Time-Series Load Prediction and Deep Deterministic Policy Gradient-Based Task Scheduling Optimization for Cloud-Edge Collaborative Networks

Abstract

Cloud-edge collaborative networks have become an important computing paradigm for latency-sensitive and resource-intensive applications, but dynamic workload variation makes efficient task scheduling difficult. Existing scheduling methods often rely on current system states and cannot proactively respond to future load fluctuations, leading to congestion, delayed execution, and unbalanced resource utilization. This paper proposes PL-DDPG, a prediction-enhanced load-aware deep deterministic policy gradient algorithm for cloud-edge task scheduling. The method first predicts multi-horizon edge load trajectories from historical workload sequences, then integrates predicted loads, current observations, task features, and prediction residuals into an augmented reinforcement learning state. A deterministic actor-critic scheduler generates continuous task assignment, CPU allocation, and bandwidth allocation decisions, while a feasibility projection layer and risk-aware reward improve constraint satisfaction and service reliability. Experiments on four public workload datasets show that PL-DDPG consistently improves delay, energy efficiency, SLA satisfaction, load balancing, and scheduling stability. These results demonstrate that prediction-enhanced continuous control provides an effective solution for proactive cloud-edge resource orchestration.