基于时空选择性相互作用的物联网可信认证

Existing Internet of things (IoT) devices often use cipher/password-based authentication, which is vulnerable to clone attacks, complicated and not easy to update keys. The hardware fingerprint-based authentication is complicated, with coarse-grained distinguishability, and cannot resist replay attacks. Both are with fatal security flaws in IoT. For performing the trustworthy IoT authentication, we propose to leverage the spatiotemporal selective interaction among IoT devices as well as the ones between IoT devices and physical environments. Our research is based on two classic IoT systems, passive RFID and smart phones. Correspondingly, we study two such interactions, the mutual inductance between a pair of tags and the influence from environmental illumination on the facial appearance of users, respectively. We aim to achieve breakthroughs in the interaction-oriented modeling, feature extraction, fingerprint generation, temporal and spatial selectivity of the environment, as well as the quantification, evaluation and balance of randomness of random sources. As consequence, we can generate device fingerprints with spatiotemporal correlation and high-randomness and prior-knowledge-free shared keys for secure and reliable IoT authentication methods. We will deploy prototypes in real IoT systems to verify the proposed theories and methods. This research will facilitate to break through the bottleneck of conventional authentications, overcome their shortcomings, such as poor compatibility, accuracy, and distinguishability. The research results will help to promote the trustworthy authentication in large-scale IoT systems.

现有的物联网(IoT)设备常采用密码/口令的认证方式，易被克隆，运算复杂，不易更新密钥；而设备硬件指纹的认证方式，产生复杂、区别度低、无法抵御重放攻击，均存在巨大安全隐患。课题组提出利用IoT设备之间以及设备与物理环境之间具有时空选择特性的相互作用，面向被动式无源标签和智能手机两种典型IoT设备，具体研究标签对间的互感耦合作用和环境光照对用户面部外观的影响为代表的相互作用，在相互作用建模、特征提取、指纹生成，时空环境的选择性与随机源的随机性的量化、评估及平衡等关键技术上取得突破，生成时空相关、高随机设备指纹和无先验知识的共享密钥，实现安全可靠的IoT可信认证方法，通过在真实的IoT系统中开发和部署示范性原型系统，验证提出的理论和方法。本课题的研究将有助于突破IoT认证的旧有模式，克服现有方法兼容性差、准确率低、区分度小的技术瓶颈，推动IoT系统可信认证长期大规模应用。

针对现有的物联网(IoT)认证方式存在的易被克隆、运算复杂、不易更新密钥、无法抵御重放和伪造攻击等安全隐患，课题组提出利用人（用户）、机（设备）、物（物理世界）之间具有时空选择特性的相互作用，面向典型IoT设备，研究基于时空的物理层相互作用，在相互作用建模、特征提取、指纹生成，时空环境的选择性与随机源的随机性的量化、评估及平衡等关键技术上取得突破，实现系列安全可靠的IoT可信认证方法，通过在真实的IoT系统中开发和部署示范性原型系统，验证了提出的理论和方法。.本项目总体完成情况良好，所涉及的研究内容均已完成，预期指标均已实现，并取得了较为丰富的成果。其中，研发原型系统4套；发表国际高水平学术论文28篇（其中SCI期刊10篇、会议论文17篇，CCF推荐A类会议10篇，CCF推荐B类会议6篇），申请技术发明专利5项，其中授权4项，登记软件著作权2项；项目进展过程中产出的成果获得CCF A类会议IEEE INFOCOM 2019年最佳论文奖、2021年最佳论文提名奖，CCF B类会议ACM Sensys 2021年最佳论文提名奖，CCF C类会议GlobeCom 2019年最佳论文奖。.本项目的研究将有助于突破IoT认证的旧有模式，克服现有方法兼容性差、准确率低、区分度小的技术瓶颈，推动IoT系统可信认证长期大规模应用。