Secure Codesign: Achieving Optimality Without Revealing

摘要

Information leakage can lead to loss of intellectual property and competitive edge. One of the primary sources of information leakage in collaborative design is sharing confidential information with collaborators, who may be also collaborating with competitors. Hiding information from collaborators is challenging in codesign because it can lead to inferior and suboptimal solutions. Therefore, there is a need for techniques that enable designers to protect confidential information from their collaborators while achieving solutions that are as good as those obtained when full information is shared. To address this need, we propose a secure codesign (SCD) framework that enables designers to achieve optimal solutions without sharing confidential information. It is built on two principles: adding/multiplying a parameter with a large random number hides the value of the parameter, and adding/multiplying a large number is orders of magnitude faster than using existing cryptographic techniques. Building on the protocols for basic arithmetic computations, developed in our earlier work, we establish protocols for higher order computations involved in design problems. The framework is demonstrated using three codesign scenarios: requirements-driven codesign, objective-driven codesign, and Nash noncooperation. We show that the proposed SCD framework enables designers to achieve optimal solutions in all three scenarios. The proposed framework is orders of magnitude faster than competing (but impractical for engineering design) cryptographic methods such as homomorphic encryption, without compromising on precision in computations. Hence, the proposed SCD framework is a practical approach for maintaining confidentiality of information during codesign.