Deep Learning Approaches to Cloud Security. Группа авторов

this facility helps make data available and manages the cost or storage area flexibility and scalability management. It reduces the complexity of managing the database and makes it easily available to each tenant and workspace.

      2 2. One database and many schemas means all tenants use the same database in a different manner. Each tenant uses their requirement of data and every tenant requirement is different as the work is different. According the tenant, the services are provided to the tenant. The complexity and cost will also affect the structure used by the tenant.

      3 3. Many databases and schemas means, in this type of model, that tenant data is stored in the database in different locations in a different database or that the tenant can create different databases as they are required. Accordingly, the new database will create a cost and complexity increase [6].

      2.2.3 Concept of Multi-Tenancy with Cloud Computing

In the Cloud Computing system, the concept of Multi-Tenancy is very important because IT provides the facility to all tenants through shared computer resources in a cloud environment. The cloud environment has two types: a public cloud environment and a private cloud environment. The tenant chooses one of the cloud environments. All tenant data is isolated and inaccessible to each other shown in Figure 2.3.

Figure 2.3 Multi-Tenancy models [5].

      Multi-Tenancy cloud computing systems create a sparse work area for each tenant for storage or project data and login privacy policies. The tenant only uses their personal and secure area for work and accesses only know work or data. In the case of other requirements of data, another tenant’s permission or access key is required for access.

One type of cloud computing is in the private cloud environment and uses groups of more than two tenants belonging to a single company. They work on the same data and use resources through the cloud environment. The other type is the public cloud environment where many companies can share their data and services with different tenants. The public cloud environment is used more than the private cloud environment.

      The Single Tenant Cloud Computing environment does not provide to all facilities. When compared to a Multi-Tenancy system, it provides more storage, access features, and security and privacy policies.

      It provides a virtual environment for work easily, with less complexity maintaining work and hardware and no restrictions to devices or location [7].

       Example of Multi-Tenancy with Cloud Computing:

Let us take a very common example of a colony. In a colony there are many houses; each house has many flats, each house is playing a role as part of a single company, each company has different departments, each department works only at their area, and each department shares common utility services (i.e. water, electricity, PNG etc.). Similar facilities are used by companies shown in Figure 2.4.

Figure 2.4 Multi-Tenant Cloud structure [6].

2.3 Privacy in Cloud Environment Using Deep Learning

      In Multi-Tenant cloud based systems, security uses the Deep Learning Method to overcome all the requirements of a tenant, including privacy policies and services required. Using Deep Learning Methods for developing the privacy structure of Cloud Computing provides better services according to the requirements of a tenant. Accordingly, at the request of the tenant, resources and service availability fulfil requests and the privacy and security services are developed and maintained. In the cloud computing environment, there are public, private, single, and multi-tenant structures available to use according to the requirements of a tenant. Different structures have different needs for privacy and security services. If better services are not available, the tenant may not use the structure, therefore, using Deep Learning methods develops privacy and security services. The services are developed according to the organisation’s needs using Deep Learning [8].

      Privacy in Cloud Computing is maintained by a distributed system concept in a Multi-Tenancy based application using encryption techniques to protect data, synchronize work, and regulate base data modification if the tenant is authorized or not, so all factors are required to check and manage [9]. The privacy management system is used to manage the security of sophisticated data access.

The information stored in the cloud environment is directly encrypted in a format that the tenant is authorised to access and the data can only decrypt that data securely [10]. Using Deep Learning techniques increases privacy and security levels. A new protection algorithm has been developed using Deep Learning Methods for safe transformation and using data in a database in a cloud environment. In a Cloud Computing environment, a privacy service system has been developed to stop unauthorized access and increase the capabilities for the security of private data, minimizing the hacking of private data in the cloud environment shown in Figure 2.5 and Figure 2.6 [11].

Figure 2.5 Concept of Cloud Security [7].

Figure 2.6 Securities in Multi-Cloud environments [7].

2.4 Privacy in Multi-Tenancy with Deep Learning Concept

There is a need for privacy in a Multi-Tenancy system because of the risk of low privacy policies and weak security for data when tenants work on multitenant applications. The organisation and tenant are not able to self-secure all data [12]. Using Deep Learning creates a concept to secure data, providing there is a secure and private environment for the tenant to accept the Multi-Tenant application and work on that system freely. This is the reason for weak privacy and security polices, a risk of data loss, a risk of hacking of data, and the wrong use of information, so it is very necessary to secure the entire task before starting work on a Multi-Tenant system. For security or privacy, the first step is maintaining the concept of a unique ID [13]. In this model, all tenants have an individual, unique ID for login. If the organisation is very large and it is complicated to manage all the IDs, then each individual department will have a single ID to login. This concept is also used for using Deep Learning. The second step is to make access limitations for each tenant. The access of each tenant is dependent on the organisation or authorised department deciding the access limitation. For example, a company whose tenants are working in an account department are able to access only the account department data, they are not able access other departments’ (admin, security, etc.) data. According to this concept, all department access criteria is decided or fixed and department tenant access limitation is decided so the authorisation is checked and only authorised users can access the limited data. The third step is to isolate the database into tables according to department. The database is then separated and isolated into tables according to tenant access and limitation-isolated data is provided to the tenant.