Table 1 Summary of findings from the literature.

Zhao Tong et al.¹⁸

DQN-based offloading algorithm with principal component analysis weighting method (D2OP) Architecture

Enhances scheduling efficiency via dynamic task and data movement optimization.

Improved time reduction, load balancing, and task success ratio compared to three similar algorithms.

IoT

Dynamic

D2OP

Computationally expensive due to DQN updates; may struggle with scalability in larger IoT environments.

Hamdi Kchaou et al.¹⁹

Fuzzy Clustering with PSO

Combines fuzzy clustering with PSO for efficient data location and job scheduling.

Reduced data transfer overhead during task execution.

IoT

Static

PSO

Static scheduling limits adaptability to dynamic workloads; may fail to respond efficiently in real-time scenarios.

Prasanta Kumar Bal et al.²⁰

Enhanced Cat Swarm Optimization

Focused on reducing make-span and increasing throughput.

Improved resource usage, energy efficiency, and response time.

Cloud

Dynamic

Cat Swarm Optimization

Requires high computational power for parameter tuning; effectiveness can drop with increased task complexity.

Neetu Sharma et al.²¹

Neural Network-Aided Ant Colony Optimization

Modified ACO for optimal global exploration.

Increased availability times, regular task allocations, and statistical improvement over traditional methods.

Cloud

Static

ACO

Neural network integration increases overhead; may face difficulties with large-scale task networks or highly dynamic environments.

Xiaokang Zhou et al²³.

Deep Reinforcement Learning Two-Stage Scheduling (DRL-TSS)

Uses deep RL to optimize task scheduling in edge-enabled IoE environments.

Enhanced learning efficiency and scheduling performance by 1.1x for specific IoE applications.

IoE

Dynamic

DRL-TSS

High training complexity and potential convergence issues in large-scale IoE systems; may require extensive parameter tuning.

Sahar Badri et al²³.

CNN-Optimized Butterfly Algorithm

Combines CNN with modified butterfly optimization for efficient cloud job scheduling.

Maximized throughput and reduced make-span in job scheduling.

Cloud

Dynamic

CNN

Integration of CNN adds computational overhead; performance might degrade with resource-constrained edge devices.

Xiuhong Li et al.²⁴

Unified Resource Management Architecture

Employs dynamic task scheduling to address resource consumption and task prioritization issues.

Near-optimal task execution performance with high system productivity and profitability.

Cloud

Dynamic

Unified Resource Management Architecture

Complexity increases with the number of tasks and resources; potential bottlenecks in real-time resource reallocation.

N. R. Rajalakshmi et al.²⁴

Parallel Task Scheduling in the Cloud

Proposes a mathematical model to allocate workloads among multiple virtual machines.

Efficient completion of parallel tasks within deadlines, reducing operational costs.

Cloud

Static

Parallel Task Scheduling

Static nature limits scalability and adaptability to fluctuating workloads; lacks energy efficiency considerations.

Zubair Sharif et al.²⁵

Priority-Based Task Scheduling (PTS-RA)

Prioritizes tasks based on urgency in a health monitoring system using mobile edge computing.

Reduced task completion time and data transmission overhead.

Mobile Edge Computing

Dynamic

PTS-RA

Task prioritization might overlook long-term resource optimization, leading to potential under-utilization of edge resources.

Rongli Chen et al.²⁶

Energy-Efficient Cloud Task Management

Breaks tasks into sub-tasks allocated to virtual machines to minimize energy consumption.

Reduced energy usage and job wastage in cloud computing environments.

Cloud

Dynamic

Energy-Efficient Task Management

Task splitting may introduce communication overhead and latency; the number of virtual machines available for task distribution limits scalability.