Dependable Internet of Things

Abstract

Wireless networked embedded systems are increasingly used for safety-critical applications such as smart production or networked cars, where failures may have severe impact. Therefore, strict dependability requirements have to be met. This is difficult to achieve, however, as these applications often operate in harsh environments or are exposed to attacks. In this talk we present recent research results obtained in the Dependable Things research center at TU Graz which aims at increasing the dependability of the IoT for safety-critical applications. Specifically, we present a single-anchor approach to robust and accurate localization using UWB, a method to analyze software for potential side-channel leakage, and an approach to automatically learn models of protocols used in the IoT in order to formally verify their correct implementation and interoperability. There is consensus among the relevant industry and standardization communities that a key element in 5G mobile networks will be network slicing. The idea is to allow the mobile infrastructure to be “sliced” into logical networks, where each slice is a collection of resources and functions that includes software modules running at different locations as well as the nodes' computational and communication resources. The intention is to tailor each slice to support a specific service, providing only what is necessary for the service while avoiding unnecessary overheads and complexity. This provides a basis for efficient infrastructure sharing among diverse entities, ranging from classical or virtual mobile network operators to new players that simply view connectivity as a service, where each of these entities may be running one or more slices. This talk will focus on the key enablers for network slicing and the research challenges involved in realizing this technology. Current standardization activities will be reviewed along with the contributions of major research projects, such as the H2020 5G-NORMA and 5G-MoNArch projects. A key problem underlying network slicing is enabling efficient sharing of mobile network resources. Various approaches considered in 3GPP will be analysed, ranging from per-reservation based schemes (where network slices reserve the required resources in advance) to others based on network shares (where resources are allocated based on predetermined shares). The performance and behavior of the various approaches will be studied based on analytical tools including optimization, game theory and machine learning. Buildig on these analyses, we will provide some insights on the stability, peformance, optimality and level of customization enabled by the various approaches.