Applied Science and Innovative Research ISSN 2474-4972 (Print) ISSN 2474-4980 (Online) Vol. 7, No. 2, 2023 www.scholink.org/ojs/index.php/asir 30 Original Paper Task Scheduling Optimization in Cloud Computing by Jaya Algorithm Ahmed Y. Hamed1, M. Kh. Elnahary1 & Hamdy H. El-Sayed1 1 Faculty of Computers and Artificial Intelligence, Department of Computer Science, Sohag University, Sohag, 82524, Egypt Received: March 2, 2023 Accepted: March 16, 2023 Online Published: March 20, 2023 doi:10.22158/asir.v7n2p30 URL: http://doi.org/10.22158/asir.v7n2p30 Abstract Cloud computing provides resources to its consumers as a service. The cloud computing paradigm offers dynamic services by providing virtualized resources via the internet for enabling applications, and these services are provided by large-scale data centers known as clouds. Cloud computing is entirely reliant on the internet to provide its services to consumers. Cloud computing offers several advantages, including the fact that users only pay for what they use weekly, monthly, or yearly, that anybody with an internet connection may use the cloud, and that there is no need to purchase resources, hardware, or software on their own. This paper proposes an efficient task scheduling algorithm based on the Jaya algorithm for the cloud computing environment. We evaluate the performance of our method by applying it to three instances. The recommended technique produced the optimal solution in makespan, speedup, efficiency, and throughput, according to the findings. Keywords Heterogeneous resources, Jaya algorithm, Task scheduling, Cloud Computing 1. Introduction Task scheduling in heterogeneous computing systems linked by high-speed networks has received a lot of attention. Such approaches promise the quick processing of computationally heavy applications with a wide range of calculation requirements. A significant application can be divided into several smaller subtasks before parallel processing. These smaller subtasks usually involve dependencies that indicate precedence restrictions, in which the outcomes of other subtasks are necessary before a particular subtask may be done. Decomposing a computation into smaller subtasks and performing the subtasks on several processors might lower the computation’s overall execution time, i.e., the makespan. As a result, a task scheduling algorithm usually schedules all subtasks on a given number of available www.scholink.org/ojs/index.php/asir Applied Science and Innovative Research Vol. 7, No. 2, 2023 31 Published by SCHOLINK INC. processors to minimize makespan while respecting precedence restrictions. The development of task scheduling algorithms that allocate subtasks of an application to processors is a difficulty in heterogeneous computing systems. As a result, various techniques for minimizing makespan for parallelizing subtasks with precedence connections have been presented. The precedence connections are represented as a directed acyclic graph (DAG) with vertices representing computations and directed edges describing dependencies between those vertices. DAGs have been demonstrated to be expressive for various applications (Xu, Li, Hu, & Li, 2014). This paper introduced an efficient algorithm based on the Jaya algorithm called the efficient Jaya algorithm (EJA) to lower the makespan and optimize the speedup, efficiency, and throughput to handle the task scheduling problem successfully. The paper is organized as follows: The notations are presented in section 2. Related work is presented in Section 3. The problem description is given in Section 4. The Jaya algorithm is given in Section 5. Section 6 describes the EJA approach. The evaluation of the proposed algorithm is presented in section 7. Section 8 concludes and offers future work. 2. Notation 3. Related Work A