Adaptive cost and energy aware secure peer-to-peer computational offloading in the edge-cloud enabled healthcare system

Now-a-days computational offloading and quick disease prediction are identified as major difficult troubles of smart healthcare systems. As cloud based healthcare is primarily far away from remote patients which leads to high latency, bandwidth and computational usage at some point in real-time monitoring and diagnosis. Due to the expanded demand of real-time healthcare applications, there's a want for higher quality of experience to carry out the disease prediction undertaking via low latency computational processing. So, the proposed research studies introduces an edge-cloud enabled healthcare system to carry out quick disease prediction in peer-to-peer edge-cloud integrated computing platform by minimizing the latency, bandwidth and computational power metrics. Consequently, a brand new fashion of adaptive cost and energy aware computational offloading scheme and privacy preserving communication protocol using hybrid encryption scheme are introduced. It will perform the real-time secure data offloading and processing among the edge nodes from the region where the affected patient records were initially collected. Also, an Adaptive Fuzzy Optimized k-Nearest Neighbor (AFO-k-NN) classifier model is introduced to predict Parkinson disease and primarily based on severity further diagnosis and rehabilitation process might be achieved by the healthcare system. Experimental evaluation on various offloading procedures and classifier models are made with respect to energy consumption, utility cost, response time, prediction time and prediction accuracy. Our expertise best, the proposed offloading scheme takes very less energy consumption, utility cost and response time while comparing to current cost-based, energy-based and learning-based schemes. In addition, the proposed privacy preserving communication protocol provides more privacy and security during offloading and sharing of patient data. Similarly, the proposed classifier model obtains less prediction time and greater prediction accuracy at the same time as compared to current classifier models.