Abstract
The paper proposed a network scheduling in cloud computing based on intelligence Particle Swarm Optimization algorithm aimed at the disadvantages of cloud computing network scheduling. Firstly, on the basis of cloud model, used intelligence Particle Swarm Optimization algorithm with strong ability of global searching to find the better solution of cloud computing network scheduling then turned the better solution into the initial pheromone of improved Particle Swarm Optimization algorithm, and found out the cloud computing network scheduling and the algorithm's global optimal solution through improved Particle Swarm Optimization information communications and feedbacks. Finally, made comparison test of the three benchmark function on the basis of MATLAB, the results showed, compared with traditional intelligence Particle Swarm Optimization algorithms, the improved algorithm can preferably allocate the resources in cloud computing model, the effect of prediction model time is more close to actual time, can efficiently limit the possibility of falling into local convergence, the optimal solution's time of objective function value is shorten which meet the user's needs more.
Introduction
Cloud computing relies on the efficient distribution of tasks across a massive pool of virtualized resources. Poor scheduling leads to "hotspots" (overloaded servers) and "cold spots" (idle hardware), significantly impacting response times and operational costs. This resource allocation is a classic NP-hard problem; as the number of tasks and nodes increases, the search space for an optimal schedule expands exponentially. Traditional schedulers often rely on simple priority queues, which fail to account for the dynamic and uncertain nature of network conditions.
Methodology
The proposed approach utilizes a multi-layered optimization strategy combining a mathematical "Cloud Model" with Quantum-behaved Particle Swarm Optimization (QPSO).
The Cloud Model
The Cloud Model serves as a bridge between qualitative concepts (e.g., "high load") and quantitative data. It utilizes "cloud drops" characterized by Expectation (Ex), Entropy (En), and Hyper-entropy (He) to represent the inherent uncertainty and randomness in network traffic.
Quantum Particle Swarm Optimization (QPSO)
Standard Particle Swarm Optimization (PSO) mimics social behavior like bird flocking. However, it is prone to "premature convergence," where the swarm gets trapped in a local minimum. To address this, the authors implement QPSO, where particles are governed by wave functions rather than fixed velocity vectors. This allows particles to "tunnel" through barriers in the search space, ensuring a more robust global search. The best solutions from the initial global search are used to initialize pheromone levels for subsequent optimization stages, effectively hybridizing PSO with Ant Colony Optimization principles.
Key Results
The algorithm was evaluated using MATLAB simulations against three benchmark functions: Sphere, Goldstein-Price, and Ackley. These functions provide complex mathematical landscapes to test global optimization capabilities.
As shown in the table above, the improved algorithm consistently outperformed standard intelligence PSO:
- Precision: On the Sphere function, the proposed algorithm achieved a best solution of 0.00196, an order of magnitude better than the standard PSO's 0.02125.
- Stability: For the Goldstein-Price function, the "worst" solution of the new algorithm (3.00003) was superior to the "best" solution of the standard version (3.00051).
- Efficiency: Simulation results indicated that as the number of sub-networks increased, the gap between predicted model time and actual execution time narrowed, demonstrating improved scalability.
Conclusion
The integration of Cloud Models and QPSO provides a more resilient and precise framework for cloud resource scheduling. By accounting for network uncertainty and leveraging quantum-inspired search mechanics, the proposed algorithm avoids local convergence and optimizes task allocation more effectively than traditional swarm intelligence methods. Future research should focus on validating these results in live, heterogeneous cloud environments to account for real-world hardware failures and latency.
Summary updated