heuristic-based task scheduling approach in parallel and distributed heterogeneous computing systems generally consists of two phases: job prioritization and processor selection. Varying priority produces different makespan on a heterogeneous computing system in a heuristic-based job scheduling method. As a result, an intelligent scheduling algorithm should be able to give priority to each subtask based on the resources required to reduce makespan. This work (Xu, Li, Hu, & Li, 2014) proposes a job scheduling approach for heterogeneous computing systems based on a multiple priority queues genetic algorithm (MPQGA). The primary concept behind our method is to use the benefits of both evolutionary and heuristic algorithms while avoiding their downsides. The suggested technique uses a GRT It refers to the graph of tasks TASi It refers to the task i VTMi It refers to the virtual machine i NVTM It refers to a virtual machine’s number NTAS It refers to the number of tasks COM_CO(TASi, TASj) It refers to the communication cost between TASi and TASj Sta_Time(TASi, VTMj) It refers to the start time of task i on a VTMj Fts_Time(TASi, VTMj) It refers to the finish time of task i on a VTMj Rey_Time(VTMi) It refers to the VTM’s ready time i DALT It refers to a list of tasks arranged in topological order of DAG Data_Arr(TSi, VMj) It refers to the time of task’s i data arrival to VTMj www.scholink.org/ojs/index.php/asir Applied Science and Innovative Research Vol. 7, No. 2, 2023 32 Published by SCHOLINK INC. genetic algorithm (GA) approach to prioritize each subtask while searching for a solution for the task-to-processor mapping using a heuristic-based earliest completion time (EFT). The MPQGA approach also creates crossover, mutation, and fitness functions appropriate for directed acyclic graph (DAG) scheduling. Cloud computing provides resources to its consumers as a service. The cloud computing paradigm offers dynamic services by providing virtualized resources via the internet for enabling applications, and these services are provided by large-scale data centers known as clouds. Cloud computing is entirely reliant on the internet to provide its services to consumers. Cloud computing offers several advantages, including the fact that users only pay for what they use (weekly, monthly, or yearly), that anybody with an internet connection may use the cloud, and that there is no need to purchase resources (hardware, software) on their own. This study (June, 2014) introduces a novel approach, Hybrid enhanced particle swarm optimization with mutation crossover, to get the most out of resources. Optimized resource use is vital, and scheduling plays a significant role. Cloud computing has lately experienced rapid growth and has emerged as a commercial reality in information technology. Cloud computing is a supplement, consumption, and delivery model for internet-based Information Technology services charged per usage. The scheduling of cloud services influences the cost-benefit of this computing paradigm by service providers to users. Tasks should be scheduled efficiently in such a circumstance to decrease execution cost and time. In this research (Kaur, & Verma, 2012), the authors suggested a meta-heuristic-based scheduling method that reduces execution time and cost. An enhanced genetic algorithm is created by combining two existing scheduling methods for scheduling activities while considering their computational complexity and computing capability of processing elements. Due to the expansion of data centers’ size, complexity, and performance, client needs in execution time and throughput has become increasingly complicated. Against this backdrop, this paper introduces a new resource allocation model that improves task scheduling by combining a multi-objective optimization (MOO) and particle swarm optimization (PSO) technique. The authors create a novel multi-objective PSO (MOPSO) algorithm based on a unique ranking technique. This algorithm’s fundamental idea is that jobs are assigned to virtual machines to reduce waiting time and maximize system throughput (Alkayal, Jennings, & Abulkhair, 2016). When excellent efficiency is required, task scheduling is one of the essential concerns in heterogeneous contexts. Because task scheduling is a Nondeterministic Polynomial (NP)-hard issue, various evolutionary methods have been developed to address it. Because population-based algorithms have a slow convergence rate, they are combined with local search algorithms. Thus, in this study (Dordaie & Navimipour, 2018), a hybrid particle swarm optimization and hill-climbing method are suggested to improve the task scheduling makespan. www.scholink.org/ojs/index.php/asir Applied Science and Innovative Research Vol. 7, No. 2, 2023 33 Published by SCHOLINK INC. 4. Problem Description In cloud computing, task scheduling is represented as a graph with NTAS tasks (TAS1, TAS2, TAS3, ..., TASNTAS). Each node (task) with GRT and E-directed edges represents a subset of the tasks’ requests (Hamed & Alkinani, 2021). Each node (task) represents an instruction that may be executed sequentially on the same virtual machine as other instructions; it may have one or more inputs. The availability of the inputs triggers the execution of an exit or entry task. A partial request with a precedence constraint (TASi → TASj), i.e., TASi precedes TASj in the implementation process. The execution time of a task TASi is denoted by (TASi) weight. Let COM_CO(TASi, TASj) be the cost of communication of an edge, and it will be equal to zero if TASi and TASj are scheduled on the same virtual machine. Start and finish times are denoted by Sta_Time(TASi, VTMj) and Fts_Time(TASi, VTMj), respectively [6]. The Data_Arr of TASi at virtual machine VTMj is given by: Data_Arr(TASi. VTMj) = max{Fts_Time(TASk, VTMj) + COM_CO(TASi, TASk)} (1) Where k = 1.2, ..., number of Parents The task scheduling problem in cloud computing may be defined as determining the best time to allocate or schedule the start times of the specified tasks on virtual machines. While maintaining precedence is restricted, the completion time (schedule length) and execution cost decrease. The completion time is defined as the schedule length or finishes time calculated as follows: Scheduled Length=max(Fts_Time(TASi, VTMj)) (2) Fts_Time(TASi,VTMj)=Sta_Time(TASi,VTMj)+ WETij (3) Where i = 1.2. ...., NTAS, and j = 1,2, …NVTM Algorithm 1: To find the schedule length (Hamed & Alkinani, 2021) Input the schedule of tasks as shown in Table 1 Rey_Time[VTMj] = 0 where j = 1, 2,…NVTM. For i = 1 : NTAS { From DALT take the first task TASi to be executed and remove it from DALT. For j = 1 : NVTM { If TASi is scheduled to virtual machine VTMj Sta_Time(TASi,VTMj)=max{Ret_Time(VTMj),Data_Arr(TASi, VTMj)} Fts_Time(TASi, VTMj) = Sta_Time(TASi, VTMj) + WET(TASi, VTMj) Rey_Time(VTMj) = Fts_Time(TASi, VTMj) End If } } Schedule length = max(Fts_Time) www.scholink.org/ojs/index.php/asir Applied Science and Innovative Research Vol. 7, No. 2, 2023 34 Published by SCHOLINK INC. 5. Jaya Algorithm Let G(y) be the objective function that has to be minimized (or maximized). Assume that there are ‘m’ number of design variables (i.e. j=1,2,...,m) and ‘n’ number of possible solutions (i.e. population size, k=1,2,...,n) at any iteration i. Let the best candidate acquire the best value of G(y) (i.e., G(y)best) in all candidate solutions, and the worst candidate obtain the worst value of G(y) (i.e., G(y)worst) in all candidate solutions. If the value of the jth variable for the kth candidate during the ith iteration is Yj,k,i, then this value is updated