Advances like Industry 4.0 lead to a rising number of Internet-connected industrial deployments and thus an Industrial Internet of Things with growing attack vectors. To uphold a secure and safe operation of these deployments, industrial protocols nowadays include security features, e.g., end-to-end secure communication. However, so far, it is unclear how well these features are used in practice and which obstacles might prevent operators from securely running their deployments. In this research description paper, we summarize our recent research activities to close this gap. Specifically, we show that even secure-by-design protocols are by far no guarantee for secure deployments. Instead, many deployments still open the doors for eavesdropping attacks or malicious takeovers. Additionally, we give an outlook on how to overcome identified obstacles allowing operators to configure their deployments more securely.

Internet-wide studies provide extremely valuable insight into how operators manage their Internet of Things (IoT) deployments in reality and often reveal grievances, e.g., significant security issues. However, while IoT devices often use IPv6, past studies resorted to comprehensively scan the IPv4 address space. To fully understand how the IoT and all its services and devices is operated, including IPv6-reachable deployments is inevitable-although scanning the entire IPv6 address space is infeasible. In this paper, we close this gap and examine how to best discover IPv6-reachable IoT deployments. To this end, we propose a methodology that allows combining various IPv6 scan direction approaches to understand the findability and prevalence of IPv6-reachable IoT deployments. Using three sources of active IPv6 addresses and eleven address generators, we discovered 6658 IoT deployments. We derive that the available address sources are a good starting point for finding IoT deployments. Additionally, we show that using two address generators is sufficient to cover most found deployments and save time as well as resources. Assessing the security of the deployments, we surprisingly find similar issues as in the IPv4 Internet, although IPv6 deployments might be newer and generally more up-to-date: Only 39% of deployments have access control in place and only 6.2% make use of TLS inviting attackers, e.g., to eavesdrop sensitive data.

Security configurations remain challenging for trained administrators. Nowadays, due to the advent of the Internet of Things (IoT), untrained users operate numerous and heterogeneous Internet-facing services in manifold use case-specific scenarios. In this work, we close the growing gap between the complexity of IoT security configuration and the expertise of the affected users. To this end, we propose ColPSA, a platform for collective and privacy-aware security advice that allows users to optimize their configuration by exchanging information about what security can be realized given their IoT deployment and scenario.

Containerization allows bundling applications and their dependencies into a single image. The containerization framework Docker eases the use of this concept and enables sharing images publicly, gaining high momentum. However, it can lead to users creating and sharing images that include private keys or API secrets—either by mistake or out of negligence. This leakage impairs the creator’s security and that of everyone using the image. Yet, the extent of this practice and how to counteract it remains unclear. In this paper, we analyze 337,171 images from Docker Hub and 8,076 other private registries unveiling that 8.5% of images indeed include secrets. Specifically, we find 52,107 private keys and 3,158 leaked API secrets, both opening a large attack surface, i.e., putting authentication and confidentiality of privacy-sensitive data at stake and even allow active attacks. We further document that those leaked keys are used in the wild: While we discovered 1,060 certificates relying on compromised keys being issued by public certificate authorities, based on further active Internet measurements, we find 275,269 TLS and SSH hosts using leaked private keys for authentication. To counteract this issue, we discuss how our methodology can be used to prevent secret leakage and reuse.

The ongoing trend to move industrial appliances from previously isolated networks to the Internet requires fundamental changes in security to uphold secure and safe operation. Consequently, to ensure end-to-end secure communication and authentication, (i) traditional industrial protocols, e.g., Modbus, are retrofitted with TLS support, and (ii) modern protocols, e.g., MQTT, are directly designed to use TLS. To understand whether these changes indeed lead to secure Industrial Internet of Things deployments, i.e., using TLS-based protocols, which are configured according to security best practices, we perform an Internet-wide security assessment of ten industrial protocols covering the complete IPv4 address space. Our results show that both, retrofitted existing protocols and newly developed secure alternatives, are barely noticeable in the wild. While we find that new protocols have a higher TLS adoption rate than traditional protocols (7.2 % vs. 0.4 %), the overall adoption of TLS is comparably low (6.5 % of hosts). Thus, most industrial deployments (934,736 hosts) are insecurely connected to the Internet. Furthermore, we identify that 42 % of hosts with TLS support (26,665 hosts) show security deficits, e.g., missing access control. Finally, we show that support in configuring systems securely, e.g., via configuration templates, is promising to strengthen security.