Deep Reinforcement Learning Based Intelligent Resource Allocation Techniques with Applications to Cloud Computing

Resources allocation in cloud computing needs an efficient and accurate prediction of network workload. To predict the high dimensional and high variable data is difficult to predict. It leads to failing services level agreement and quality of services. To address these challenges deep reinforcement learning techniques are used for predicting the heterogeneous workload on the network. This paper presents a comprehensive review of deep reinforcement learning techniques. It includes the use of deep reinforcement learning in the field of resource allocation, task scheduling, traffic identification, future prediction and to fulfil the QoS and SLA. Some parameters such as SLA violation, Resource Allocation (RA), cost, the response time (RT), and resource prediction (RP) are also discussed. A case study related to deep reinforcement learning for automated task scheduling in cloud computing is also included which encourages us to use these techniques for task scheduling without any response delay and proper resource utilization.