Actuellement, nous vivons dans une société numérique. L'avènement de l'Internet et l'arrivée du multimédia et des supports de stockage numériques, ont transformé profondément la façon dont nous communiquons. L'image en particulier occupe une place très importante dans la communication interpersonnelle moderne. Toutefois, elle présente l'inconvénient d'être représentée par une quantité d'information très importante. De ce fait, la transmission et le stockage des images soulèvent certains problèmes qui sont liés essentiellement à la sécurité et à la compression d'images. Ce sont ces considérations qui ont guidé cette thèse. En effet, la problématique que nous posons dans cette thèse est de proposer une solution conduisant à la crypto-compression d'images afin d'assurer un archivage et un transfert sécurisés tout en conservant les performances de la méthode de compression utilisée. En effet, nos travaux de recherche ont porté essentiellement sur la compression et le cryptage des images numériques. Concernant la compression, nous avons porté un intérêt particulier au codage arithmétique vu sont efficacité en terme de ta ux de compression et son utilisation par les nouvelles normes et standards de compression tel que JPEG2000, JBIG, JBIG2 et H.264/AVC. Quant au cryptage, nous avons opté pour l'utilisation du chaos combiné avec les fractions continues afin de générer des flux de clés ayant à la fois de bonnes propriétés cryptographiques et statistiques. Ainsi, nous avons proposé deux nouvelles méthodes de compression sans perte basées sur le codage arithmétique tout en introduisant de nouveaux paramètres de codage afin de réduire davantage la taille en bits des images compressées. Deux autres méthodes s'appuient sur l'utilisation du chaos et des fractions continues pour le développement d'un générateur de nombres pseudo-aléatoires et le cryptage par flot d'images. Enfin, nous proposons une nouvelle méthode qui emploie conjointement le cryptage avec la compression. Cette dernière méthode se base sur l'échange des sous-intervalles associés aux symboles d'un codeur arit hmétique binaire de façon aléatoire tout en exploitant notre générateur de nombres pseudo-aléatoire. Elle est efficace, sécurisée et conserve le taux de compression obtenu par le codage arithmétique et ceci quelque soit le modèle statistique employé : statique ou adaptatif. / Actually, we live in a digital society. The proliferation of the Internet and the rapid progress in information technology on multimedia, have profoundly transformed the way we communicate. An enormous amount of media can be easily exchanged through the Internet and other communication networks. Digital image in particular occupies an important place in modern interpersonal communication. However, image data have special features such as bulk capacity. Thus, image security and compression issues have became exceptionally acute. It is these considerations that have guided this thesis. Thus, we propose throw this thesis to incorporating security requirements in the data compression system to ensure reasonable security without downgrading the compression performance.For lossless image compression, we have paid most attention to the arithmetic coding (AC) which has been widely used as an efficient compression algorithm in the new standards including JBIG, JBIG2, JPEG2000 and H.264/AVC. For image encryption, we are based on the combination of a chaotic system and the Engel continued fraction map to generate key-stream with both good chaotic and statistical properties. First, we have proposed two new schemes for lossless image compression based on adding new pre-treatment steps and on proposing new modeling methods to estimate probabilities for AC. Experimental results demonstrate that the proposed schemes give mean compression ratios that are significantly higher than those by the conventional AC. In addition, we have proposed a new pseudo-random bit generator (PRBG). The detailed analysis done by NIST statistical test Suite demonstrates that the proposed PRGB is suitable for cryptography. The proposed PRBG is used to develop a new symmetr ic stream cipher for image encryption. Theoretic and numerical simulation analyses indicate that our image encryption algorithm is efficient and satisfies high security. Finally, we have proposed a new scheme which performs both lossless compression and encryption of image. The lossless compression is based on the binary AC (BAC) and the encryption is based on the proposed PRBG. The numerical simulation analysis indicates that the proposed compression and encryption scheme satisfies highly security with no loss of the BAC compression efficiency.

Nowadays, network security plays a major role in the field of wireless communications. Wired networks propagate electrical signals or pulses through cables. Whereas wireless signals propagate through the air. If wireless networks are left open and exposed to the outside world, there are high chances of being misused by others. The intruders take advantage of this, to intercept the wireless signals. This is the reason why an extra level of security is required for wireless networks. The physical layer is one of the important layers of the Open System Interconnection (OSI) model which plays an important role in the network’s physical connections like wireless transmission, cabling, connections etc. The physical layer supports the bit-level transmission between various devices by connecting to the physical medium for synchronized communication.In this thesis, a method is studied for exchanging secret key [1] bits using a pseudo-random sequence generator based on Frequency Division Duplex (FDD) systems. The principle of this method is to generate a secret key in a manner that produces low correlation at the intruder. By uniquely relating the secret key bits to the channel in a private version of the universal codebook, a robust key exchange between the transmitter and the receiver is then performed.

The thesis deals with establishing uncertainties of indirect measurements. It focuses primarily on random number generators in software enabling the calculation of mea-surement uncertainties using Monte Carlo. Then it focuses on the uncertainty calculati-on indirect measurement as the Monte Carlo method and the classical numerical met-hod. The practical part deals with the verification of randomness generators numbers contained in various softwares. It also deals with the determination of uncertainties indi-rect current measurements by both above-mentioned methods and then comparing and evaluating the values achieved.