according to the following Eq (4) (Venkata Rao, 2016). Y’j,k,i = Yj,k,i + ran1,j,i (Yj,best,i - │Yj,k,i│) - ran2,j,i (Yj,worst,i - │Yj,k,i│) (4) where Yj,best,i is the variable j value for the best candidate and Yj,worst,i is the variable j value for the worst candidate Y’j,k,i is the updated value of Yj,k,i, and ran1,j,i and ran2,j,i are the two random values for the jth variable in the range [0, 1] during the ith iteration. The word "ran1,j,i ((Yj,best,i - Yj,k,i)" denotes the solution’s propensity to get closer to the best solution, whereas the term "- ran2,j,i (Yj,worst,i - Yj,k,i)" denotes the solution’s inclination to avoid the worst solution. If Y’j,k,i yields a superior function value, it is accepted. At the end of the iteration, all of the acceptable function values are kept, and these values form the input to the following iteration (Venkata Rao, 2016). Jaya algorithm Initialize population size, number of design variables, and termination criterion Iteration=1 While (iteration <= termination criterion) Identify the best and the worst solutions in the population Modify the solutions based on the best and the worst solutions by using Eq.(4) If the solution of Y’j,k,i is better than Yj,k,i Update the old solution with the new obtained solution End if Iteration =iteration +1 End while 6. The Proposed Algorithm Because it is evident that the vector representation in the Jaya algorithm is in continuous value form, we will utilize the five ways to transform these continuous values into discrete values. The first rule is the Smallest Position Value (SPV) (Dubey & Gupta, 2017), the second is the Largest Position Value (LPV) (Wang, Pan, & Tasgetiren, 2011), the third is the round nearest function, the fourth is the floor nearest function, and the fifth is the Ciel nearest function. Table 1 shows how we will utilize the modulus function with the number of virtual machines in the SPV and LPV to raise the result by one. www.scholink.org/ojs/index.php/asir Applied Science and Innovative Research Vol. 7, No. 2, 2023 35 Published by SCHOLINK INC. Table 1. Convert Continuous Values to Discrete Values Population 1.0 1.6 1.4 3.0 2.3 1.9 2.0 SPV rule 1 3 2 6 7 5 4 modulus with SPV and NVRM=3 2 1 3 1 2 3 2 LPV rule 4 5 7 6 2 3 1 modulus with LPV and NVRM=3 2 3 2 1 3 1 2 round nearest function 1 2 1 3 2 2 2 floor nearest function 1 1 1 3 2 1 2 ceil nearest function 1 2 2 3 3 2 2 Algorithm 2: The function that converts a continuous value to a discrete value Function convert_to_discrete (u) Rand=random number between [1…5] If (Rand == 1) Transform the continuous values by the SPV rule Else if (Rand == 2) Transform the continuous values by the LPV rule Else if (Rand == 3) Transform the continuous values by round the nearest function Else if (Rand == 4) Transform the continuous values by the nearest function Else Transform the continuous values by ceil nearest function End if End function Algorithm 3: EJA Input DAG with communication and computation cost Initialize population size, number of design variables, lower bound, upper bound, and termination criterion Initialize population by using population = lower bound + rand * (upper bound – lower bound) Convert the Initialize population by using Algorithm 2 Calculate the schedule length by using Algorithm 1 Iteration=1 While (iteration <= termination criterion) Identify the best and the worst solutions in the population www.scholink.org/ojs/index.php/asir Applied Science and Innovative Research Vol. 7, No. 2, 2023 36 Published by SCHOLINK INC. Modify the solutions based on the best and the worst solutions by using Eq.