Implementation of Visual and Quantum techniques for Secure Information Flow in Content Delivery Networks

P. JAGADEESWARA RAO1, P. SUNITHA2, CH. RUPA3, V. SANDHYA4 1 2Assistant Professor, Department of computer science & engineering, 2 Scholars, JNTUK 2 3 4 Dhanekula Institute of Engineering & Technology, Vijayawada. 3Professor, VR Siddhartha Engineering College, Vijayawada.

Abstract— , a science of visual 1. INTRODUCTION cryptography, was historically originated to write Steganography is the field of research that utilizes a secret message and it was impossible to be the insensitivity of the human sense organs revealed, Though Steganography is an old (sensory redundancy) and multimedia signal concept it is a powerful one and can be practiced inherent redundancy (data redundancy) to hide where long classical secret keys need to be the secret message in a carrier signal. avoided and sense of simplicity while strong Steganography hides the fact of communicating cryptography is vital. One of the biggest and therefore it is a good choice of secure challenges of Steganography and other Visual communication. Steganography message can be Cryptography techniques is that an attacker in the retrieved by reverse Steganography middle can hijack the image and forward a fake image. This again means that we need large However one problem of Steganography and classical keys or so to secure the channel. other visual cryptography techniques [3] is that However large keys are no longer going to be there can be an eavesdropper Eve between two suitable for future next generation networks as users, Alice and Bob, and Eve can hijack the they require light weight distribution systems image taken from Alice and forward a fake image which are also free from eavesdropping. Besides, to Bob. Again it means we need a key to secure in recent studies it is suggested that in near future the channel. However if we think of symmetric quantum computers would be able to break keys, large keys are no longer going to be suitable classical cryptographic techniques such as RSA. for future next generation networks as it With quantum computers, essentially requires light weight key distribution also came into light and now seems to be systems and it also has to be free from promising based on its unique features that make eavesdropping. Besides, some recent studies such it free from eavesdropping or any other third as Shor’s algorithm also suggest that in near party intrusion in the key distribution system. Our future quantum computers can break the paper proposes a technique where we show how factorization problems in classical cryptographic an old technique such as Steganography can be techniques such as RSA [2]. promising when we combine it with the unique Following that Quantum Cryptography came key distribution mechanism of Quantum in town and now based on its unique features such Cryptography with some other additional security as non-clone ability or non-copying properties, it features. promises establishing a channel free from Keywords—Steganography, Visual Cryptography, eavesdropping as discussed later or any other Quantum Cryptography, Image hijacking intrusion in its key distribution mechanism [1][5].

Thus such unique properties of Quantum As can been seen from the above figure not all Cryptography make it a potential key distribution 8-bit value’s last bit needed to be changed as technique for the future generation networks as some were the same with the original bit value. well. As can be seen only the maroon bits (with 1) and Therefore, our paper focuses on proposing yellow bits (with 0) had to be changed. Therefore mechanism which combines an old technique if we consider in the worst case all 32 bits needed such as Steganography with the unique key to be inserted one after another as the last bits for distribution mechanism of Quantum each pixel’s bytes which would make about 32 of Cryptography. Besides, some additional features those 8*3*11 =264 (about 12%). are also proposed to make a strong information However in this case we needed to flow channel between two next generation change only 15 of 264 bits (less than 6%) [8], the networks primarily with a focus on fast and unaltered and altered image is shown in Fig. 2. efficient QoS providing networks such as Content Now when we consider a fairly small image of Delivery Networks. 240*160 pixels with each pixel having 24 bits composed of RGB, we get 240*160*24 = 921600 II. RELATED WORKS bits the hidden message becomes even a smaller A. Steganography in Practice portion and hard to identify at all. Steganography strips less important information from digital content and inserts hidden data in its place. This is done over the spectrum of the entire image. One of the ways it is practiced is described as follows. As shown in Fig. 1, a small message called “Aha!” is hidden in a dark green background composed of dark green pixels. A dark green pixel can be Fig. 2. Hidden image before and after altering with represented by three 8-bit values for the three Steganography. colours, Red, Green and Blue (RGB). In this case, the last bit of each 8-bit value is replaced by the B. Key Distribution with Quantum Cryptography consecutive bit value of the message “Aha!” The In physics, a photon is considered as the following Fig. 1 shows how it is done in practice. smallest particle of light which is usually unpolarized. It can be polarized by 2 kinds of filters to make it spin steady which are namely the cross and diagonal filter. The cross filter filters out photons with (|) and (--) polarization. The diagonal filter filters out photons with (/) and ( ) polarization. (|), (--), (/) and () can be assigned logic bits 0 and 1 in agreement. In this case the logic bits 1 and 0 are rather replaced with photons polarization and can be considered as qubits (quantum bits) [2][5]. There are three unique features of qubits as follows:

1) If a qubits is filtered with the wrong filter type Fig. 1. Steganography Mechanism. or copied, the photon’s polarization is destroyed.

