Today's Internet frequently suffers from unreliable packet forwarding and security vulnerabilities, while it largely depends on the Border Gateway Protocol (BGP) that obscures routing complexities from end-hosts. Path-Aware Networking (PAN) is a novel paradigm that aims to overcome BGP's limitations by offering end-hosts control over packet routing. However, an optimal path choice based on PAN requires realtime performance data about the different path options. Programmable data planes and Software-Defined Networking (SDN) enable to add real-time telemetry data to user packets. However, existing implementations are limited to single administrative domains, lacking the security and privacy required for interdomain telemetry. This paper proposes the first hardware implementation of Inter-Domain In-band Network Telemetry (ID-INT), extending the P4-based SCION border router targeting Tofino 2. Leveraging SCION's path-aware features and robust security mechanisms, like the DRKey system for secure key derivation, our ID-INT implementation on Tofino ensures telemetry data authenticity and integrity. We demonstrate the benefits of ID-INT on Tofino by evaluating the Deadline-aware Multipath Transport Protocol (DMTP) as an example application utilizing ID-INT data to improve its responsiveness to network conditions. Our testbed experiments show that ID-INT enables DMTP to dynamically adjust the sending rate based on instantaneous queue length data, highlighting ID-INT's practical benefits in enhancing an application's performance.

This work in progress paper presents an architecture for an IoT gateway that can be seamlessly integrated into an industrial setting running older generations PLCs to enable IoT connectivity and data analytics for sensors to applications with minimal setup. Internet of Things (IoT) is a major paradigm of Industry 4.0 where cyber-physical systems are connected to the Internet. By integrating the measurement data from sensors into an IoT environment, more value can be derived from the data through advanced analytics using machine learning or deep learning. At present, most of the sensors in industrial plants are connected to Programmable Logic Controllers (PLCs). To enable IoT connectivity for sensors connected to PLCs, the controller often needs to be retrofitted with IoT gateways, especially in the case of older generation PLCs. The cost of retrofitting older generation PLCs with IoT gateways and the extensive configuration and setup required to use the sensor data in running environments can prevent industries and factories from utilizing the benefits of IoT. The presented architecture addresses these issues by describing an approach to capture sensor data from PLCs through Modbus protocol, transform the data into an IoT communication standard and provides contextual information of data sources on demand to enable plug and play IoT connectivity. A method to store, process and visualize the sensor data is also presented. Finally, a prototypical implementation of the presented architecture and method is introduced.