Bandwidth-Optimized Secure Two-Party Computation of Minima

We optimize the communication (and, indirectly, computation) complexity of two-party secure function evaluation (SFE). We propose a new approach, which relies on the information-theoretic (IT) Garbled Circuit (GC), which is more efficient than Yao’s GC on shallow circuits. When evaluating a large circuit, we “slice” it into thin layers and evaluate them with IT GC. Motivated by the client-server setting, we propose two variants of our construction: one for semi-honest model (relatively straightforward), and one secure against a semi-honest server and covert client (more technically involved). One of our new building blocks, String-selection Oblivious Transfer (SOT), may be of independent interest.Our approach offers asymptotic improvement over the state-of-the-art GC, both in communication and computation, by a factor logκ, where κ is a security parameter. In practical terms, already for today’s κ ∈ {128,256} our (unoptimized) algorithm offers approximately a factor 2 communication improvement in the semi-honest model, and is only a factor ≈ 1.5 more costly in setting with covert client.KeywordsRandom OracleSecure ComputationHomomorphic EncryptionOblivious TransferSelection StringThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Secure two-party computation allows two untrusting parties to jointly compute an arbitrary function on their respective private inputs while revealing no information beyond the outcome. Existing cryptographic compilers can automatically generate secure computation protocols from high-level specifications, but are often limited in their use and efficiency of generated protocols as they are based on either garbled circuits or (additively) homomorphic encryption only.

Alice and Bob with their private inputs xn and yn respectively, want to compute fn(xn, yn) for some publicly known function fn without disclosing information regarding their private inputs more than what can be inferred from fn(xn, yn). This problem is referred to as a secure two-party computation and Yao proposed a solution to privately compute fn using garbled circuits. In this paper, we improve the efficiency of circuit by hardwiring the input of Alice in the circuit without compromising privacy. Using a typical two-party computation problem, namely, the Millionaire Problem, we show that our method reduces circuit size significantly specially for circuits whose fan-in is bounded by 2. We also show that the protocol using the reduced circuit is provably secure.

Secure two-party computation (2PC) based on homomorphic encryption (HE) achieves formal data privacy protection and gets increasing adoption for private deep neural network (DNN) inference. As modern HE schemes usually operate on polynomials, existing works rely on manually-designed HE kernels for representative DNN operations. However, this is not only unscalable considering the diverse operator types, shapes, polynomial orders, etc, but also misses important optimization opportunities. In this paper, we introduce FlexHE, a flexible kernel generation framework to enable automatic generation and optimization of HE kernels for 2PC-based private inference. Given a high-level description of DNN operations, FlexHE can systematically define the HE kernel design space considering various optimization dimensions, including loop tiling, reordering, etc. We also analyze the communication and computation impact of different optimization dimensions for design space reduction. To search for the best kernel design, a two-level optimization problem is formulated and iteratively solved with an integer linear programming (ILP) formulation. With extensive experimental results, we not only demonstrate a better coverage of DNN operations including depth-wise Conv3D and dilated Conv3D, but also achieve more than 100×, 7.9×, and 4.2× latency reduction compared to prior-art HElayers, Cheetah, and Falcon, respectively.

The complexity of many cryptographic protocols, including today’s most efficient Secure Function Evaluation (SFE) protocols, depends linearly on the size of the boolean circuit representation of the function to be evaluated. In this chapter we show how to optimize boolean circuits for cryptographic protocols and give efficient circuit constructions. We first motivate protocols that allow “free” evaluation of XOR gates: (SFE), Homomorphic Encryption (HE), and zero-knowledge (Introduction). Afterwards, we give optimization algorithms that reduce the size of boolean circuits in particular when XOR gates can be evaluated “for free” in circuit optimizations. In circuit constructions we show how our optimizations allow one to construct more efficient circuits for standard functionalities such as integer comparison. We also give efficient circuit constructions for computing minima (or maxima) and for fast multiplication. As applications we consider efficient protocols for secure comparison and first-price auctions in applications: secure comparison auctions.

Secure two-party computation provides a way for two parties to compute a function, that depends on the two parties’ inputs, while keeping them private. Known since the 1980s, Yao’s garbled circuits appear to be a general solution to this problem, in the semi-honest model. Decades of optimizations have made this tool a very practical solution. However, it is well known that a malicious adversary could modify a garbled circuit before submitting it. Many protocols, mostly based on cut-&-choose, have been proposed to secure Yao’s garbled circuits in the presence of malicious adversaries. Nevertheless, how much an adversary can modify a circuit and make it still executable has not been studied yet. The main contribution of this paper is to prove that any modification made by an adversary is equivalent to adding/removing NOT gates arbitrarily in the original circuit, otherwise the adversary can get caught. Thereafter, we study some evaluation functions for which, even without using cut-&-choose, no adversary can gain more information about the inputs by modifying the circuit. We also give an improvement over most recent cut-&-choose solutions by requiring that different circuits of the same function are used instead of just one.