(4) Convert the obtained solution by using Algorithm 2 Calculate the schedule length by using Algorithm 1 If the solution of Y’j,k,i is better than Yj,k,i Update the old solution with the new obtained solution End if Iteration =iteration +1 End while 7. Evaluation of the EJA We demonstrate the EJA’s performance by applying it to three instances. The first scenario has ten tasks and three heterogeneous virtual machines, and the second scenario has ten tasks and three heterogeneous virtual machines. The third is made up of three heterogeneous virtual machines and eleven tasks. Speedup = min VTMj ( ∑ WETi,j schedule lengthTASi ) (5) Efficiency = Speedup NVTM (6) Throughput = NTAS Schedule Length (7) 7.1 Case 1 In this case, the tasks {TAS1, TAS2, TAS3, TAS4, TAS5, TAS6, TAS7, TAS8, TAS9, TAS10} are executed on three heterogeneous virtual machines {VTM1, VTM2, VTM3}. The cost of executing each task on different virtual machines is shown in Table 2 (Younes, Ben Salah, Farag, Alghamdi, & Badawi, 2019). The schedule obtained by EJA is shown in Table 3. The results obtained by the EJA are compared with those obtained by the Whale Optimization Algorithm (WOA) (Thennarasu, Selvam, & Srihari, 2021), Gravitational Search Algorithm (GSA) (Biswas, Kuila, Ray, & Sarkar, 2019), and Hybrid Heuristic and Genetic-based scheduling task algorithm for heterogeneous computing (HHG) (Sulaiman, Halim, Lebbah, Waqas, & Tu, 2021). The results obtained by the EJA and WOA, GSA, and HHG are illustrated in Table 4. The task priority of EJA {TAS1, TAS6, TAS4, TAS5, TAS2, TAS3, TAS8, TAS9, TAS7, TAS10}. Figure 1, Figure 2, Figure 3, and Figure 4 represent the results obtained by the EJA, WOA, GSA, and HHG in terms of makespan, speedup, efficiency, and throughput. www.scholink.org/ojs/index.php/asir Applied Science and Innovative Research Vol. 7, No. 2, 2023 37 Published by SCHOLINK INC. Table 2. Computation Cost for Case 1 TAS/ VTM VTM1 VTM2 VTM3 TAS1 22 21 36 TAS2 22 18 18 TAS3 32 27 43 TAS4 7 10 4 TAS5 29 27 35 TAS6 26 17 24 TAS7 14 25 30 TAS8 29 23 36 TAS9 15 21 8 TAS10 13 16 33 Table 3. Schedule Obtained by EJA for Case 1 VTM1 VTM2 VTM3 Sta_Time Fts_Time Sta_Time Fts_Time Sta_Time Fts_Time TAS1 - - 0 21 - - TAS2 - - 21 39 - - TAS3 - - 39 66 - - TAS4 64 71 - - - - TAS5 - - - - 34 69 TAS6 38 64 - - - - TAS7 - - 66 91 - - TAS8 71 100 - - - - TAS9 - - - - 78 86 TAS10 100 113 - - - - Table 4. The Comparative Results for Case 1 Algorithm Makespan WOA 122 GSA 122 HHG 117 EJA 113 www.scholink.org/ojs/index.php/asir Applied Science and Innovative Research Vol. 7, No. 2, 2023 38 Published by SCHOLINK INC. Figure 1. Comparison of Makespan for Case 1 Figure 2. Comparison of Speedup for Case 1 Figure 3. Comparison of Efficiency for Case 1 Figure 4. Comparison of Throughput for Case 1 105 110 115 120 125 WOA GSA HHG EJA M ak es p an 1.6 1.65 1.7 1.75 1.8 1.85 WOA GSA HHG EJA S p ee d u p 0.54 0.56 0.58 0.6 0.62 WOA GSA HHG EJA E ff ic ie n cy 0.075 0.08 0.085 0.09 WOA GSA HHG EJA T h ro u g h p u t www.scholink.org/ojs/index.php/asir Applied Science and Innovative Research Vol. 7, No. 2, 2023 39 Published by SCHOLINK INC. 7.2 Case 2 In this case, the tasks {TAS1, TAS2, TAS3, TAS4, TAS5, TAS6, TAS7, TAS8, TAS9, TAS10} are executed on three heterogeneous virtual machines {VTM1, VTM2, VTM3}. The cost of executing each task on different virtual machines is shown in Table 5 (Younes, Ben Salah, Farag, Alghamdi, & Badawi, 2019). The schedule obtained by EJA is shown in Table 6. The results obtained by the EJA are compared with those obtained by the Ant Colony Optimization (ACO) (Tawfeek, El-Sisi, Keshk, & Torkey, 2015), Heterogeneous Earliest Finish Time (HEFT) (Topcuoglu, Hariri, & Wu, 2002), and Critical Path on Processor (CPOP) (Topcuoglu, Hariri, & Wu, 2002). The results obtained by the EJA and ACO, HEFT, and CPOP are illustrated in Table 7. The task priority of EJA { TAS1, TAS4, TAS3, TAS2, TAS5, TAS6, TAS9, TAS8, TAS7, TAS10}. Figure 5, Figure 6, Figure 7, and Figure 8 represent the results obtained by the