2) Qubits can be in two opposite states at the forward it to the other party. On the other hand same time. the eavesdropper doesn’t now the filter basis to 3) To read information of qubits we need to be used to measure the photons. In that case only measure it and results of measurement are 50% time the eavesdropper will get the right always probabilistic. result and can only forward the result of the photons to the other party which will again have another 50% possibility to get right result. When some test bits are matched with the shared filter basis very few results will match due to the eavesdropping in the middle. This part is discussed in the analysis section of our paper. III. PROPOSED MECHANISM Our proposed mechanism focuses on combining the old image concepts of Steganography with the new concepts of Quantum Cryptography in terms of establishing a free from eavesdropping. The idea lies in first establishing a Service Level Agreement (SLA) between the two active Fig. 3. Photos polarization and translation into keys Content Delivery Networks, say for example [2]. between CDN1 and CDN2. The Service Level In quantum cryptography the secret key bits Agreements includes the image type, for example are sent as qubits (which is basically the spin of a meaningful image of a ship or so, and the angle the photon). So for example, four logical of the image in which the image is encrypted and impressions can be derived from the four possible needs to be decrypted on the other side. Next the qubits states as shown in Fig. 4 [2]. two CDNs authenticate each other by exchanging the quantum generated photon based key which the other CDN acknowledges. When CDN1 wants to send the image it sends the meaningful image in accordance to the SLA with the secret quantum key randomly distributed into the image which was previously exchanged and agreed upon. Then CDN2 can acknowledge and verify the image based on the image type, angle and secret quantum key embedded in the image. In the case where CDN1 sends an image and an attacker captures the image in the middle and forwards a fake image to CDN2, CDN2 may receive some message but the quantum secret key will ensure that the image is not the real image sent from

CDN1. The above procedure is demonstrated as Fig. 4. Qubits states and logical impressions. follows in Fig. 5. As described in the first feature, an eavesdropper cannot just copy a photon and

generated in a completely secure quantum mode free from eavesdropping or any other intrusion in the middle. IV. ANALYSIS AND SIMULATION In general, in case of a CDN2 has to measure the qubits with a random filter which can be either cross or diagonal. And there will be about 50% possibility that CDN2 will guess the right filter as the qubits can be either cross or diagonal both with equal probabilities. To check whether CDN2 has got the right qubits, 25% of those 50% matched filter’s qubits will be shared between the two CDNs as test bits. Out of this 25% if less than a certain threshold, say for example 30% bits don’t match, then eavesdropping can be assumed to have happened. This is because an eavesdropper Fig. 5. Proposed mechanism of information flow cannot just forward the qubits due to the non- between two CDNs. clone ability property of qubits. Therefore the Therefore, as explained in the above eavesdropper can only forward the measurement diagram, CDN2 first need to check with the of the qubits which will also be about 50% previously agreed SLA terms and information correct. Also any measurement collapses the received according to that. Thus it checks the type quantum superposition of the qubits. So, CDN2 of the image and the angle of the image. Next only receives the result of measurements, not the after decrypting using the associated algorithm, it original qubits. Therefore, when CDN2 measures checks whether the image comes with a secret the message coming from the eavesdropper, there key and whether the secret key matches with the is another layer of applying wrong random bases, quantum key previously distributed between the which decreases the probability of getting the two parties. If it matches the image can be original message by more than 50%. The process verified and the information can be extracted will continue as long as all the test bits don’t from the image. If not then the image will be match and a secure channel is not established. discarded and the information of the existence of And the rest 25% of the bits Secret key (Ks) sent some attacker in the middle will be disseminated through quantum photons polarization is then to secure the channel. used as the secret key following the process as The idea of this paper is to increase the secrecy depicted Secret Key Acknowledged in Fig. 6. and verifiability of the Steganography generated image which can securely and simply send and receive information hidden inside a Steganography generated image following a certain algorithm. Some properties of the image such as image type, image angle are utilized in the process to identify the image before finally verifying it with the quantum key that was

size of our sent key.

