Summary The arrival of cloud computing technology promises innovative solutions to the problems inherent in existing vehicular ad hoc network (VANET) networks. Because of the highly dynamic nature of these networks in crowded conditions, some network performance improvements are needed to anticipate and disseminate reliable traffic information. Although several approaches have been proposed for the dissemination of data in the vehicular clouds, these approaches rely on the dissemination of data from conventional clouds to vehicles, or vice versa. However, anticipating and delivering data, in a proactive way, based on query message or an event driven has not been defined so far by these approaches. Therefore, in this paper, a VANET‐Cloud layer is proposed for traffic management and network performance improvements during congested conditions. For the traffic management, the proposed layer integrates the benefits of the connected sensor network (CSN) to collect traffic data and the cloud infrastructure to provide on‐demand and automatic cloud services. In this work, traffic services use a data exchange mechanism to propagate the predicted data using a fuzzy aggregation technique. In the evaluation phase, simulation results demonstrate the effectiveness of the proposed VANET‐Cloud layer to dramatically improve traffic safety and network performance as compared with recent works.

despite the growing trend in intelligent transportation systems applications. Besides, there still many problems waiting for an accurate solution such as traffic flow forecasting. In this paper, based on real-time data provided by dual loop speed traps detectors at given slot of time; we propose a cloud data collection method aimed to improve prediction accuracy. To reach this accuracy, two traffic parameters was introduced: average speed and foreseen arrival time between two vehicles. By adopting Choquet integral operator, these parameters can subsequently aggregated to busiest parameters. Afterwards, a simple linear regression is applied for a dual purpose: forecasting and proving that there is a relationship between derived busiest arrival time and the traffic flow (q). Moreover, simulation flowcharts results illustrate that the forecasts by the Choquet operator ensure an accurate results to the real-time data. In contrast, weighted average operator results weak accuracy forecast compared to the real-time data.

In urban areas, the cost of road congestion has paid great attention to the sociological, technological and environmental aspects, such as the optimal route and fuel consumption. This step is towards a smarter vehicle mobility where the travel time will be planned and dynamically adapted to changes with actual status of the traffic flow. In this article a multi-objective ACO algorithm is proposed to solve the daily carpooling problem. In particular, a set of decision variables are proposed in order to minimize three objective functions subject to a set of constraints on these objectives.

The concept of data dissemination has become an inherent problem with Vehicular Ad hoc NETwork (VANET), especially when traffic jams occur. The main idea of data dissemination is how to handle a huge amount of traffic data collected in the vicinity. Therefore, different strategies have been proposed by the research community in this field to solve this problem, including routing protocols, virtual machines, data aggregation, clustering and so on. In this paper, a virtual cluster strategy is proposed to select candidate vehicles acting as getaways in the network. In particular, the formation of the virtual clusters is based on dominant set algorithm to create virtual network for data dissemination. In the evaluation phase, the simulation results demonstrate the effectiveness of the proposed clustering strategy to significantly improve network performances.

This article describes how anticipating unforeseen road events reveal a serious problem in intelligent transportation systems. Due to the diversity of causes, road incidents do not require regular traffic conditions and accurate prediction of these incidents in real-time becomes a complicated task not defined so far. In this article, a smart traffic management system based cloud-assisted service is proposed to preserve the traffic safety by controlling the road segments and predicts the probability of incoming incidents. The proposed cloud-assisted service includes a predictive model based on logistic regression to predict the occurrence of unforeseen incidents. The sudden slowdown of vehicles speeds is the practical case of the article. The classification task of the predictive model incorporates four explained variables, including vehicle speed, the travel time and estimated delay time. The prediction accuracy is proved by checking the model relevance according to the quality of fit and the statistical significance of each explained variable.