Historically, cryptography was created and used to deliver messages between people in such a way that only the intended receiver would be able to read them. The two persons communicating trusted each other, but not the rest of the world - for instance, not the message bearer. But what happens if we assume instead that even the receiver of a message may not be trusted by the sender, while both still have an interest in communicating? This problem is addressed by secure multi-party computation, that studies communication and shared computation between mutually distrusting parties. In a world where most of the communication happens over the Internet, and where an ever-increasing amount of computation is outsourced, for example to the cloud, multi-party computation is a subject of crucial importance. However, unconditionally secure multi-party protocols have never found practical application: the lack of realistic noisy channel models, that are required to achieve security against computationally unbounded adversaries, prevents implementation over real-world, standard communication protocols.\n\nIn this thesis, we study the properties and characteristics of noisy channels, and we introduce new models based on real-world error sources that have never been studied in this context, such as delays. On the contrary, previous works mostly focused on theoretical noisy channels that have little in common with the reality of network communication. We show that the models we propose are suitable for multi-party computation by building on each of them a protocol for oblivious transfer, a fundamental cryptographic primitive that implies any secure computation. Thanks to the realism of our noisy channels, we are able to provide the first working implementation of unconditionally secure oblivious transfer over both wired and wireless networks and different Internet protocols, which we test thoroughly in an extensive set of experiments. Our constructions are not only more realistic, but also more flexible, as they allow larger ranges of error probabilities. This will eventually open the way for the widespread use of unconditionally secure multi-party computation.
Historically, cryptography was created and used to deliver messages between people in such a way that only the intended receiver would be able to read them. The two persons communicating trusted each other, but not the rest of the world - for instance, not the message bearer. But what happens if we assume instead that even the receiver of a message may not be trusted by the sender, while both still have an interest in communicating? This problem is addressed by secure multi-party computation, that studies communication and shared computation between mutually distrusting parties. In a world where most of the communication happens over the Internet, and where an ever-increasing amount of computation is outsourced, for example to the cloud, multi-party computation is a subject of crucial importance. However, unconditionally secure multi-party protocols have never found practical application: the lack of realistic noisy channel models, that are required to achieve security against computationally unbounded adversaries, prevents implementation over real-world, standard communication protocols.\n\nIn this thesis, we study the properties and characteristics of noisy channels, and we introduce new models based on real-world error sources that have never been studied in this context, such as delays. On the contrary, previous works mostly focused on theoretical noisy channels that have little in common with the reality of network communication. We show that the models we propose are suitable for multi-party computation by building on each of them a protocol for oblivious transfer, a fundamental cryptographic primitive that implies any secure computation. Thanks to the realism of our noisy channels, we are able to provide the first working implementation of unconditionally secure oblivious transfer over both wired and wireless networks and different Internet protocols, which we test thoroughly in an extensive set of experiments. Our constructions are not only more realistic, but also more flexible, as they allow larger ranges of error probabilities. This will eventually open the way for the widespread use of unconditionally secure multi-party computation.
