Abstract

International audience; Semi-autonomous and fully-autonomous systems must have knowledge about the objects in their environment to ensure a safe navigation. Modern approaches implement deep learning techniques to train a neural network for object detection. This project will study the effectiveness of using several promising algorithms such as Faster R-CNN, SSD, and different versions of YOLO, to detect, classify, and track objects in near real-time fluvial domain. Since no dataset is available for this purpose in literature, we first started by annotating a dataset of 2488 images with almost 35 400 annotations for training the convolutional neural network architectures. We made this data set openly accessible for the community working on this area. The other contribution of this research is the adaptation and the configuration of deep learning techniques used in other domains such as maritime and road domain to fluvial domain for autonomous vessels in which high accuracy and fast processing are vital. Experiments demonstrated that detecting objects in such environment is plausible in near real time with the selected algorithms.

Abstract

end-to-end encryption on the Internet is becoming more prevalent, techniques such as deep packet inspection (DPI) can no longer be expected to be able to classify traffic. In many cellular networks a large fraction of all traffic is video traffic, and being able to divide flows in the network into video and non-video can provide considerable traffic engineering benefits. In this study we examine machine learning based flow classification using features that are available also for encrypted flows. Using a data set of several several billion packets from a live cellular network we examine the obtainable classification performance for two different ensemble-based classifiers. Further, we contrast the classification performance of a statistical-based feature set with a less computationally demanding alternate feature set. To also examine the runtime aspects of the problem, we export the trained models and use a tailor-made C implementation to evaluate the runtime performance. The results quantify the trade-off between classification and runtime performance, and show that up to 1 million classifications per second can be achieved for a single core. Considering that only the subset of flows reaching some minimum flow length will need to be classified, the results are promising with regards to deployment also in scenarios with very high flow arrival rates. HITS

Abstract

The European Commission Horizon 2020 AUTOPILOT (AUTOmated driving Progressed by Internet Of Things) is aiming to exploit the IoT ecosystem to integrate connected cars and transform them in automated moving "objects". One of the key challenges encountered in the project is to ensure the interoperability of the different components and IoT platforms serving e.g. in-vehicle and road-side devices and sensors. The adopted solution is the use of Federated IoT platforms, with the oneM2M Interoperability Platform used to ensure that all components are able to communicate with each other. This paper provides a high-level description of the project and its goals and then concentrates on the importance to ensure interoperability support for IoT platforms using the standard IoT platform provided by oneM2M. © 2019 IEEE.

Abstract

Wireless vehicular ad-hoc networks comprised solely of city taxis are investigated for their ability to deliver data across an urban environment. Openly available taxi trace datasets for Rome (Italy) and San Francisco (USA) are combined with respective building footprint and road network topology data from OpenStreetMap, to generate a realistic systems level model of a taxi V2V network. Analysis of LOS and NOLOS constraints on wireless transmission range suggests a minimum threshold of 50m is applicable to ensure LOS in over 90% of cases. Variations in taxi location sampling frequency and filtering techniques for the taxi trace datasets are also investigated. Overall vehicular network performance is computed for an all-to-one transmission scenario for both cities with varying taxi fleet size. Results suggest a non-linear relationship between increases in taxi fleet sizes and the reduction of end-to-end delay; doubling taxi fleet size (using a randomised data folding technique) reduces end-to-end delay by a factor of 0.6–0.7. However, doubling the fleet does not increase the fraction of delivered source messages, which saturates at 0.67–0.71 in most simulations. Finally it appears that taxi networks for delivering messages across urban environments are limited more by their routing than by the number of possible V2V exchanges. In a simulated one-to-all continuous V2V broadcast scenario, over 90% of the taxis within the fleet receive the source message within one hour of the original taxi passing the source node.