Article Preview
Top1. Introduction
Nowadays, mobile devices, such as smartphones and tablets, have gained enormous popularity. Many sophisticated mobile applications have boosted opportunities for increased efficiency and real innovation in several domains, such as medical, face recognition and mobile augmented reality domains (Veazie et al., 2018). These applications require high computational resources to run intensive tasks. Due to computing and storage resource limitations, mobile devices migrate these applications to rich-resources on the cloud infrastructure. Thus, running mobile applications on the cloud servers addresses the mobile device resource restrictions, which in turn, have produced the tremendous paradigm of Mobile Cloud Computing (MCC) (Mollah, Azad, & Vasilakos, 2017; Zhou & Buyya, 2018).
The cloud enables on-demand access to a shared pool of virtualized computing resources that users can exploit for handling and deploying their mobile device applications (Vaezi & Zhang, 2017). Cloud computing allows users to leverage virtualization technology, in which server hardware resources are divided into multiple virtual machines (VMs) for mobile users, which also enable several executions to be carried out at the same time. It permits offloading of all types of computational tasks on a VM (phone clone) system level. Each VM can execute some computational-intensive tasks on behalf of the mobile device, and then, can send the result back (Huang & Wu, 2017). For cloud providers, virtualization helps them to increase hardware utilization efficiency. For mobile users, virtualization allows outsourcing the maintenance of mobile devices resources, enables easy data sharing, increases scalability and provides on-demand resources (Huang & Wu, 2017).
The users’ VMs are isolated from each other even when they run on the same cloud physical machines in order to avoid sharing VMs’ sensitive data, which may risk exposing confidential data. Unfortunately, attackers are finding new ways to circumvent weaker isolation of VMs, such as damaging the data confidentiality of VMs and violating privacy by building attackers' channels (Mollah et al., 2017). Researchers have shown that a malicious user can break the isolation by building many side-channels between VMs for obtaining private data that are processed on users’ VMs. The malicious users’ goal is to deliver their VMs on the same server that allocates legitimate users’ VMs in order to co-locate and steal their information (Ristenpart, Tromer, Shacham, & Savage, 2009; Wang & Lee, 2007; Wang & Lee, 2006; Aviram, Hu, & Ford, 2010; Vattikonda, 2011; Wu, Ding, Lin, & Wang, 2012; Han, Chan, Alpcan, & Leckie, 2015; Idrissi, Ennahbaoui, Souidi, & Hajji, 2015; AbdElnapi et al. 2016; Zhang et al. 2017; Gutub & AlKhodaidi, 2020; Jyothi, Bhargavi, Mani, Kumari & Lydia, 2020). With the motive of securing VMs, the hypervisor is the software level that creates, runs and controls the VMs. This layer resides between the hardware and the users’ VMs. However, researchers (Sgandurra & Lupu, 2016) have proven that a malicious user can circumvent the hypervisor and allocate their malicious VMs. Other issues that virtualization brings when applied to MCC environment are unauthorized access of users to cloud services and VMs communication within a virtualized environment, which lead to the risk of losing users' data in cloud (Modi & Acha, 2017; Mollah et al., 2017).