The security of digital goods buyers and sellers is unbalanced. Of course, the property of sellers is protected; for example, when customers acquire digital books or films from Internet's merchants, they only receive the products they have paid for. Unfortunately, the buyers' privacy is rarely respected: purchases are often — without the buyers' knowledge — monitored, recorded, analysed, and sometimes sold to marketing companies. As a consequence, even if the customers do not intend to acquire additional products, their computer screens are later invaded with targeted advertisements. The main purpose of this thesis is to propose some methods to restore the balance and guarantee the buyers' privacy, while protecting the interests of the sellers. To this end, it is worth looking into the area of cryptography and more specifically, it is worth studying and designing some protocols called distributed oblivious transfer (DOT) protocols. A DOT protocol allows a party A to obtain one of the secret pieces of information (a secret for short) held by a party B so that the following two fundamental conditions are satisfied: • A chooses the secrets she wishes to obtain and does not obtain anything on the secrets she has not chosen, • B does not learn which secret was obtained by A. Furthermore, to improve the availability of the information, the protocol is distributed. That is, the party B transmits his secret information to m servers and the party A needs to contact at least k of these servers to obtain the chosen secret. The servers are not fully trusted, neither by A, nor by B. Therefore, from the information exchanged with A and B, no coalition of servers should be able to learn the secrets of B or the choice of A. The results of a preliminary literature review are surprising. In fact, the number of publications on DOT protocols is small (fewer than 20) compared to, for example, the number of publications on a similar concept, secret sharing (100s of publications). And yet, oblivious transfer is a fundamental component of more complex cryptographic protocols such as multi-party computation protocols, which allow a group of participants to securely calculate any function of their joint secret inputs. So, one could expect many variants, for example of the original DOT protocol introduced in 2002 by Naor and Pinkas [74], to fulfil the requirements of specific scenarios. The design of variants of DOT protocols in traditional cryptography has been the guiding thread of my research. My contribution mainly consists in (a) a critical analysis of the existing protocols, demonstrating their limitations, weaknesses, and in some cases, flaws; and (b) the design of the following protocols,well adapted to some specific situations: A Strongly Secure DOT Protocol. This DOT protocol addresses the most important weakness of unconditionally secure, one-round, polynomial interpolation-based DOT protocols: after the protocol has been executed, if the party A corrupts only one server, she can learn all the secrets of the party B. The protocol is secure even if A corrupts up to k - 1 servers. A Verifiable DOT Protocol. The party A should obtain the secret she has chosen, even when some servers are controlled by a malicious adversary whose objective is to sabotage the protocol. This is the case with this protocol, assuming that the adversary cannot control more than k - 1 servers. A Multiple Secrets DOT Protocol. When the party A wishes to obtain n > 1 secrets, the current protocols have to be executed n times. In this context, they are inefficient. The DOT protocol introduced here allows the party A, by contacting in the same session k - 1 + n servers, to collect n secrets. Adaptive DOT Protocols. The previous protocol allows the party A to request several secrets. However, the request of one secret may depend on the values of secrets already obtained. Two efficient protocols are presented in this scenario. The first one allows A to receive a limited number of secrets and therefore, is well adapted to a single receiver. For several receivers, a second protocol is proposed. This second protocol accepts an unlimited number of queries, but requires communications amongst the servers. A Threshold DOT protocol. Most existing DOT protocols rely on threshold secret sharing schemes. In a k-threshold protocol or scheme, security is guaranteed not only when k parties are contacted, but also when more than k parties are contacted. However, the existing DOT protocols based on k-threshold secret sharing schemes require an additional mechanism to control that exactly k servers are contacted, which is an under-utilisation of the underlying functionality. The proposed protocol is the first k-threshold DOT protocol which allows the party A to contact as many servers as she wishes to obtain the chosen secret, provided that at least k servers are contacted. This research is limited to unconditionally secure protocols, i.e., protocols whose security does not depend on mathematical (unproven) assumptions; within the limits of the given security models, the protocols are secure even against an adversary with unlimited computing power and time. In brief, the results presented in this thesis are a significant advance to the state of the art research on DOT protocols because on one hand, they point out the weaknesses of the DOT protocols most commonly accepted by the cryptographic community and on the other hand, they contribute to the cryptographic field through the design of new protocols, secure and efficient.
Oblivious transfer is a primitive of paramount importance in cryptography or, more precisely, two- and multi-party computation due to its universality. Unfortunately, oblivious transfer cannot be achieved in an unconditionally secure way for both parties from scratch. Therefore, it is a natural question what information-theoretic primitives or computational assumptions oblivious transfer can be based on. The results in our thesis are threefold. First, we present a protocol that implements oblivious transfer from a weakened oblivious transfer called universal oblivious transfer, where one of the two players may get additional information. Our reduction is about twice as efficient as previous results. Weak oblivious transfer is an even weaker form of oblivious transfer, where both players may obtain additional information about the other player's input, and where the output can contain errors. We give a new, weaker definition of weak oblivious transfer, as well as new reductions with a more detailed analysis. Finally, we carry over our results to the computational setting and show how a weak oblivious transfer that is sometimes incorrect and only mildly secure against computationally bounded adversaries can be strengthened.