Fig. 7. Final Secret key size based on initial key size. Fig. 6. Final key generation after dropping wrong The data from the above graph data can be filters and test bits. summarized in the following table. Graphical Analysis As stated earlier, in our mechanism if an eavesdropper Eve tries to tap the channel, this will automatically show up in CDN2’s measurements. In those cases where CDN1 and CDN2 have used the same basis, CDN2 is likely to obtain an incorrect measurement. Thus Eve’s measurements are bound to affect the states of the photons. As eavesdropper intercepts CDN1’s Table. 1. Final Secret key size based on initial key photons, she has to measure them with a random size. basis and send new photons to CDN2 since the So for these final keys, if we consider the photon states cannot be cloned (non-clone ability percentage of the image that it is covering then it property of quantum). Thus, eavesdropper’s can be seen it is a very small portion. For a fairly presence is always detected since measuring a small image of size 240*160 pixels with each quantum system irreparably alters its state and the pixel having 24 bits composed of 8 bits from each expected probabilities. Red, Green and Blue combination, we get the For our proposed mechanism, the main goal is percentage of the key in the image formula as , to detect an eavesdropping of the Steganography generated image by the concepts of quantum cryptography and some other possible SLA features. So based on a threshold, say for δ = 30%, if we find the percentage of matches in test In the worst case the percentage will be bits are less than 30% for δ value we can assume the exact percentage calculated from the above the possibilities of eavesdropping. If not, the rest formula but in the best case the percentage can 25% of the bits is used as the secret key. also reduce to 0% in a scenario where all the bits Therefore, if we start with a small 64 bit key then, of the key match with the image bits and so we based on the required calculations as described don’t need to change any bits in the image. The before, only 16 bits remain as the final secret key following graph shows the best case and worst at the end. The following graph shows how the case scenarios for the final secret key values that final key size varies when we keep increasing the we achieved from our previous graph

calculations. In a real scenario, since we can have the angle in which the image is sent. The latter only two bits, 1 and 0, we can assume about a would be helpful in the case where we consider 50% possibility that the bit would not need to be the secret key is somehow stolen (by company changed and therefore it would always be less espionage); the image sent would still be than the worst case. meaningless if it is not decrypted in the required angle. In some research, hijacking of the image in visual cryptography is being tried to be solved by signatures or logos embedded with shares in a visual cryptography scheme [3]. In the case of Steganography, if we use such processes an eavesdropper hijacking the image can analyze the shares and make minimal changes, yet devastating, to fake an image. Since in our mechanism the secret message along with the unique one-time key is randomly distributed Fig. 8. Percentage of the Final Secret key in the following the Service Level Agreements for Steganography image. verifying the image, we believe our approach As can be seen from the above graph, even in stands out in terms of its unique verification the worst case for a couple of secret keys ranging properties and is promising for a futuristic next from the size of 16 to 2048, the percentage generation network in terms of its adoption of covered in the image is less than 25% or 0.25 latest technologies. which makes the key quite light weight and V. CONCLUSION efficient for fast next generation networks while We proposed a mechanism where we providing the same level of security. The combined the old concepts of Steganography following graph information can also be with the new concept of Quantum Cryptography represented in table form as follows. to create a strong and secure channel between next generation networks such as Content Delivery Networks. Our theoretical and experimental results suggest that the communication can be technologically upgraded to provide a simplified but strong communication with light weight but eavesdropping free key distribution mechanism offered by Quantum Cryptography while utilizing the simplicity offered by Steganography for information flow based on our previously set Service Level Agreements. Table. 2. Percentage of the Final Secret Key in the Steganography image.

Therefore in our mechanism, while an image sent with the secretly shared quantum key VI. REFERENCES would prove the image is not a hijacked image, another level of security can also be provided by

[1]. D. Hjelme, L. Lydersen, “ A Multidisciplinary Introduction to information security” Chapter 5: Quantum Cryptography.

[2]. J. Clark, “How Quantum Cryptography Works”, available at “http://science.howstuffworks.com/science-vs- myth/everyday-mthy/quantum-cryptology3.htm.

[3] K. Singh, S. Nandi, S. Singh, L. Singh, “Stealth Steganography in visual cryptography for half tone images”, Proceedings of the International conference on computer and communication Engineering, pp. 1217-1221,May 2008.

[4] M. Vincent, E. Helena, “Securing multiple colour secrets using visual cryptography”, International Conference on Modelling Optimization and computing”, Vol.38, pp.806- 812, 2010.

[5]. N. Dunn, “Quantum Key Distribution”, available at http://asgc.ualr.edu/documents/2014/presentatio ns/oral/engineering/saun_dunn.pdf.