DeepSeaNet: A Multi-Modal Deep Learning Framework for Reliable Data Transmission and Event Detection in Underwater IoT Environments
Nellore Kapileswar, Judy Simon
Abstract
With cloud computing at the forefront of digital services, resource management needs to be done well and at a high capacity to handle changing user requests. Using traditional and fixed resource allocation strategies does not allow for changes in the workload, which often leads to either wasting resources or giving more than is needed. To solve these challenges, this paper proposes AutoScaleNet, a brand new DRL framework built for managing resources in cloud computing. The PPO approach in AutoScaleNet lets it learn proper scaling policies that balance cost with the need for a high level of service quality. The framework can adapt its resources to changes in demand, which lowers the number of SLA disagreements and lowers operation costs. Results from experiments on benchmark cloud environments show that AutoScaleNet is better than both traditional rules and deep Q-Iearning when it comes to using resources, speed of response, and energy costs. This research allows for intelligent and adaptive handling of cloud resources by using the latest DRL techniques.