Machine Learning Protocol for Secure 5G Handovers

The fifth generation (5G) networks are characterized with ultra-dense deployment of base stations with limited footprint. Consequently, user equipment’s handover frequently as they move within 5G networks. In addition, 5G requirements of ultra-low latencies imply that handovers should be executed swiftly to minimize service disruptions. To preserve security and privacy while at the same time maintaining optimal performance during handovers, numerous schemes have been developed. However, majority of these techniques are either limited to security and privacy or address only performance aspect of the handover mechanism. As such, there is need for a novel handover authentication protocol that addresses security, privacy and performance simultaneously. This paper presents a machine learning protocol that not only facilitates optimal selection of target cell but also upholds both security and privacy during handovers. Formal security analysis using the widely adopted Burrows–Abadi–Needham (BAN) logic shows that the proposed protocol achieves all the six formulated under this proof. As such, the proposed protocol facilitates strong and secure mutual authentication among the communicating entities before generating the shares session key. The derived session key protected the exchanged packets to avert attacks such as forgery. In addition, informal security evaluation of the proposed protocol shows that it offers perfect forward key secrecy, mutual authentication any user anonymity. It is also demonstrated to be robust against attacks such as denial of service (DoS), man-in-the-middle (MitM), masquerade, packet replays and forgery. In terms of performance, simulation results shows that it has lower packets drop rate and ping–pong rate, with higher ratio of packets received compared with improved 5G authentication and key agreement (5G AKA’) protocol. Specifically, using 5G AKA’ as the basis, the proposed protocol reduces the handover rate by 94.4%, hence the resulting handover signaling is greatly minimized.

     

Large-Scale Gaussian Process Inference with Generalized Histogram Intersection Kernels for Visual Recognition Tasks

We present new methods for fast Gaussian process (GP) inference in large-scale scenarios including exact multi-class classification with label regression, hyperparameter optimization, and uncertainty prediction. In contrast to previous approaches, we use a full Gaussian process model without sparse approximation techniques. Our methods are based on exploiting generalized histogram intersection kernels and their fast kernel multiplications. We empirically validate the suitability of our techniques in a wide range of scenarios with tens of thousands of examples. Whereas plain GP models are intractable due to both memory consumption and computation time in these settings, our results show that exact inference can indeed be done efficiently. In consequence, we enable every important piece of the Gaussian process framework—learning, inference, hyperparameter optimization, variance estimation, and online learning—to be used in realistic scenarios with more than a handful of data.     

amePLM: a platform providing information provision in engineering

The procedure for the engineering of products and production systems in manufacturing companies is often distributed over several organisational units. The different units involved in these procedures use their respective methods and tools, resulting in silo-like and distributed information and data sources for engineering. In this paper, the open platform amePLM is introduced. This platform is based on a semantic data model. The ontology as explicitly formulated data model allows an integrated view on data and information available in the systems used along the product lifecycle, and the automatic provision of suitable information to the user. Furthermore, this open approach allows the linking of the solution to existing engineering software systems in the sense of a continuous flow of information.     

Electrocatalytic applications of platinum-decorated TiO<Subscript>2</Subscript> nanotubes prepared by a fully wet-chemical synthesis

Pt-decorated \(\hbox {TiO}_{2}\) nanotubes Pt@TiO₂ are prepared only by applying a set of facile wet-chemical redox reactions to ion track-etched polycarbonate templates. First, a homogeneous layer of Pt nanoparticles is deposited onto the complex template surface by reducing potassium tetrachloroplatinate with absorbed dimethylaminoborane. Second, the template is coated with a conformal \(\hbox {TiO}_{2}\) layer, using a chemical bath deposition reaction based on titanium(III) chloride. After the removal of the template, the rutile-type \(\hbox {TiO}_{2}\) nanotubes remain decorated with Pt nanoparticles and nanoparticle-clusters on their outside. During the process, neither vacuum techniques nor external current sources or addition of heat are employed. The crystallinity, composition, and morphology of the composite nanotubes are analysed by X-ray diffraction, scanning and transmission electron microscopy as well as by energy-dispersive X-ray spectroscopy. Finally, the obtained materials are examplarily applied in the electrooxidation of ethanol and formic acid, and their performances have been evaluated. Compared to conventional carbon black-supported Pt nanoparticles, the Pt@TiO₂ nanotubes show higher reaction rates. Mass activities of 2.36 \(\hbox {A}\hbox { mg}_{\rm Pt}^{-1}\hbox { cm}^{-2}\) are reached in ethanol oxidation and 7.56 \(\hbox {A}\hbox { mg}_{\rm Pt}^{-1}\hbox { cm}^{-2}\) in the formic acid oxidation. The present structures are able to exploit the synergy of Pt and \(\hbox {TiO}_{2}\) with a bifunctional mechanism to result in powerful but easy-to-fabricate catalyst structures. They represent an easily producible type of composite nanostructures which can be applied in various fields such as in catalytics and sensor technology.     

Bootstrapping Inference in the IDP Knowledge Base System

Declarative systems aim at solving tasks by running inference engines on a specification, to free their users from having to specify how a task should be tackled. In order to provide such functionality, declarative systems themselves apply complex reasoning techniques, and, as a consequence, the development of such systems can be laborious work. In this paper, we demonstrate that the declarative approach can be applied to develop such systems, by tackling the tasks solved inside a declarative system declaratively. In order to do this, a meta-level representation of those specifications is often required. Furthermore, by using the language of the system for the meta-level representation, it opens the door to bootstrapping: an inference engine can be improved using the inference it performs itself.One such declarative system is the IDP knowledge base system, based on the language \(\rm FO(\cdot)^{\rm IDP}\), a rich extension of first-order logic. In this paper, we discuss how \(\rm FO(\cdot)^{\rm IDP}\) can support meta-level representations in general and which language constructs make those representations even more natural. Afterwards, we show how meta-\(\rm FO(\cdot)^{\rm IDP}\) can be applied to bootstrap its model expansion inference engine. We discuss the advantages of this approach: the resulting program is easier to understand, easier to maintain, and more flexible.