Towards healthcare system integrity using fault-tolerant-based scheduling in edge data center
Masnida Binti Hussin, A. I. Muhammad · AIP conference proceedings · 2022
Effective healthcare services are one of the vital issues nowadays that required ICT for better management. It is because the volume of data in healthcare services is growing rapidly where the physical computing resources are not anymore enough to handle it. Cloud computing by its adaptive infrastructure is helping healthcare organizations to effectively manage large volumes of data, also supporting better clinical responses. Such computing employed a confederation data center for handling massive users’ requests through edge computing where network devices can improve data processing, analysis speed, and reduce network costs. But the issue arises when it comes to unpredictable deficiencies in network performance such as increases in delay and overhead. It needs a better provisioning mechanism for isolating the deficiency. Due to the healthcare data is comprised of sensitive and non-sensitive data hence the fault isolation process needs to thoroughly manage. It means to sustain dataset privacy and trustworthiness. In this work, the fault-tolerant scheduling approach is proposed for realizing consistency in the data processing. It aims to ensure any network performance issues will not be affected by users’ requests. Our edge computing environment is designed by using the multi-tenant model where it represents diversity in healthcare organizations. The integration between Cloud and edge computing in our environment represents global service (friend) and local service (family). Our fault-tolerant family-and-friend (FT-FnF) scheduling approach is used fuzzy logic for realizing rescheduling and re-allocating decisions. The square matrix multiplication is employed to develop the logic structure and related migration threshold. This strategy intends to identify the suitable nodes prior that to be used as alternative edge processing or storage. By incorporating such strategies, fault isolation and tolerance able to be done effectively while enduring data integrity. Our experimental results show better execution time and computing overhead compared to the existing algorithm. Implicitly, our fault-tolerant approach improves system performance and able to deal with large-scale healthcare communities.