EJA, ACO, HEFT, and CPOP in terms of makespan, speedup, efficiency, and throughput. Table 5. Computation Cost for Case 1 TAS/ VTM VTM1 VTM2 VTM3 TAS1 14 16 9 TAS2 13 19 18 TAS3 11 13 19 TAS4 13 8 17 TAS5 12 13 10 TAS6 13 16 9 TAS7 7 15 11 TAS8 5 11 14 TAS9 18 12 20 TAS10 21 7 16 Table 6. Schedule Obtained by EJA for Case 2 VTM1 VTM2 VTM3 Sta_Time Fts_Time Sta_Time Fts_Time Sta_Time Fts_Time TAS1 - - - - 0 9 TAS2 - - - - 9 27 TAS3 21 32 - - - - TAS4 - - 18 26 - - TAS5 - - 26 39 - - TAS6 - - - - 27 36 TAS7 32 39 - - - - TAS8 - - 55 66 - - www.scholink.org/ojs/index.php/asir Applied Science and Innovative Research Vol. 7, No. 2, 2023 40 Published by SCHOLINK INC. TAS9 - - 43 55 - - TAS10 - - 66 73 - - Table 7. The Comparative Results for Case 2 Algorithm Makespan CPOP 86 HEFT 80 ACO 78 EJA 73 Figure 5. Comparison of Makespan for Case 2 Figure 6. Comparison of Speedup for Case 2 Figure 7. Comparison of Efficiency for Case 2 65 70 75 80 85 90 CPOP HEFT ACO EJA M ak es p an 1.3 1.4 1.5 1.6 1.7 1.8 CPOP HEFT ACO EJA S p ee d u p 0.45 0.5 0.55 0.6 CPOP HEFT ACO EJA E ff ic ie n cy www.scholink.org/ojs/index.php/asir Applied Science and Innovative Research Vol. 7, No. 2, 2023 41 Published by SCHOLINK INC. Figure 8. Comparison of Throughput for Case 2 7.3 Case 3 In this case, the tasks {TAS1, TAS2, TAS3, TAS4, TAS5, TAS6, TAS7, TAS8, TAS9, TAS10} are executed on three heterogeneous virtual machines {VTM1, VTM2, VTM3}. The cost of executing each task on different virtual machines is shown in Table 8 (Keshanchi, Souri, & Navimipour, 2017). The schedule obtained by EJA is shown in Table 9. The results obtained by the EJA are compared with those obtained by the Multiple Priority Queues and a Memetic Algorithm (MPQMA) (Keshanchi, Souri, & Navimipour, 2017), a New Genetic Algorithm (NGA) (Keshanchi, Souri, & Navimipour, 2017). The results obtained by the EJA, MPQMA, and NGA are illustrated in Table 10. The task priority of EJA {TAS0, TAS2, TAS3, TAS4, TAS1, TAS6, TAS8, TAS7, TAS5, TAS9, TAS10}. Figure 9, Figure 10, Figure 11, and Figure 12 represent the results obtained by the EJA, MPQMA, and NGA in terms of makespan, speedup, efficiency, and throughput. Table 8. Computation Cost for Case 3 TAS / VTM VTM1 VTM2 VTM3 TAS0 10 11 12 TAS1 11 12 13 TAS2 12 8 13 TAS3 14 10 18 TAS4 27 20 19 TAS5 15 12 18 TAS6 9 14 19 TAS7 19 12 14 TAS8 14 10 15 TAS9 15 12 15 TAS10 18 10 17 0 0.05 0.1 0.15 CPOP HEFT ACO EJA T h ro u g h p u t www.scholink.org/ojs/index.php/asir Applied Science and Innovative Research Vol. 7, No. 2, 2023 42 Published by SCHOLINK INC. Table 9. Schedule OBTained by EJA for CAse 3 VTM1 VTM2 VTM3 Sta_Time Fts_Time Sta_Time Fts_Time Sta_Time Fts_Time TAS0 - - 0 11 - - TAS1 36 47 - - - - TAS2 - - 11 19 - - TAS3 - - 19 29 - - TAS4 - - 29 49 - - TAS5 47 62 - - - - TAS6 - - - - 29 48 TAS7 - - 59 71 - - TAS8 - - 49 59 - - TAS9 - - 71 83 - - TAS10 - - 83 93 - - Table 10. The Comparative Results for Case 3 Algorithm Makespan MPQMA 101 NGA 101 EJA 93 Figure 9. Comparison of Makespan for Case 3 Figure 10. Comparison of Speedup for Case 3 85 90 95 100 105 MPQMA NGA EJA M ak es p an 1.2 1.25 1.3 1.35 1.4 1.45 MPQMA NGA EJA S p ee d u p www.scholink.org/ojs/index.php/asir Applied Science and Innovative Research Vol. 7, No. 2, 2023 43 Published by SCHOLINK INC. Figure 11. Comparison of Efficiency for Case 3 Figure 12. Comparison of Throughput for Case 3 8. Conclusion and Future Work The suggested efficient Jaya algorithm allocates or schedules subtasks to available virtual machines in a cloud computing context. According to the findings obtained on DAGs of various situations, the efficient Jaya algorithm outperforms other algorithms in terms of makespan, speedup, efficiency, and throughput. In the future, we will create an algorithm based on DAGs that will consider resource load balancing. References Alkayal, E. S., Jennings, N. R., & Abulkhair, M. F. (2016). Efficient Task Scheduling Multi-Objective Particle Swarm Optimization in Cloud Computing. Proc. - Conf. Local Comput. Networks, LCN. (2016), 17-24. https://doi.org/10.1109/LCN.2016.024 Biswas, T., Kuila, P., Ray, A. K., & Sarkar, M. (2019). Gravitational search algorithm based novel workflow scheduling for heterogeneous computing systems. Simul. Model. Pract. Theory. 