I-Corps: Hardening Programs Against Cyber Attacks

该项目从编程语言的角度解决网络安全问题。在当今最常用的五种编程语言中，有三种(包括最常用的C语言)导致程序极易受到网络攻击。其目标是将软件多样性确立为一种可行的网络防御方法，与其他防御措施协同使用。将该团队的研究商业化可能会改变软件的分发方式，以及当前对已部署软件的威胁背后的许多假设和模型。这个团队？S的“多编译器方法”可能会为很多安全问题提供解决方案。多编译器方法在将二进制文件分发给最终用户之前提供自动强化。使用强化的二进制文件并不能消除通过修补源代码和分发软件来修复安全漏洞的需要。然而，这大大增加了攻击者利用漏洞的难度。这一解决方案可能会吸引广泛的客户：软件供应商以及美国政府等大型机构。该团队的项目旨在将自动化的软件多样化机制商业化，提供一种既可以扩展到潜在的数百万用户(即数百万个截然不同的软件版本)的解决方案，也可以扩展到非常大的程序，甚至包括操作系统。为此，他们计划通过“云计算”服务提供变异机制，在这种服务中，后续请求者下载相同软件的不同唯一变体，而不是完全相同的二进制文件。此外，该团队还致力于通过“标准化”用户返回的错误报告，使他们的解决方案对开发人员几乎完全透明。。

This project addresses cyber security from a programming language perspective. Of the five most used programming languages today, three (including the most used language C) lead to programs that are highly susceptible to cyber-attacks. The goal is to establish software diversity as a viable cyber defense approach, to be used in concert with other defenses. Commercializing this team's research may have the effect of changing both the way that software is distributed as well as many of the assumptions and models underlying current threats to deployed software. This team?s 'multi-compiler approach' may provide solutions for many security problems. The multi-compiler approach provides automatic hardening of binaries before they are distributed to end users. The use of hardened binaries does not remove the need to fix security vulnerabilities by patching the source code and distributing software. However, it makes it substantially harder for attackers to exploit vulnerabilities. This solution may appeal to a broad spectrum of customers: software vendors as well as large institutions such as the US government. This team's project aims to commercialize automated software diversification mechanisms, providing a solution that can scale both to potentially millions of users (i.e., millions of distinctly different software versions) as well as very large programs, including even operating systems. To this end, they plan to provide the variation mechanism through a "cloud computing" service in which subsequent requesters download different unique variants of the same software, rather than identical binaries. In addition, this team is also working on making their solution almost completely transparent to developers by "normalizing" error reports that flow back from users. .