A Hardware in the Loop Benchmark Suite to Evaluate NIST LWC Ciphers on Microcontrollers
A Hardware in the Loop Benchmark Suite to Evaluate NIST LWC Ciphers on Microcontrollers
复制标题
用于评估微控制器上的 NIST LWC 密码的硬件在环基准套件
DOI：
--
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
J. Mottok
中科院分区：
文献类型：
--
作者：
Sebastian Renner;Enrico Pozzobon;J. Mottok
.美国国家标准与技术研究院（NIST）于2018年启动了轻量级密码算法的标准化进程。在第一轮结束时，32份提交的文件被选为第二轮候选人。NIST允许第二轮提交的设计人员提供其规格和实现包的小更新。在这项工作中，我们引入了一个基准框架，用于评估NIST轻量级密码（LWC）候选人在嵌入式平台上的性能。我们展示了该框架的功能和应用，并解释了其设计原理。此外，我们还提供了有关我们如何在NIST LWC竞赛中展示最新性能数据的信息。在本文中，我们提出了一个摘录，我们的软件基准测试结果的速度和内存的要求选定的密码。所有最新的结果，包括在五个不同的微控制器上对每个第二轮算法的多个变体进行基准测试的不同测试用例，都会定期发布到公共网站上。虽然最初只有参考实现可用，但随着更优化的实现的开发，在多个平台上自动测试候选算法性能的能力变得特别重要。最后，我们展示了该框架如何在不同的方向上扩展：可以很容易地添加对更多目标平台的支持，可以测试不同类型的算法，以及可以获得其他测试指标。本文的重点应该放在框架设计和测试方法上，而不是当前的结果，特别是参考代码。
. The National Institute of Standards and Technology (NIST) started the standardization process for lightweight cryptography algorithms in 2018. By the end of the first round, 32 submissions have been selected as 2nd round candidates. NIST allowed designers of 2nd round submissions to provide small updates on both their specifications and implementation packages. In this work, we introduce a benchmarking framework for evaluating the performance of NIST Lightweight Cryptography (LWC) candidates on embedded platforms. We show the features and application of the framework and explain its design rationale. Moreover, we provide information on how we aim to present up-to-date performance figures throughout the NIST LWC competition. In this paper, we present an excerpt of our software benchmarking results regarding speed and memory requirements of selected ciphers. All up-to-date results, including benchmarking different test cases for multiple variants of each 2nd round algorithm on five different microcontrollers, are periodically published to a public website. While initially only the reference implementations were available, the ability of automatically testing the performance of the candidate algorithms on multiple platforms becomes especially relevant as more optimized implementations are developed. Finally, we show how the framework can be extended in different directions: support for more target platforms can be easily added, different kinds of algorithms can be tested, and other test metrics can be acquired. The focus of this paper should rather lay on the framework design and testing methodology than on the current results, especially for reference code.