CheapSMC: A Framework to Minimize Secure Multiparty Computation Cost in the Cloud

Abstract

Secure multi-party computation (SMC) techniques are increasingly becoming more efficient and practical thanks to many recent novel improvements. The recent work have shown that different protocols that are implemented using different sharing mechanisms (e.g., boolean, arithmetic sharings, etc.) may have different computational and communication costs. Although there are some works that automatically mix protocols of different sharing schemes to fasten execution, none of them provide a generic optimization framework to find the cheapest mixed-protocol SMC execution for cloud deployment. In this work, we propose a generic SMC optimization framework CheapSMC that can use any mixed-protocol SMC circuit evaluation tool as a black-box to find the cheapest SMC cloud deployment option. To find the cheapest SMC protocol, CheapSMC runs one time benchmarks for the target cloud service and gathers performance statistics for basic circuit components. Using these performance statistics, optimization layer of CheapSMC runs multiple heuristics to find the cheapest mix-protocol circuit evaluation. Later on, the optimized circuit is passed to a mix-protocol SMC tool for actual executable generation. Our empirical results gathered by running different cases studies show that significant cost savings could be achieved using our optimization framework.

HyCC

“…The computation and communication costs of a hybrid MPC protocol depend on the combined costs to evaluate each module in the selected protocol plus the time to convert between modules, when evaluating them with different protocols. However, in contrast to previous works, i.e., [32] and [44], we observe that the optimization problem, i.e., achieving minimal costs for a given decomposition, is not only a protocol selection problem but also a scheduling problem. Namely, the evaluation order of parallel modules, i.e., modules without sequential dependencies, can significantly influence the effectiveness of protocol selection, and thus the overall protocol runtime.…”

“…A very interesting use case is the optimization of the protocol's online runtime, yet, various other cost models are also of interest. For example optimizing the cloud computing costs, i.e., the price to pay for computational power and traffic to perform a computation using cloud service providers, has been discussed in [44]. Further examples are the total protocol runtime including or excluding the time spent on preprocessing depending on the use case of the application, the pure communication costs when considering a constrained network connection, or the power consumption when considering mobile devices.…”

“…Optimal protocol selection is an optimization problem, where the best mapping is identified in regard to the cost model that considers the cost to evaluate each circuit in the respective MPC protocol as well as the conversion costs between the different representations. The concept of protocol selection has previously been studied independently from compilation in [32,44]. Kerschbaum et al [32] investigated protocol selection for a combination of Yao's garbled circuits and additive homomorphic encryption.…”

“…Second, they presented a greedy optimization algorithm, which is capable of optimizing larger functionalities. Pattku et al [44] used similar heuristics to optimize the protocol selection for minimal cloud computing costs, i.e., the price to pay a cloud provider to perform a computation, as a cost model.…”

Fog Computing as Privacy Enabler

“…While we agree that there are indeed some privacy (and security) risks in fog computing, we believe that these are largely technical challenges that can be surmounted with existing technology. On the other hand, fog computing provides invaluable privacy opportunities that are primarily of structural nature and can valuably complement existing, especially cryptographic techniques such as homomorphic encryption [7], secure multi-party computation [9], or mechanisms for differential privacy [10]. These structure-based opportunities can be used to significantly heighten data privacy, as compared to well-established cloud-centric models of data processing.…”