In secure multi-party computation n parties jointly evaluate an n-variate function f in the presence of an adversary which can corrupt up till t parties. Almost all the works that have appeared in the literature so far assume the presence of authenticated channels between the parties. This assumption is far from realistic. Two directions of research have been borne from relaxing this (strong) assumption: (a) The adversary is virtually omnipotent and can control all the communication channels in the network, (b) Only a partially connected topology of authenticated channels is guaranteed and adversary controls a subset of the communication channels in the network. This work introduces a new setting for (unconditional) secure multiparty computation problem which is an interesting intermediate model with respect to the above well studied models from the literature (by sharing a salient feature from both the above models). We consider the problem of (unconditional) secure multi-party computation when 'some' of the communication channels connecting the parties can be corrupted passively as well as actively. For this setting, some honest parties may be connected to several other honest parties via corrupted channels and may not be able to authentically communicate with them. Such parties may not be assured the canonical guarantees of correctness or privacy. We present refined definitions of security for this new intermediate model of unconditional multiparty computation. We show how to adapt protocols for (Unconditional) secure multiparty computation to realize the definitions and also argue the tightness of the results achieved by us.
In this thesis, we are interested in the theory of unconditional secure two-party computations of which Oblivious Transfer (OT) and Bit Commitment (BC) are the central primitives. On one hand, my works are inspired from Crépeau's et al.'s framework of building of OT protocol from noisy communication channels. The principle of this framework is to conceive, from noisy channels, an intermediate erasure model which is a variant of OT. We contributed to this framework by proposing a more general intermediate model, the Binary Symmetric Multi-Error-Rate Channel, which also can be built from noisy channels. With this intermediate model, we can build OT protocol from the noisy channels more effectively. In addition, we expose some case studies on emulating noisy models by a quantum nonorthogonal coding (QNOC) scheme which uses two non-orthogonal pure states for encoding two values of the classical bit. On the other hand, we revise the quantum model for general two-party protocols concerning classical and quantum computations and communications. We state that in the general model, a classical channel is inevitably macroscopic and its decoherence is so strong that quantum information is not accepted to be transfered on it. Thus, the quantum model for two-party protocols becomes three-party, including an environment of the channel. Indeed, with the faithful interpretation of general quantum two-party protocols in this three-party model, we reaffirm the no-go theorems of Mayers and Lo & Chau on the impossibilities of quantum OT, BC. In addion, we can go further to apply these negative results to protocols using some quantum trusted oracles, such as Coin-Flipping.
Motivated by the applications of secure multi-party computation as a privacy-protecting data analysis tool, and identifying oblivious transfer as one of its main practical enablers, we propose a practical realization of randomized quantum oblivious transfer. By using only symmetric cryptography primitives to implement commitments, we construct computationally secure randomized oblivious transfer without the need for public-key cryptography or assumptions imposing limitations on the adversarial devices. We show that the protocol is secure under an indistinguishability-based notion of security and demonstrate an experimental implementation to test its real-world performance. Its security and performance are then compared to both quantum and classical alternatives, showing potential advantages over existing solutions based on the noisy storage model and public-key cryptography.
