Internet of Things (IoT) devices are increasingly deployed nowadays in various security-sensitive contexts, e.g., inside homes or in critical infrastructures. The data they collect is of interest to attackers as it may reveal living habits, personal data, or the operational status of specific targets. This paper presents an approach to counter software manipulation attacks against running processes, data, or configuration files on an IoT device, by exploiting trusted computing techniques and remote attestation. We have used a Raspberry Pi 4 single-board computer device equipped with Infineon Trusted Platform Module (TPM) v2, acting as an attester. A verifier node continuously monitors the attester and checks its integrity through remote attestation protocol and TPM-enabled operations. We have exploited the Keylime framework from MIT Lincoln Laboratories as remote attestation software. Through tests, we show that remote attestation can be performed within short time (in order of seconds), allowing to restrict the window of exposure of such devices to attacks against the running software and/or hosted data.

Smart cities deploy large numbers of sensors and collect a tremendous amount of data from them. For example, Advanced Metering Infrastructures (AMIs), which consist of physical meters that collect usage data about public utilities such as power and water, are an important building block in a smart city. In a typical sensor network, the measurement devices are connected through a computer network, which exposes them to cyber attacks. Furthermore, the data is centrally managed at the operator’s servers, making it vulnerable to insider threats.Our goal is to protect the integrity of data collected by large-scale sensor networks and the firmware in measurement devices from cyber attacks and insider threats. To this end, we first develop a comprehensive threat model for attacks against data and firmware integrity, which can target any of the stakeholders in the operation of the sensor network. Next, we use our threat model to analyze existing defense mechanisms, including signature checks, remote firmware attestation, anomaly detection, and blockchain-based secure logs. However, the large size of the Trusted Computing Base and a lack of scalability limit the applicability of these existing mechanisms. We propose the Feather-Light Blockchain Infrastructure (FLBI) framework to address these limitations. Our framework leverages a two-layer architecture and cryptographic threshold signature chains to support large networks of low-capacity devices such as meters and data aggregators. We have fully implemented the FLBI’s end-to-end functionality on the Hyperledger Fabric and private Ethereum blockchain platforms. Our experiments show that the FLBI is able to support millions of end devices.