一种基于安全多方计算的边缘学习协议

An Edge Learning Protocol Based on Secure Multi-party Computation

提出一种基于椭圆曲线的安全多方计算协议,旨在解决边缘学习中边缘设备算力和能耗受限难以通过强加密传输抵御网络攻击的问题。该协议采用椭圆曲线加密算法,在同等安全条件下具有较低的计算复杂度,实现了一种在加密状态下矩阵内积的计算方法。该协议主要应用于边缘学习中的云边缘协作计算,以提高其安全性。实验结果表明,与联邦学习和差分隐私等方法相比,该协议具有更低的计算复杂度和计算时间,同时能够保证边缘学习场景下的隐私和准确性。

An improved elliptic curve-based secure multiparty computation protocol is proposed to address the challenge of limited computing power and energy consumption of edge devices in edge learning,making it difficult to resist network attacks through strong encrypted transmission.The protocol is mainly used for cloud-edge collaborative computing in edge learning and realizes a method for calculating matrix inner products in an encrypted state.This method uses elliptic curve encryption algorithm and has lower computational complexity under the same security conditions,which can improve the security of cloud-edge collaborative computing at lower computational cost.Experimental analysis on different datasets shows that compared with other methods such as federated learning and differential privacy,the proposed protocol can effectively reduce computational complexity and computation time while ensuring privacy and accuracy in edge learning scenarios.