Distance education techniques are developing at a rapid pace. Often, program designers are fascinated by the latest technology without addressing the underlying issues related to the needs of learners. This type of system could guarantee equal access to systems in the educational process and adapt to the new roles of the training institution and the learner. An important point is the trust, the integrity and the authenticity of assessment results exams of the distance learning process. Higher education encourages schools and universities to struggle against online exam cheats more effectively and is more concerned with the quality and integrity of distance learning.Online proctoring includes all automated processes which help to make remote assessment systems more secure. Indeed, several organizations finally recognize the weaknesses of traditional surveillance security. The rapid growth of new technologies that an increasing number of people around the world can access has enabled the availability and the proliferation of online training offerings. At the same time, this rapid development undoubtedly increased the number of fraud attempts. In addition, there are various possible attacks threatening the privacy of individuals. The learner may face identity spoof (an attacker can use various methods to steal his identity) or an attempt to perform fraudulent actions, it is therefore obligatory to counter them and to define the necessary requirements for the management of the exams and the protection of personal data.The aim of this thesis is to meet the need for monitoring a remote examination by integrating a solution based on current technology with different biometric modalities and the implementation of an automated proctoring system to monitor and verify the identity of learners during an online exam. In this work, we have implemented an innovative multimodal biometric application with two modalities: keystroke dynamics and facial recognition in order to verify the identity of learners during online exams.Biometric is an active research topic because of its utility for a wide range of applications, it refers to the analysis of physiological or behavioral characteristics of an individual for the purpose of authentication or identification. Our biometric system based on the collection and processing of personal data complies with the European General Data Protection Regulation (GDPR). An identity detection and verification module have been integrated and used on an industrial scale for several remote exams. We have also implemented a fraud detection system to assess and monitor the learners' environment during a remote examination. Indeed, several techniques based on image analysis, sound signal processing and keyboard events have been used to detect fraudulent actions with a solution based on machine learning. A multimodal biometric system with a cryptographical tool for data protection and behavioral proctoring can play a decisive role in improving learner identity verification and detecting unusual events without compromising learner privacy. The proposed biometric system is effective against identity theft and solves a significant and a large portion of remote examination fraud with a high level of accuracy by detecting unusual behavior. We demonstrate on databases of literature and operational data sets the interest of the approach proposed in this thesis.