | 332.063 |

Case 3 | 265.622 | 287.031 | 311.393 | 324.681 | 328.372 | 335.754 |

Case 4 | 341.66 | 356.425 | 361.593 | 377.834 | 372.666 | 385.216 |

Deadline hit count | ||||||
---|---|---|---|---|---|---|

Cases | Min-min | FTMM | BSA | LBFT | MLFT | IGSS-DHOA |

Case 1 | 212.2046 | 238.4996 | 362.8033 | 381.927 | 380.7318 | 391.4888 |

Case 2 | 152.4431 | 196.6666 | 282.723 | 295.8705 | 300.6515 | 310.2133 |

Case 3 | 169.1763 | 227.7425 | 292.2849 | 318.5799 | 319.7751 | 330.5322 |

Case 4 | 230.133 | 243.2805 | 298.261 | 312.6037 | 350.8511 | 366.389 |

For sample, under case 1, the IGSS-DHOA algorithm has achieved an enhanced resource utilization of 93.57213% and Min-min, FTMM, BSA, LBFT, and MLFT techniques have gained a lower resource utilization of 70.393%, 77.00725%, 78.0809%, 90.04442%, and 92.03834% correspondingly. Moreover, under case 4, the IGSS-DHOA approach has reached a maximum resource utilization of 94.10900% and Min-min, FTMM, BSA, LBFT, and MLFT algorithms have achieved a reduced resource utilization of 69.9518%, 76.3170%, 78.54100%, 88.81740%, and 92.07670% correspondingly.

An average resource utilization analysis of the IGSS-DHOA technique with existing techniques take place in

Resource utilization (%) | ||||||
---|---|---|---|---|---|---|

Cases | Min-min | FTMM | BSA | LBFT | MLFT | IGSS-DHOA |

Case 1 | 70.10521 | 77.00725 | 78.0809 | 90.04442 | 92.03834 | 93.57213 |

Case 2 | 67.95791 | 73.93968 | 75.01333 | 89.12415 | 90.96469 | 93.41875 |

Case 3 | 72.86603 | 80.22820 | 83.14239 | 92.19172 | 95.25929 | 97.25321 |

Case 4 | 68.87819 | 74.09306 | 77.92752 | 83.90928 | 90.04442 | 92.19172 |

Average | 69.9518 | 76.3170 | 78.54100 | 88.81740 | 92.07670 | 94.10900 |

Processing cost | ||||||
---|---|---|---|---|---|---|

Cases | Min-min | FTMM | BSA | LBFT | MLFT | IGSS-DHOA |

Case 1 | 98224.68 | 95677.6 | 94949.87 | 85489.31 | 70206.88 | 63657.27 |

Case 2 | 52741.24 | 58199.26 | 54196.71 | 43280.69 | 33092.40 | 27998.26 |

Case 3 | 48010.97 | 45827.76 | 45100.03 | 39642.02 | 25087.32 | 19993.17 |

Case 4 | 26178.92 | 22904.11 | 23631.85 | 20720.91 | 18901.57 | 15262.90 |

Average | 56289 | 55652.2 | 54469.6 | 47283.2 | 36822 | 31727.9 |

Finally, an average processing cost of IGSS-DHOA technique with other techniques is provided in

This study has developed an effective IGSS-DHOA technique to generate optimal schedules in the grid environment. The IGSS-DHOA algorithm is mainly inspired by the hunting characteristics of humans towards deer. The IGSS-DHOA algorithm has derived an objective function with the minimization of makespan in the grid environment with the candidate solution as input and makespan value as output. In order to assess the optimal scheduling performance of the IGSS-DHOA technique, an extensive simulation analysis is carried out. The resultant values ensured the betterment of the IGSS-DHOA technique over the recent state of art techniques interms of different evaluation parameters. As a part of future scope, improved metaheuristic algorithms can be designed to further lessen the makespan in the grid environment.