In this paper, we propose a probably secure key agreement scheme based on module learning with rounding (MLWR) problem by modifying the reconciliation mechanism to obtain a shared key. In addition, compared with other schemes with the same properties, the error probability is relatively small. The detailed security analysis is provided against a passive adversary. Then, a modified software package is developed to generate a parameter set for any security level. Primal and dual attack costs are also computed. The main contribution of this paper is to construct a modified reconciliation mechanism and obtain agreed values with low error probability and bandwidth.

在本文中，我们通过修改协调机制以获得共享密钥，提出了一种基于舍入模块学习（MLWR）问题的可能安全的密钥协商方案。此外，与其他具有相同属性的方案相比，错误概率相对较小。针对被动对手提供详细的安全分析。然后，开发修改后的软件包以生成任何安全级别的参数集。还计算了原始和双重攻击成本。本文的主要贡献是构建了一种改进的协调机制，并以较低的错误概率和带宽获得了一致的值。