DYNAMIC AND ADAPTIVE REINFORCED MACHINE LEARNING FOR IMPROVED EFFICIENCY AND RELIABLE ROUTING IN SENSOR-CLOUD ENVIRONMENT

Abstract

Sensor-Cloud and Industrial Internet of Things (IIoT) architectures have brought a phenomenal change in industrial sector particularly in hospital industry both for indoor and outdoor applications. Both are boosting the hospital industry sector by taking into account the concept of smart city, IoT and communication. Many techniques and algorithms are applied to improve network reliability and to reduce latency that eventually help optimized energy consumption and increase in network lifetime. There are problems with network failures that is required to be addressed keeping in mind the dynamics and limiting constraints of the network. Under network breakdown conditions, an alternative route must be selected to have a reliable end-to-end information delivery. This work proposes a reinforced machine learning with dynamic and adaptive approach (DAML), a low-latency reliable routing algorithm for cluster based hospital environment so that a patient’s critical parameters are monitored continuously. The DAML method comes with optimized reward, monitor, and risk factor concepts and the patient’s critical parameters are interfaced to the sensor-cloud environment. The information is buffered as text files and then transmitted over wireless channels after encoding. The proposed method is evaluated with respect to network throughput, packet delivery ratio (PDR), and end-to-end delay. The proposed method is expected to provide a significant improvement in throughput, PDR and decrease in end-to-end delay as compared to existing approaches.