96(2019), 101932. https://doi.org/10.1016/j.simpat.2019.101932 Dordaie, N., & Navimipour, N. J. (2018). A hybrid particle swarm optimization and hill climbing algorithm for task scheduling in the cloud environments. ICT Express, 4(2018), 199-202. https://doi.org/10.1016/j.icte.2017.08.001 Dubey, I., & Gupta, M. (2017). Uniform mutation and SPV rule based optimized PSO algorithm for TSP problem. Proc. 2017 4th Int. Conf. Electron. Commun. Syst. ICECS 2017, 17(2017), 168-172. https://doi.org/10.1109/ECS.2017.8067862 0.41 0.42 0.43 0.44 0.45 0.46 0.47 MPQMA NGA EJA E ff ic ie n cy 0.095 0.1 0.105 0.11 0.115 MPQMA NGA EJA T h ro u g h p u t www.scholink.org/ojs/index.php/asir Applied Science and Innovative Research Vol. 7, No. 2, 2023 44 Published by SCHOLINK INC. Hamed, A. Y., & Alkinani, M. H. (2021). Task scheduling optimization in cloud computing based on genetic algorithms, Comput. Mater. Contin, 69(2021), 3289-3301. https://doi.org/10.32604/cmc.2021.018658 June, A. (2014). Research Paper on Optimized Utilization of Resources Using PSO and Improved Particle Swarm Optimization (IPSO). Algorithms in Cloud Computing, 2(2014), 499-505. Kaur, S., & Verma, A. (2012). An Efficient Approach to Genetic Algorithm for Task Scheduling in Cloud Computing Environment. Int. J. Inf. Technol. Comput. Sci. 4(2012), 74-79. https://doi.org/10.5815/ijitcs.2012.10.09 Keshanchi, B., Souri, A., & Navimipour, N. J. (2017). An improved genetic algorithm for task scheduling in the cloud environments using the priority queues: Formal verification, simulation, and statistical testing. J. Syst. Softw. 124(2017), 1-21. https://doi.org/10.1016/j.jss.2016.07.006 Sulaiman, M., Halim, Z., Lebbah, M., Waqas, M., & Tu, S. (2021). An Evolutionary Computing-Based Efficient Hybrid Task Scheduling Approach for Heterogeneous Computing Environment. J. Grid Comput. 19(2021). https://doi.org/10.1007/s10723-021-09552-4 Tawfeek, M., El-Sisi, A., Keshk, A., & Torkey, F. (2015). Cloud task scheduling based on ant colony optimization. Int. Arab J. Inf. Technol. 12(2015), 129-137. Thennarasu, S. R., Selvam, M., & Srihari, K. (2021). A new whale optimizer for workflow scheduling in cloud computing environment. J. Ambient Intell. Humaniz. Comput. 12(2021), 3807-3814. https://doi.org/10.1007/s12652-020-01678-9 Topcuoglu, H., Hariri, S., & Wu, M. Y. (2002). Performance-effective and low-complexity task scheduling for heterogeneous computing. IEEE Trans. Parallel Distrib. Syst. 13(2002), 260-274. https://doi.org/10.1109/71.993206 Venkata Rao, R. (2016). Jaya: A simple and new optimization algorithm for solving constrained and unconstrained optimization problems. Int. J. Ind. Eng. Comput. 7(2016), 19-34. https://doi.org/10.5267/j.ijiec.2015.8.004 Wang, L., Pan, Q. K., & Tasgetiren, M. F. (2011). A hybrid harmony search algorithm for the blocking permutation flow shop scheduling problem. Comput. Ind. Eng. 61(2011), 76-83. https://doi.org/10.1016/j.cie.2011.02.013 Xu, Y., Li, K., Hu, J., & Li, K. (2014). A genetic algorithm for task scheduling on heterogeneous computing systems using multiple priority queues. Inf. Sci. (Ny). 270(2014), 255-287. https://doi.org/10.1016/j.ins.2014.02.122 Younes, A., Ben Salah, A., Farag, T., Alghamdi, F. A., & Badawi, U. A. (2019). Task scheduling algorithm for heterogeneous multi processing computing systems. J. Theor. Appl. Inf. Technol. 97(2019), 3477-3487.