Minimum Response Time Optimization in Cloud-Based IoT Systems

Abstract

 Cloud computing and IoT technologies are evolving quickly, leading to a growing need for intelligent systems that can process the dynamic workloads with minimal latency. However, effective response time and resource utilization are still huge challenges in cloud-IoT environments. This study suggests a minimum response time optimization framework for cloud-based IoT systems to achieve latency-aware resource scheduling and adaptive optimization under dynamic workloads. Data preprocessing, feature engineering, exploratory data analysis, and normalization follow the utilization of the Multi-Tier IoT Resource Allocation Dataset from Kaggle. The metrics MAE, MSE, RMSE, and R² are used for training and evaluation of Random Forest (RF), Gradient Boosting (GB), Stochastic Gradient Descent (SGD), and Deep Neural Network (DNN). Experimental outcomes show that the DNN model outperforms the base models including RF, GB, and SGD with an MAE of 5.9605 ms and an R² score of 0.9830. Results show that the suggested strategy is more efficient and performs better overall in terms of response time minimization, latency optimization, throughput improvement, energy efficiency, and SLA-aware QoS enhancement. The novelty of this work is in combining the three elements of response time prediction, adaptive resource allocation, and optimization analysis in a single framework, which can effectively enhance the performance and reliability of cloud-based IoT systems.