Abstract:

On March 11, 2011, eastern Japan experienced a catastrophic earthquake and tsunami, from which the east coast suffered severe damage. The tsunami caused significant damage to the infrastructure, including several educational assistance systems built by universities. Other recent significant earthquakes include the Kumamoto earthquake, which occurred in the northern Kyushu region in 2017, and the large-scale earthquake which occurred in southern Hokkaido in 2018. Japan has also suffered serious weather-related disasters. During July 1–10, 2018, a special heavy-rain warning was issued for 10 prefectures in western Japan. Much infrastructure was destroyed and power grids and communication networks were shut down. For this reason, troubles occurred in the operation of many educational assistance systems.

Moreover, several factors serve to create instability in East Asia’s security. Many missile tests were conducted on the Korean Peninsula in 2017. If missiles were to fall in metropolitan areas, the damage could reach an almost unimaginable level. Therefore, the threats we must address have diversified, as must the field of information systems and the educational system.

The current educational environment cannot be realized without educational assistance systems, such as learning management systems. Students’ educational records are stored in these educational assistance systems. If an educational assistance system, along with students’ records, was lost due to a crisis, it could represent a loss of sustainability for educational activity. Therefore, a plan must be devised to protect educational assistance systems and records in the event of crises. However, an effective protection plan must function differently in response to the characteristics of various disasters or crises.

In this study, we built an adaptive risk management framework for e-learning systems. In particular, in order to protect educational support systems from multiple threats, this study clarifies effective protective mechanisms and proposes an algorithm to process container-based virtual machines with multipoint cloud providers. This algorithm uses machine learning to identify natural disasters and the history of regional security emergencies to determine how to cope with specific crises. This algorithm makes it possible to adaptively determine system-operation strategies according to the types of disasters and crises rather than using simple rule-based judgment criteria. Thus, this algorithm could help sustain educational assistance systems by adaptively protecting private cloud infrastructure and public cloud services.

The paper first describes the impact of frequent natural disasters and dynamic regional security threats on educational support systems and clarifies our approach to these issues. Second, we describe related studies describing the disaster recovery methods used by hybrid cloud architecture and identify problems. Third, we describe the design of the proposed algorithm using disaster histories and machine-learning analytics. Finally, we describe the results of experimental use of a prototype system employing the proposed algorithm and analyze its effectiveness.

Keywords:

e-learning infrastructure, crisis management, container-based virtualization, emergency alert, machine learning.