CAMO: Counterfeit Attestation MOdule for Electronics Supply Chain Tracking and Provenance

供应链全球化和电子商务导致假冒电子产品贩运大幅增加，而流行病引起的微芯片短缺只会加剧这种情况。当市场上没有高需求的电子产品时，它们就成为造假者的主要目标。除了因知识产权（IP）侵权而损失数十亿美元外，假冒电子产品通常质量不合格，从而为医疗保健、食品和农业、通信、交通、能源、制造业、应急服务、国防等关键部门的系统和基础设施带来风险。最近的立法，如CHIPS法案，强调了打击假冒行为的程序的必要性，但主要利益相关者多年来一直抵制采用必要的技术。假冒检测和避免可以通过物理检查、电气测试或防伪设计（DfAC）来实现。前两者一直是主流，但不太可靠，更昂贵，而且缺乏自动化。虽然DfAC原语只能包含在全新的电子产品中，但它们在准确性、认证成本和可扩展性方面更有前途。此外，包括区块链在内的集中式和分散式分类账技术正在发展，以支持DfAC原始验证以及电子产品的不可变可追溯性和来源。该项目旨在解决DfAC原始集成到微芯片和系统中的主要障碍。将开发伪造认证模块（简称CAMO），以有效地收集印刷电路板（PCB）上所有微芯片的DfAC输出，验证PCB本身没有被篡改，并保护结果与数字分类账的通信，以进行加密认证。此外，还将探索可以方便地集成到常见微芯片模块中的轻量级DfAC原语。 该项目可扩展、方便且廉价地检测假冒电子产品，将为商业和国防行业节省资金，并将这些资金转移给消费者和纳税人。上述目标将通过四项任务实现。首先，将探索一种称为谐振频率锁定（Res-FLO）的新颖电路用于PCB的自包含认证。Res-FLO可从PCB配电网络捕获独特的、依赖于工艺变化的特征，以检测克隆和篡改的PCB。接下来，为了支持芯片认证并在CAMO通信协议中提供安全性，将研究轻量级DfAC原语以检测两种最常见的伪造芯片类型-回收和克隆。前者将通过测量模拟和数字低压差调节器（LDO）的退化来检测，而后者可以通过采用物理不可克隆功能（PUF）来实现。第三，为PCB内的通信和账本/区块链提供机密性和完整性的廉价协议将由前两项任务的元素组成。也就是说，CAMO将利用大的输入/输出空间、机器学习（ML）攻击抵抗能力和对强PUF的多次测量来增加真实/伪造分类的安全性和置信度，而不需要昂贵的加密和纠错方案。最后，CAMO原语和协议将被录制出来用于概念验证，并捕获开销、老化和噪声的影响、对攻击的抵抗力等。该奖项反映了NSF的法定使命，并被认为值得通过使用基金会的知识价值和更广泛的影响审查标准进行评估来支持。

Supply chain globalization and e-commerce have caused a substantial rise in counterfeit electronics trafficking, which was only exacerbated by the pandemic-induced microchip shortage. When high-demand electronics are not available in the marketplace, they become prime targets for counterfeiters. Aside from billions of dollars lost from intellectual property (IP) infringement, counterfeit electronics are often of substandard quality, thereby creating risks for systems and infrastructure in critical sectors such as healthcare, food and agriculture, communications, transportation, energy, manufacturing, emergency services, defense, and more. Recent legislation, such as the CHIPS Act, stresses the need for procedures to combat counterfeiting, but key stakeholders have resisted adoption of the requisite technologies for years. Counterfeit detection and avoidance can be accomplished by physical inspection, electrical tests, or design-for-anti-counterfeit (DfAC). The former two have been the mainstay, but are less reliable, more expensive, and lack automation. Although DfAC primitives can only be included in brand new electronics, they are more promising with respect to accuracy, authentication cost, and scalability. Further, centralized and decentralized ledger technologies, including blockchains, are advancing to support DfAC primitive verification and for immutable traceability and provenance of electronics. This project aims to address the major barriers for DfAC primitive integration into microchips and systems. Counterfeit attestation modules (CAMOs for short) will be developed to efficiently collect DfAC outputs from all the microchips on a printed circuit board (PCB), verify that the PCB itself has not been tampered, and protect communication of results to a digital ledger for en masse authentication. In addition, lightweight DfAC primitives that can be conveniently incorporated into common microchip modules will also be explored. Scalable, convenient, and inexpensive detection of counterfeit electronics from this project shall result in savings for commercial and defense industries that will be passed on to consumers and taxpayers.The above goals will be achieved through four tasks. First, a novel circuit called Resonant Frequency Lock-On (Res-FLO) will be explored for self-contained authentication of PCBs. Res-FLO captures unique, process variation dependent signatures from PCB power distribution networks to detect cloned and tampered PCBs. Next, to support chip authentication and provide security in CAMO communication protocols, lightweight DfAC primitives will be investigated for detecting the two most prevalent counterfeit chip types – recycled and cloned. The former will be detected by measuring degradation in analog and digital low dropout regulators (LDOs) while the latter can be achieved by adopting physical unclonable functions (PUFs). Third, inexpensive protocols that provide confidentiality and integrity for communications within the PCB and to a ledger/blockchain will be composed from elements of the first two tasks. Namely, CAMO will exploit the large input/output space, machine learning (ML) attack resistance, and multiple measurements of strong PUFs to increase the security and confidence of authentic/counterfeit classification without the need for expensive cryptographic and error correction schemes. Finally, the CAMO primitives and protocols will be taped out for proof-of-concept and to capture overheads, impacts of aging and noise, resistance to attacks, etc.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.