Misbehavior Detection in C-ITS Using Deep Learning Approach

Cooperative Intelligent Transport Systems (C-ITS) aim to enhance road safety, traffic efficiency and environmental performance of road transportation by enabling communication between the vehicles and the road infrastructure. Today, vehicle manufacturers and road operators are investing significantly in C-ITS technology and infrastructure implementation to realise its potential benefits. In Europe, the C-Roads platform aims to harmonise the deployment of C-ITS infrastructure and services across Europe. As part of the C-Roads platform, Transport Infrastructure Ireland implemented C-ITS pilot infrastructure to deliver C-ITS services to road users via ITS-G5 and cellular communications. For the deployment of ITS-G5 Roadside Units along a 60 km distance of the pilot routes in Ireland, a radio coverage planning exercise was undertaken to optimise the deployment. Several technical and practical factors were taken into consideration during this exercise. This paper outlines Transport Infrastructure Ireland’s C-ITS pilot infrastructure, discusses the key technical and practical factors taken into consideration and summarises the results achieved. As a result of this radio coverage planning exercise, we were able to achieve over 50% cost savings on infrastructure deployment for C-ITS services via ITS-G5 technology on the M50 and M1 motorways.

Urban air quality index forecasting using multivariate convolutional neural network based customized stacked long short-term memory model

A deep learning based misbehavior classification scheme for intrusion detection in cooperative intelligent transportation systems

Today, more and more vehicles on the road are being connected via roadside communication networks for a wide range of services, ranging from road safety and navigation applications to in-vehicle entertainment. Cooperative Intelligent Transport Systems (C-ITS) services are aiming to improve road safety, traffic management and environmental performance of road transportation by sharing data between vehicles and road operator’s infrastructure via roadside communication networks. This paper investigates the capacity requirements of roadside ITS-G5 and cellular communication networks for supporting the emerging C-ITS applications, based on the C-ITS pilot implemented by Transport Infrastructure Ireland (TII). The results indicate that although the current communication networks are able to support the capacity demand of C-ITS applications at a low penetration rate of C-ITS equipped vehicles during the early uptake of C-ITS, new measures are required to improve the capacity of roadside communication networks when the number of C-ITS applications or the penetration rate of C-ITS equipped vehicles on the road goes high.

In this demo, we introduce a Human-Interactive Hardware-In-the-Loop Simulation (HI2LS) framework for evaluating Cooperative Intelligent Transportation Systems (C-ITS) and road safety services. C-ITS is a complex system consisting of Wireless Access in Vehicular Environments (WAVE) communication devices, Human-Machine Interface (HMI), and computing elements to evaluate collision risk. Furthermore, C-ITS and services require the high reliability in a complicated operating environments including the human factors, and thus system testing is very important and essential process. In this demo, we propose a HI2LS framework for designing and evaluating road safety systems based on vehicle-to-vehicle/-infrastructure (V2X) communications. It enables developers to evaluate their road safety system designs in a wide range of virtual traffic environments and road conditions, which can replace dangerous and time-consuming field tests while guaranteeing the high reliability.

The single-layer perceptron, introduced by Rosenblatt in 1958, is one of the earliest and simplest neural network models. However, it is incapable of classifying linearly inseparable patterns. A new era of neural network research started in 1986, when the backpropagation (BP) algorithm was rediscovered for training the multilayer perceptron (MLP) model. An MLP with a large number of hidden nodes can function as a universal approximator. To date, the MLP model is the most fundamental and important neural network model. It is also the most investigated neural network model. Even in this AI or deep learning era, the MLP is still among the few most investigated and used neural network models. Numerous new results have been obtained in the past three decades. This survey paper gives a comprehensive and state-of-the-art introduction to the perceptron model, with emphasis on learning, generalization, model selection and fault tolerance. The role of the perceptron model in the deep learning era is also described. This paper provides a concluding survey of perceptron learning, and it covers all the major achievements in the past seven decades. It also serves a tutorial for perceptron learning.

This research compares the performance of Long Short-Term Memory (LSTM) and Multilayer Perceptron (MLP) neural networks to determine the most effective approach for accurate power loss prediction in three-phase distribution transformers. The study utilized 100,000 data sets from a 100 kVA transformer rated at 22 kV-400/230 V, collected by varying current flow from 1% to 100% under controlled temperatures from 30°C to 75°C. Only four features were employed, low voltage resistance (R<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">LV</inf>), core loss (P<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Core</inf>), low voltage current (I<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">LV</inf>), and magnetizing current (I<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Test</inf>). Both models were optimized through grid search, testing different layer configurations (3, 5, 9) and nodes per layer (32, 64, 128). Results demonstrate that LSTM significantly outperforms MLP across all evaluation metrics, with LSTM achieving lower MSE (0.001117) and RMSE (0.033427) compared to MLP's 0.004072 and 0.063812, respectively. LSTM also yielded a higher R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> value (0.986804 versus 0.951912). Although LSTM required longer training time (585.7 seconds compared to 102.22 seconds) on a system equipped with a GPU 10-core and RAM 16 GB, at low error thresholds (1%-10%), LSTM showed markedly superior accuracy. At the 1% threshold, LSTM achieved 62.13% accuracy compared to MLP's 34.22%. These findings demonstrate the advantage of recurrent architecture in capturing temporal relationships in transformer operational data, making LSTM more suitable for applications requiring high precision in predicting distribution transformer losses.

