The rapid growth and widespread adoption of cloud computing have led to significant electricity costs and environmental impacts. Traditional approaches that rely on static utilization thresholds are ineffective in dynamic cloud environments, and simply consolidating virtual machines (VMs) to minimize energy costs does not necessarily result in the lowest carbon footprints. In this paper, a deep reinforcement learning (DRL) based framework called CFWS is proposed to enhance the energy efficiency of renewable energy sources (RES) supplied data centers (DCs). CFWS incorporates an adaptive thresholds adjustment method TCN-MAD by evaluating the predicted probability of a physical machine (PM) being overloaded to prevent unnecessary VM migrations and mitigate service level agreement (SLA) violations due to imbalanced workload distribution. Additionally, CFWS introduces a novel action space in the DRL algorithm by representing VM migrations among geo-distributed cloud data centers as flattened indices to accelerate its execution efficiency. Simulation results demonstrate that CFWS can achieve a superior optimization of energy costs and carbon footprints, saving 5.67% to 13.22% brown energy with maximized RES utilization. Furthermore, CFWS reduces VM migrations by up to 86.53% and maintains the lowest SLA violations within suboptimal execution time in comparison to the state-of-art algorithms.

The widespread adoption of cloud data centers has led to a rise in energy consumption, with the associated carbon emissions posing a further threat to the environment. Cloud providers are increasingly moving towards sustainable data centers powered by renewable energy sources (RES). The existing approaches fail to efficiently coordinate IT and cooling resources in such data centers due to the intermittent nature of RES and the complexity of state and action spaces among different devices, resulting in poor holistic energy efficiency. In this paper, a reinforcement learning (RL) based framework is proposed to optimize the holistic energy consumption of sustainable cloud data centers. First, a joint prediction method MTL-LSTM is developed to accurately evaluate both energy consumption and thermal status of each physical machine (PM) under different optimization scenarios to improve the state space information of the RL algorithm. Then, this framework designs a novel energy-aware approach named BayesDDQN, which leverages Bayesian optimization to synchronize the adjustments of VM migration and cooling parameter within the hybrid action space of the Double Deep Q-Network (DDQN) for achieving the holistic energy optimization. Moverover, the pre-cooling technology is integrated to further alleviate hotspot by making full use of RES. Experimental results demonstrate that the proposed RL-based framework achieves an average reduction of 2.83% in holistic energy consumption and 4.74% in brown energy, which also reduces cooling energy consumption by 13.48% with minimal occurrences of hotspots. Furthermore, the proposed MTL-LSTM method reduces the root mean square error (RMSE) of energy consumption and inlet temperature predictions by nearly half compared to LSTM and XGBoost.

Combinatorial testing (CT) can efficiently detect failures caused by interactions of parameters of software under test. The CT study has undergone a transition from traditional CT to constrained CT, which is crucial for real‐world systems testing. Under this scenario, constrained covering array generation (CCAG), a vital combinatorial optimisation issue targeted with constructing a test suite of minimal size while properly addressing constraints, remains challenging in CT. To the authors’ best knowledge, this paper presents a synergic method first based on quantum particle swarm optimisation (QPSO) for the CCAG problems. Three auxiliary strategies, including contraction‐expansion coefficient adaptive change strategy, differential evolution strategy, and discretisation strategy, are proposed to improve the performance of QPSO. Meanwhile, the improved QPSO method combines with the three different constraint handling strategies and an enhanced one‐test‐at‐a‐time strategy as a synergic QPSO method named QPIO to solve the CCAG problem. In the experiment, we investigate the impacts of parameter settings on the performance of the QPIO. Extensive experimental results show that the QPIO algorithm is a competitive method compared to the representative methods for CCAG. Besides, the QPIO method enriches the application of the QPSO algorithm in the context of CT.