In this thesis we focus on various aspects of secure message transmission protocols. Such protocols achieve the secure transmission of a message from a sender to a receiver - where the term “secure” encapsulates the notion of privacy and reliability of message transmission. These two parties are connected using an underlying network in which a static computationally unlimited active adversary able to corrupt up to t network nodes is assumed to be present. Such protocols are important to study as they are used extensively in various cryptographic protocols and are of interest to other research areas such as ad-hoc networks, military networks amongst others. Optimal bounds for the number of phases (communication from sender to receiver or vice versa), connectivity requirements (number of node disjoint network paths connecting sender and receiver - denoted by n), communication complexity (complexity of the number of field elements sent - where F is the finite field used and jFj = q) and transmission complexity (proportion of communication complexity to complexity of secrets transmitted) for secure message transmission protocols have been proven in previous work. In the one-phase model it has been shown that n 3t+1 node disjoint paths are required to achieve perfect communication. In the two phase model only n 2t + 1 node disjoint paths are necessary. This connectivity is also the required bound for almost perfectly secure one-phase protocols - protocols which achieve perfect privacy but with a negligible probability may fail to achieve reliability. In such cases the receiver accepts a different message to that transmitted by the sender or does not accept any message. The main focus of recent research in secure message transmission protocols has been to present new protocols which achieve optimal transmission complexity. This has been achieved through the transmission of multiple messages. If a protocol has a communication complexity of O(n3) field elements, to achieve optimal transmission complexity O(n2) secrets will have to be communicated. This has somewhat ignored the simplification and improvement of protocols which securely transmit a single secret. Such improvements include constructing more efficient protocols with regards to communication complexity, computational complexity and the number of field elements sent throughout the whole protocol. In the thesis we first consider one-phase almost perfectly secure message transmission and present two new protocols which improve on previous work. We present a polynomial time protocol of O(n2) communication complexity which at the time of writing this thesis, is computationally more efficient than any other protocol of similar communication complexity for the almost perfectly secure transmission of a single message. Even though our first almost perfectly secure transmission protocol is of polynomial time, it is important to study other protocols also and improve previous work presented by other researchers. This is the idea behind the second one-phase almost perfectly secure message transmission protocol we present which requires an exponential complexity of field operations but lower (O(n)) communication complexity. This protocol also improves on previous protocols of similar communication complexity, requiring in the order of O(log q) less computation to complete - where q denotes the size of the finite field used. Even though this protocol is of exponential time, for small values of n (e.g. when t = 1, t = 2 or t = 3) it may be beneficial to use this protocol for almost perfectly secure communication as opposed to using the polynomial time protocol. This is because less field elements need to be transmitted over the whole network which connects a sender and a receiver. Furthermore, an optimal almost perfectly secure transmission protocol will be one with O(n) communication complexity and with polynomial computational complexity. We hope that in the future, other researchers will be inspired by our proposed protocol, improve on our work and ideally achieve these optimal results. We also consider multi-phase protocols. By combining various cryptographic schemes, we present a new two-phase perfectly secure single message transmission protocol. At the time of writing this thesis, the protocol is the most efficient protocol when considering communication complexity. Our protocol has a communication complexity of O(n2) compared to O(n3) of previous work thus improving on the communication complexity by an order of O(n) for the perfectly secure message transmission of a single message. This protocol is then extended to a three phase protocol where a multi-recipient broadcast end channel network setting is considered. As opposed to point to point networks where a path from a sender reaches a single receiver, this network model is new in the field of message transmission protocols. In this model each path from a sender reaches multiple receivers, with all receivers receiving the same information from their common network communication channel. We show how the use of this protocol upon such a network can lead to great savings in the transmission and computation carried out by a single sender. We also discuss the importance and relevance of such a multi-recipient setting to practical applications. The first protocols in the field of perfectly secure message transmission with a human receiver are also presented. This is a topic proposed by my supervisor Professor Yvo Desmedt for which I constructed solutions. In such protocols, one of the communicating parties is considered to be a human who does not have access to a computational device. Because of this, solutions for such protocols need to be computationally efficient and computationally simple so that they can be executed by the human party. Experiments with human participants were carried out to assess how easily and accurately human parties used the proposed protocols. The experimental results are presented and these identify how well human participants used the protocols. In addition to the security of messages, we also consider how one can achieve anonymity of message transmission protocols. For such protocols, considering a single-receiver multi-sender scenario, the presence of a t-threshold bounded adversary and the transmission of multiple secrets (as many as the number of sender), once the protocols ends one should not be able to identify the sender of a received message. Considering a passive and active adversary new protocols are presented which achieve the secure and anonymous transmission of messages in the information-theoretic security model. Our proposed solutions can also be applied (with minor alterations) to the dual problem when a single-sender multi-recipient communication setting is considered. The contributions of the thesis are primarily theoretical - thus no implementation of the proposed protocols was carried out. Despite this, we reflect on practical aspects of secure message transmission protocols. We review the feasibility of implementing secure message transmission protocols in general upon various networks - focusing on the Internet which can be considered as the most important communication network at this time. We also describe in theory how concepts of secure message transmission protocols could possibly be used in practical implementations for secure communication on various existing communication networks. Open problems that remain unsolved in the research area of the proposed protocols are also discussed and we hope that these inspire research and future solutions for the design (and implementation) of better and more efficient secure message transmission protocols.