Prediction OPEC oil price utilizing long short-term memory and multi-layer perceptron models

Cooperative intelligent transport systems (C-ITS) are expected to considerably influence road safety, traffic efficiency and comfort. Nevertheless, their market penetration is still limited, on the one hand due to the high costs of installation and maintenance of the infrastructures and, on the other hand, due to the price of support automated driving functions. A breakthrough C-ITS technological solution was studied, designed, built and tested that is based on the implementation of custom low-cost on-road platforms (named “strips”) that embed micro/nano sensors, communication technologies and energy harvesting to shift intelligence from the vehicle to the road infrastructure. The strips, through a V2X and LTE communication gateway, transmit real-time, reliable and accurate information at lane level about the environmental and road condition, the traffic and the other road users’ position and speed. The exchanged information supports a series of C-ITS functions and services extending equipped vehicles capabilities and providing similar functions to non-equipped ones (including powered two wheelers). The general framework and the technological solution proposed is presented and the results of the field trials, conducted in three pilot sites around Europe, quantify the promising system performance as well as the positive effects of the C-ITS applications developed and tested on driver/rider’s behavior.

The United States in the early 20th century became the first nation to introduce the concept of intelligence in the transportation systems followed by intensive research and development by the European Countries like (Sweden, Switzerland), Japan, South Korea in the similar area. Since 1991, Intelligent Transport Systems (ITS) have been an emerging and promising area of research. This fact is also aided by the recent advancement and developments in the field of automated driving systems over the past few years. Cooperative Intelligent Transportation Systems (C-ITS) is the latest advancement over the existing Intelligent Transportation Systems (ITS) as now the systems are making use of communication and interaction with infrastructure using connected and automated vehicles (CAV’s) in order to create helpful and exciting applications that concern road safety, environment, mobility. Trust and Node Cooperation are the most crucial factors on which C-ITS relies as the decisions that are taken by a vehicle are based on the data and information it gets from the network infrastructure that is installed on the roadsides which is basically a pool of information which has been gathered from other vehicles and smart infrastructures that are part of the VANET. Presence of any misbehaviour in the network or a compromised node in the transport network system will lead to atrocious results for both traffic and network safety. Detecting any misbehaviour in the network can save us a lot of resources that will go into creating a system that can defend the complete network against a threat. Through the medium of our thesis project we will be detecting misbehaviour in the system using Deep Learning techniques. A most Basic MLP Neuron Architecture based Artificial Neural Network (ANN) would serve as our baseline and proposed models include LSTM and Transformer Model which will be used for time series classification.

Graph convolutional network – Long short term memory neural network- multi layer perceptron- Gaussian progress regression model: A new deep learning model for predicting ozone concertation

Cooperative Intelligent Transport Systems (C-ITSs) aim to improve safety, traffic efficiency, energy efficiency, and comfort. The C-ITS uses technology to allow vehicles to communicate with each other and with roadside infrastructure. C-ITS services generate and use local information shared by vehicles and infrastructures to improve transport systems, to increase energy efficiency and road safety. Development and deployment of C-ITS services for fixed and mobile components and for road users are not linked and each service provides independent information. However, these C-ITS services need interoperability to have visibility to all system and then to take the right decision and to be monitored and controlled. In this paper, we propose a mobile agent-based framework to deploy C-ITS applications in fixed components (roadside infrastructure) in order to monitor and manage C-ITS services’ interoperability in mobile components (vehicles). This framework aims to allow the linking between some C-ITS services and then to allow the management of local information collected and generated by deployed C-ITS services. An example of a safety C-ITS services scenario is illustrated to show the usefulness of the proposed framework.

This study compares the performance of Long Short-Term Memory (LSTM) and Bidirectional LSTM (Bi-LSTM) models in predicting earthquake occurrences in the Tokai region, using data from the United States Geological Survey (USGS) dataset. Given the importance of accurate earthquake prediction, particularly in high-risk regions, this research focuses on assessing the effectiveness of each model in identifying occurrence and non-occurrence events. Both models were tuned to optimize sensitivity and specificity through adjustments in sequence length, learning rate, and additional hyperparameters, with results evaluated using metrics including sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and area under the curve (AUC). Findings reveal that while both models achieved high sensitivity, the LSTM model demonstrated superior specificity and AUC, indicating a more balanced performance in distinguishing between earthquake occurrences and non-occurrences. The results show that LSTM outperforms Bi-LSTM in terms its classification metrics. LSTM achieved an accuracy of 76%, compared to 55% for Bi-LSTM. For the AUC metric, LSTM scored 66%, while Bi-LSTM scored 67%.

Long short-term memory - Fully connected (LSTM-FC) neural network for PM2.5 concentration prediction

IntroductionDrug safety has increasingly become a serious public health problem that threatens health and damages social economy. The common detection methods have the problem of high false positive rate. This study aims to introduce deep learning models into the adverse drug reaction (ADR) signal detection and compare different methods.MethodsThe data are based on adverse events collected by Center for ADR Monitoring of Guangdong. Traditional statistical methods were used for data preliminary screening. We transformed data into free text, extracted text information and made classification prediction by using the Long Short-Term Memory (LSTM) model. We compared it with the existing signal detection methods, including Logistic Regression, Random Forest, K-NearestNeighbor, and Multilayer Perceptron. The feature importance of the included variables was analyzed.ResultsA total of 2,376 ADR signals were identified between January 2018 and December 2019, comprising 448 positive signals and 1,928 negative signals. The sensitivity of the LSTM model based on free text reached 95.16%, and the F1-score was 0.9706. The sensitivity of Logistic Regression model based on feature variables was 86.83%, and the F1-score was 0.9063. The classification results of our model demonstrate superior sensitivity and F1-score compared to traditional methods. Several important variables “Reasons for taking medication”, “Serious ADR scenario 4”, “Adverse reaction analysis 5”, and “Dosage” had an important influence on the result.ConclusionThe application of deep learning models shows potential to improve the detection performance in ADR monitoring.