Related Works:

Round Robin:
A popular scheduling algorithm called Round Robin uses quantum time to divide CPU time among processes. This technique is used in cloud and CPU
computing, where excellent QoS and effective performance are essential. Researchers have put forth a number of methods to enhance the algorithm, which can
be divided into two groups: dynamic quantum and static quantum. [1] presents a new taxonomy based on quantum time and reviews prior research on RR
algorithms in CPU and cloud computing. Subsequent research ought to concentrate on creating or refining RR algorithms in a manner that integrates static and
dynamic quantum, and assessing them through practical workloads.

1. https://ieeexplore.ieee.org/abstract/document/8769534

Multilevel Feedback Queue:

The experiments show that dynamically generating time quantum and using SJF before RR improve CPU and resource utilization. The Round Robin algorithm
is preferred for fair CPU use in the first queue, while SJF prior to RR in the next queues reduces average waiting and turned-around time. Efficient MLFQ can be
designed to manage processes considering the nature of submitted processes, and the experiment is being enhanced for a more realistic analysis.

https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7566483

Shortest Job First:

The Shortest Job First served (SJF) is an ideal scheduling algorithm that reduces average waiting time for available processes. However, it has more waiting time
for processes that require more time to complete their execution compared to FCFS. Round robin provides faire CPU sharing to all available processes, with
processes with short burst time completing their execution in a single quantum. Priority-based scheduling also introduces starvation, as processes with high
priority execute first, causing starvation for lower priority processes. The proposed algorithm, an extension of preemptive SJF, provides a reduction in context
switching with a small change in average waiting time, thereby increasing system performance.

https://www.researchgate.net/profile/Inayat-Ur-Rehman-2/publication/
286452857_An_Optimized_Shortest_job_first_Scheduling_Algorithm_for_CPU_Scheduling/links/566a8e6c08aea0892c4a11e6/An-Optimized-Shortest-
job-first-Scheduling-Algorithm-for-CPU-Scheduling.pdf
Shortest Remaining Time First:
This paper presents a task-scheduling policy based on SRJF and evaluates it in a Cloud simulator with tasks. The algorithm's effectiveness is demonstrated
through experiments. To define a policy, one must extend VmScheduler or Cloudlet Scheduler, create sharing methods, and pass a new class during object
construction. Cloud computing is a vast concept, and there is room for improvement in scheduling algorithms. Performance can be enhanced by varying
parameters.

Volume_15__Issue_06__June_2015_IJCSMSJune2015_65_68_Mohammed-libre.pdf (d1wqtxts1xzle7.cloudfront.net)

Weighted Fair Query:

The study focuses on optimizing virtual machines (VMs) and job assignments on heterogeneous cloud resources. Instead of a "one-solution-fits-all" approach, a
customized scheduling policy based on task characteristics and cloud resource capacity is needed. The researchers developed a resource allocation algorithm
using weighted fair-share queues for on-demand resource provisioning. The algorithm allows high priority jobs with aggressive deadlines to advance time-
sharing while demoting lower priority jobs. The algorithm outperforms standard scheduling policies on six different workload scenarios. The solution uses a
three-leveled queue with tuning parameters for fine-grained control of high priority jobs, additional adjuster knobs, and a combination of strategies. The results
show the success of the approach compared to standard scheduling policies like FCFS and SJF. The researchers plan to extend their work by adding a policy
library that can save past assignments and reuse them based on incoming job patterns.

https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8648643

Ant Colony optimization:

The LBACO algorithm is proposed for tasks scheduling with load balancing, outperforming FCFS and ACO algorithms in cloud computing environments.
Experimental results show that LBACO effectively balances system load, handling all task sizes. Future work should consider the assumption of mutually
independent tasks and computationally intensive tasks, as well as extending the availability vector to accommodate heterogeneous processing in cloud systems.

https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6051750

Genetic Algorithm:
The paper presents a task scheduling model in Cloud computing that uses the GA scheduling function to create task schedules based on task and virtual machine
information. The GA function evaluates the population using fitness functions, iterating to produce the best schedule. The model outperforms existing models
like round-robin, load index-based, and activity-based costing-based.