This article presents the first privacy-preserving localization method based on provably secure primitives for smart automotive systems. Using this method, a car that is lost due to unavailability of GPS can compute its location with assistance from three nearby cars, while the locations of all the participating cars including the lost car remain private. Technological enhancement of modern vehicles, especially in navigation and communication, necessitates parallel enhancement in security and privacy. Previous approaches to maintaining user location privacy suffered from one or more of the following drawbacks: trade-off between accuracy and privacy, one-sided privacy, and the need of a trusted third party that presents a single point to attack. The localization method presented here is one of the very first location-based services that eliminates all these drawbacks. Two protocols for computing the location is presented here based on two Secure Function Evaluation (SFE) techniques that allow multiple parties to jointly evaluate a function on inputs that are encrypted to maintain privacy. The first one is based on the two-party protocol named Yao’s Garbled Circuit (GC). The second one is based on the Beaver-Micali-Rogaway (BMR) protocol that allows inputs from more than two parties. The two secure localization protocols exhibit trade-offs between performance and resilience against collusion. Along with devising the protocols, we design and optimize netlists for the functions required for location computation by leveraging conventional logic synthesis tools with custom libraries optimized for SFE. Proof-of-concept implementation of the protocol shows that the complete operation can be performed within only 355ms. The fast computing time enables localization of even moving cars.

Secure multi-party computation (MPC) allows two or more distrusting parties to jointly evaluate a function over private inputs. For a long time considered to be a purely theoretical concept, MPC transitioned into a practical and powerful tool to build privacy-enhancing technologies. However, the practicality of MPC is hindered by the difficulty to implement applications on top of the underlying cryptographic protocols. This is because the manual construction of efficient applications, which need to be represented as Boolean or arithmetic circuits, is a complex, error-prone, and time-consuming task. To facilitate the development of further privacy-enhancing technology, multiple compilers have been proposed that create circuits for MPC. Yet, almost all presented compilers only support domain specific languages or provide very limited optimization methods. In this work (this is an extended and revised version of the paper ‘Secure Two-party Computations in ANSI C’ (Holzer et al., in: ACM CCS, 2012) that reflects the progress in secure computation and describes the current optimization tool chain of CBMC-GC) we describe our compiler CBMC-GC that implements a complete tool chain from ANSI C to circuit. Moreover, we give a comprehensive overview of circuit minimization techniques, which we have identified and adapted for the creation of efficient circuits for MPC. With the help of these techniques, our compilation approach allows for a high level of abstraction from the cryptographic primitives used in MPC protocols, as well as the complex design of digital circuits. By using the model checker CBMC as a compiler frontend, we illustrate the link between MPC, formal methods, and digital logic design. Our experimental results illustrate the effectiveness of the implemented optimizations techniques for various example applications. In particular, compared with other state-of-the-art compilers, we show that CBMC-GC compiles circuits from the same source code that are up to four times smaller.

The advances in sensing capabilities of smartphones give rise to a variety of mobile participatory sensing applications that collect users' personal data. Because of the existence of both sensitive, private personal data, and untrusted aggregator, serious privacy concerns on users arise. Currently, existing privacy preserving data collection methods either require bidirectional communications between an untrusted aggregator and mobile users in every aggregation period, or have high computation or communication overhead. To address these problems, we propose an efficient data aggregation approach by which an untrusted aggregator in mobile sensing can collect the statistics over the data contributed by multiple mobile users, while supporting privacy preservation of each user and data integrity verification. In this approach, information hiding and homomorphic encryption are applied to guarantee the data privacy of mobile users. In detail, a breadth-first search tree is first constructed at the initial phase among the mobile users, and then the original datum of each user is perturbed among its neighbors in ciphertext space by using information hiding and homomorphic encryption. The evaluations of our approach show that our protocol requires lower communication and computation overhead and thus more feasible for the computation constrained mobile devices. Copyright © 2016 John Wiley & Sons, Ltd.

