A PUF Application: Are sensors the key to a secure IoT environment?

puf代表物理不可克隆功能，可用于唯一标识给定的设备/组件。PUF的前提是，由于制造过程中发生的微观物理变化，没有两个组件（即RAM模块）是相同的，并且几乎不可能准确且经济有效地复制。这些物理变化可以用来帮助提高安全性。这个项目正在追求的PUF的一个应用是使用PUF对标准的基于密钥的加密进行加密，这种加密通常实现得很差（在密钥管理和存储方面），并且未针对物联网设备进行优化（在功耗和处理能力方面）。主要目标整个研究项目可以分为3个阶段（项目），每个阶段本身都是一个项目。预计只有前两个项目才能完全完成。项目1 -指纹传感器对于给定的设备（或设备类型），该项目旨在使用多个传感器的输出对整个设备进行指纹识别。技术能力要求：*机器学习-指纹识别常用技术*从传感器中提取原始值*传感器背景理论（例如校准）*数学[向量，证明]*创建一个从多极传感器生成指纹的算法（*新颖性方面）关键参考论文：SensorID：传感器校准智能手机指纹识别预期持续时间：1.5 YearsProject 2 -传感器PUF创建和证明[主要焦点]本项目直接遵循第一个项目，使用从前一个项目中提取的值来创建基于传感器的PUF。到项目结束时，预计至少会对该实现进行有限的概念验证。技术能力要求：*数学[证明，向量]*传感器PUF的创建，实现和证明（*新颖性方面）关键参考论文：纹理救援：基于纹理图案的实用纸张指纹预计持续时间：1.5年项目3 -传感器PUF应用在时间允许的不太可能的情况下，第三个项目包括应用前几个项目的发现；将传感器PUF应用于现有的物联网产品/环境。在此之后，在传感器PUF实施之前和之后，将在有关的物联网环境上实施基于身份验证的攻击，以显示PUF的有效性。技术能力要求：*进行基于身份验证的攻击*模拟物联网环境*将项目2（实施PUF）应用于现有设置预计持续时间：~1年项目新颖性项目的主要新颖性是从传感器生成用于物联网环境中身份验证的PUF。PUF通常使用RAM、处理器、FPGA和ASIC来实现，但目前没有关于传感器PUF的论文（只有一项专利）。该项目打算回答的问题是：传感器（字面上）是安全物联网环境的关键吗？传感器生成的PUF能否可靠地用于提高物联网环境的安全性，而没有标准的基于密钥的加密的缺点？

Terminology ExplainedPUF stands for Physical Unclonable Function and can be used to uniquely identify a given device/component. PUF's are based on the premise that no two components (i.e. a RAM module) are the same due to the microscopic physical variations that occur during manufacturing and are near-impossible to replicate accurately and cost-effectively. These physical variations can be used to aid in security.One application of PUF's which is being pursued by this project is the use of PUF's for encryption over the standard key-based encryption which is often poorly implemented (withregard to key-management and storage) and un-optimised for IoT devices (in regard to powerconsumption and processing power).Key ObjectivesThe research project as a whole can be broken down into 3 stages (projects) with each stage acting as a project in itself. It is only expected for the first two projects to be fully completed.Project 1 - Fingerprinting SensorsOn a given device (or type of device), this project aims to fingerprint the device as a whole using the output of multiple sensors.Technical Abilities Required:*Machine Learning - Commonly used technique for fingerprinting*Extracting Raw values from sensors*Sensor background theory (e.g. Calibration)*Maths [Vectors, proofs]*Create an algorithm for fingerprint generation from multipole sensors (*Novelty Aspect)Key Reference Paper: SensorID: Sensor Calibration Fingerprinting for SmartphonesExpected Duration: 1.5 YearsProject 2 - Sensor PUF Creation and Proof [Main Focus]This project follows directly from the first by using the extracted values from the previous project to create a sensor-based PUF. By the end of the project, a limited proof-of-concept of this implementation is expected, at a minimum.Technical Abilities Required:*Maths [Proofs, Vectors]*Sensor PUF Creation, Implementation and Proof (*Novelty Aspect)Key paper of reference: Texture to the Rescue: Practical Paper Fingerprinting Based on Texture PatternsExpected Duration: 1.5 YearsProject 3 - Sensor PUF ApplicationIn the unlikely scenario that time permits, this third project consists of applying the findings from the previous projects; applying the sensor PUF to an existing IoT product/environment. Following this, an authentication-based attack will be enacted on the IoT environment in question prior to and following the implementation of the sensor PUF to show the effectiveness of the PUF.Technical Abilities Required:*Conducting an Authentication-Based Attack*Emulating an IoT Environment*Applying Project 2 (Implementing a PUF) to Existing Set-UpExpected Duration: ~1 YearProject NoveltyThe overarching novelty of the project is the generation of PUF's from sensors to be used in authentication in an IoT environment. PUF's are commonly implemented using RAM, Processors, FPGA's and ASIC's, but there are currently no papers on sensor PUF's (only a patent). What question does project intend to answerThe question the project intends to answer is: Are sensors (literally) the key to a secure IoT environment? Can PUF's generated from sensors reliably be used to increase the security of an IoT environment without the drawbacks of standard key-based encryption?