Article Preview
Top1. Introduction
Cloud computing brought a lot of advantages such as remote computation, data processing and storing in a way that can be accessed from anywhere using any web-enabled devices (Mlgheit et al., 2017; Banyal et al., 2013). Not to mention that many Cloud services providers standardized their platforms to provide computing resources on-demand without the need for too much administration or configuration. Along with all these advantages, there are still some security concerns that halter the migration of some users to the Cloud (Mlgheit et al., 2018).
The authentication technique is an important factor in the security of the cloud, and it is the first step for protecting the data and the privacy of the user (Velciu et al., 2014). However, Yassin et al. (2012) mentioned that 75% of web applications use the username/password-based authentication technique. those techniques proved the lack of security, due to the fact, that the password can be stolen, lost or even guessed. Therefore, compromising the authentication technique will give the attacker the right to use and/or manipulate the users' data. Many authentication techniques use biometrics features due to their uniqueness and being hard to forge (Velciu et al., 2014).
In general, biometric means the use of the biological features to define the user, such as (face, palm print, fingerprint, and voice). Biometric can be categorized into two groups: behavioral and physiological. Behavioral means identifying the user based on his behavior like keystroke and voice, while physiological means identifying the user using his/her physical body features like the fingerprint, iris, etc. Particularly, the voice and speech recognition has met quite an interest in the business field like in (amazon ALEXA, apple SIRI and windows CORTANA).
At this point, the authentication techniques have been classified as Unimodal and Multimodal techniques, where the Unimodal techniques are all the ones that adopt one method for authenticating the user whether it is Biometric-based or non-Biometric, while the Multimodal techniques are those that adopt more than one level of authentication (Gupta, 2017).
Based on a comprehensive review of both Uni/Multi-modal techniques, it is noted that the multi-modal techniques increased the security but on the favor of usability and satisfactory level of the users such in Lupu (2018), Pradhan et al. (2018), Miguel-Hurtado et al. (2017) and Matsuo et al. (2018). This is due to the obligation of using special sensors to acquire the Biometric data like fingerprint sensor, iris sensor, special HD camera, and microphone, which may not be always available on the user side. On the other side, unimodal biometric techniques are based on the assumption that biometrics are secure and can’t be stolen like Islam et al. (2009) and Ranjan et al. (2013), which is not always true as proven through numerous of recent attacks. Nevertheless, although the username/password-based authentication technique possesses the highest usability rate, it always suffers from security issues and can be forgotten with the continuous change. The need for an efficient authentication technique featured by both high usability and security becomes crucial especially with the continuous increases in the amount of data on the Cloud.
The biometric that uses the voice of the user as a characteristic for the identification and verification is called voice biometric authentication. Specifically, Voice recognition met an equal interest in the research field for its advantages and scalability. Not to mention, its increasing interest in the market field like in: (telephone banking, call centers, government administration, etc...). Precisely, voice biometric is the exclusive feature for the user that is related to the human vocal tract anatomy, which adds a large spectrum of possibilities, unlike the fingerprint and iris that are constant and easy to be compromised. Moreover, the features of the voice biometrics are unique as the fingerprint features, which can be used to identify and verify the user.
The usage of voice in authentication has not been yet explored as it should be, where new techniques can be proposed to provide more secure advantages for the Cloud users and data protection all over the web.