14-libre.pdf (d1wqtxts1xzle7.cloudfront.net)

Least Slack Time:

This paper proposes a preemptive scheduler for MapReduce based on deadline and slack time to improve Hadoop cluster performance. The least slack time
preemptive deadline constraint scheduler (LSTPD) scheme avoids unnecessary preemption and improves schedulability based on available resources. However,
constraints such as runtime estimation, filter ratio estimation, data distribution, and dynamic data replication are not included. Future work will address these
issues and consider factors affecting scheduling performance in heterogeneous MapReduce systems.

https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9117131

Earliest Deadline First:

The paper proposes a new task scheduling method that introduces a waiting queue for pre-empted tasks. This method reduces the waiting time of pre-empted
tasks by placing them in a waiting queue and transferring them to another virtual machine. The queue is arranged according to priority, allowing for efficient
scheduling and reducing the execution time of pre-empted tasks.

https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6906659

Earliest Finish Time:

This paper proposes an enhancement of the Heterogeneous Earliest Finish Time algorithm (E-HEFT) for efficient task scheduling in cloud computing. The
algorithm assigns a utilization threshold for each virtual machine, specifies dataset dependencies, and sorts tasks based on the HEFT algorithm's rank. Matching
game theory is used for task execution. The algorithm balances load and takes less time to execute tasks compared to existing techniques. Future work will
consider execution cost and real-world cloud environments.

https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8514405
Priority-Based Job Scheduling:
This paper proposes a priority-based job scheduling algorithm called "PJSC" for cloud environments. The algorithm addresses issues like complexity,
consistency, and finish time. The paper surveys existing priority-based job scheduling algorithms and compares their parameters like cost, makespan, resource
utilization, and scalability. The results show reasonable complexity, but further improvement is considered for reducing finish time.

https://www.academia.edu/download/89970279/Priority-Based-Job-Scheduling-Techniques-In-Cloud-Computing-A-Systematic-
Review.pdf

Min Min:
Load Balanced Min-min is a scheduling algorithm that uses the advantages of Max-min and Min-min algorithms, but it may not produce better makespan or
resource utilization.This algorithm works on strategy in which task having minimum execution time is selected for all task.Enhanced Load Balanced Min-min, on
the other hand, selects tasks with maximum completion time and assigns them to appropriate resources, resulting in better makespan and resource utilization.

https://www.sciencedirect.com/science/article/pii/S1877050915019146

Hybrid Approaches:
This paper surveys hybrid scheduling algorithms in cloud computing, combining existing or modified PSO, GA, and ACO versions. Most hybrid algorithms use
makespan and cost as objective functions, with CloudSim simulator being the most popular for simulation results. Future research should focus on multi-
objective versions and dependent tasks, as well as real-world testing and simulation using different combinations of meta-heuristics algorithms.
https://core.ac.uk/download/pdf/478033469.pdf
Algorithm Latency Security Cost Energy Remarks

Round Robin Generally Minimal Low cost Moderate Fair distribution of tasks, may not be the most efficient
(RR) moderate considerations consumption, not
optimized for
efficiency

Multilevel Moderate Moderate Moderate Moderate Balances responsiveness for short tasks and fairness for longer tasks.
Feedback
Queue

Shortest Job Optimized for Similar to RR Moderate, Can reduce energy May lead to task starvation, not ideal for all scenarios
First (SJF) minimizing latency and FCFS depending on consumption by
task lengths completing tasks
quickly

Similar to SJN but Similar to Similar to SJN Can be more Can lead to task preemption, introducing overhead
more dynamic SJN energy-efficient
Shortest due to dynamic
Remaining adjustments
Time
Algorithm Latency Security Cost Energy Remarks

First(SRT)

Weighted Fair Can be optimized Similar to RR Moderate, Moderate, Fair distribution based on weights, but may require complex implementation
Queuing (WFQ) based on weights and FCFS depending on depending on
weight weights and task
assignment distribution

Ant Colony Can optimize Similar to RR Moderate to high Moderate, Can adapt to changing network conditions and provide near-optimal solutions
Optimization latency and FCFS depending on the
(ACO) unless efficiency of
additional pheromone-guided
security allocation
measures are
integrated

Genetic Can provide near- Similar to RR Moderate to high Moderate, Can handle complex optimization problems, but may require significant computational resour
Algorithm (GA) optimal solutions and FCFS depending on the
unless efficiency of
additional genetic algorithm
security operations
measures are
integrated
Algorithm Latency Security Cost Energy Remarks

Least Slack Similar to SJN but Similar to Moderate Can reduce energy Considers both task lengths and deadlines, minimizing slack time
Time (LST) considers deadlines SJN consumption

Earliest Prioritizes tasks Similar to Moderate Can reduce energy Potentially sacrifices fairness for meeting deadlines
Deadline First based on their SJN consumption
(EDF) deadlines

Earliest Finish Prioritizes tasks Similar to Moderate Moderate, Schedules tasks based on their estimated completion times, aiming to minimize overall compl
Time (EFT) based on their SJN depending on task
estimated finish lengths and task
times distribution
Algorithm Latency Security Cost Energy Remarks

Hybrid N/A N/A N/A N/A Combine multiple algorithms or techniques to optimize various metrics
Approaches

Priority- Depends on Depends Depends on Depends on Prioritizes tasks based on assigned priorities
Based Job priority settings on implementatio implementation
Scheduling implement n
ation

Min-Min Moderate Moderate Low to Moderate Identifies minimum execution time tasks and schedules them first.
Moderate