An improved blockchain-based multi-region Federated Learning framework for crop disease diagnosis

Secure outsourcing of computation to an untrusted (cloud) service provider is becoming more and more important. Pure cryptographic solutions based on fully homomorphic and verifiable encryption, recently proposed, are promising but suffer from very high latency. Other proposals perform the whole computation on tamper-proof hardware and usually suffer from the the same problem. Trusted computing (TC) is another promising approach that uses trusted software and hardware components on computing platforms to provide useful mechanisms such as attestation allowing the data owner to verify the integrity of the cloud and its computation. However, on the one hand these solutions require trust in hardware (CPU, trusted computing modules) that are under the physical control of the cloud provider, and on the other hand they still have to face the challenge of run-time attestation.In this paper we focus on applications where the latency of the computation should be minimized, i.e., the time from submitting the query until receiving the outcome of the computation should be as small as possible. To achieve this we show how to combine a trusted hardware token (e.g., a cryptographic coprocessor or provided by the customer) with Secure Function Evaluation (SFE) to compute arbitrary functions on secret (encrypted) data where the computation leaks no information and is verifiable. The token is used in the setup phase only whereas in the time-critical online phase the cloud computes the encrypted function on encrypted data using symmetric encryption primitives only and without any interaction with other entities.

In vehicular networks, drones can enhance performance by enabling real-time data collection, traffic monitoring, and communication optimization. However, due to the large, open, and highly dynamic nature of vehicular networks, ensuring security and achieving lightweight privacy protection are pressing challenges. Additionally, in traditional solutions, the control center (CC) often performs many tasks, leading to excessive computation and communication overhead, creating a system bottleneck. To address these issues, this paper proposes a privacy-preserving authentication scheme for FL in drone-assisted vehicular networks, where model parameter collection and model training are performed by drones in a privacy-preserving manner, significantly reducing the computation and communication overhead on the CC side. Moreover, the proposed scheme ensures strong privacy protection and efficient authentication by using federated learning, Elliptic Curve Cryptography (ECC), and the Chinese Remainder Theorem (CRT). The performance evaluation results show that the proposed scheme achieves a 17.3% reduction in computation overhead and a 22.8% reduction in communication overhead compared to existing solutions, highlighting its significant advantages.

In this paper, we present a parallel iterative solution for large banded systems of linear equations based on incomplete LU-factorization (ILU). A master—workers parallel computing scheme is used. The proposed algorithm incurs reduced storage and communication overhead as compared to previous methods. The reduction in communication overhead has been achieved by pipelining the inter-workers message passing and limiting the communication between master and workers to initial work distribution and final results collection.The performance of the proposed algorithm is evaluated analytically and experimentally using Parallel Virtual Machine (PVM) as a parallel programming environment on a cluster of Linux work-stations. The obtained performance results show that the proposed algorithm exhibits higher efficiency for coarse grain computations corresponding to large matrix sizes and reduced number of processors.

The rapid expansion of the Internet of Things (IoT) ecosystem has transformed industries but also exposed significant cybersecurity vulnerabilities. Traditional centralized methods for securing IoT networks struggle to balance privacy preservation with real-time threat detection. This study presents a Federated Learning-Driven Cybersecurity Framework designed for IoT environments, enabling decentralized data processing through local model training on edge devices to ensure data privacy. Secure aggregation using homomorphic encryption supports collaborative learning without exposing sensitive information. The framework employs GRU-based recurrent neural networks (RNNs) for anomaly detection, optimized for resource-constrained IoT networks. Experimental results demonstrate over 98% accuracy in detecting threats such as distributed denial-of-service (DDoS) attacks, with a 20% reduction in energy consumption and a 30% reduction in communication overhead, showcasing the framework’s efficiency over traditional centralized approaches. This work addresses critical gaps in IoT cybersecurity by integrating federated learning with advanced threat detection techniques. It offers a scalable, privacy-preserving solution for diverse IoT applications, with future directions including blockchain integration for model aggregation traceability and quantum-resistant cryptography to enhance security.

Secure two-party computation is a rapidly emerging field of research and enables a large variety of privacy-preserving applications such as mobile social networks or biometric identification. In the late eighties, two different approaches were proposed: Yao’s garbled circuits and the protocol of Goldreich-Micali-Wigderson (GMW). Since then, research has mostly focused on Yao’s garbled circuits as they were believed to yield better efficiency due to their constant round complexity. In this work we give several optimizations for an efficient implementation of the GMW protocol. We show that for semi-honest adversaries the optimized GMW protocol can outperform today’s most efficient implementations of Yao’s garbled circuits, but highly depends on a low network latency. As a first step to overcome these latency issues, we summarize depth-optimized circuit constructions for various standard tasks. As application scenario we consider privacy-preserving face recognition and show that our optimized framework is up to 100 times faster than previous works even in settings with high network latency.