The technological advances in the transportation and automotive industry led to the use of new types of sensing systems more cost-effective and adapted to large-scale dense deployment. Those sensing techniques allow continuously gathering traffic measurements times series in different geospatial locations. The accuracy of the obtained raw measurements is often hindered by different factors related to the sensing environment and the sensing process itself and thus fail to capture the short-term traffic variations crucial for real-time traffic monitoring. In this paper, we propose the DGC-LSTM model for area-wide traffic estimation from error-prone measurements time series. The backbone of the DGC-LSTM model is a graph convolutional Long Short Term Memory model with a dynamic adjacency matrix. The adjacency matrix is learned and optimized during the model training. The adjacency matrix values are estimated from the set of contextual features that impact the dynamicity of the dependencies in both the spatial and temporal dimensions. Experiments on a realistic synthetic labelled Bluetooth counts dataset is used for model evaluation. Lastly, we highlight the importance of transfer learning methods to improve the model applicability by ensuring model adaptation to the new deployment site while avoiding the extensive data-labelling effort.

The technological advances in the transportation and automotive industry led to the use of new types of sensing systems more cost-effective and adapted to large-scale dense deployment. Those sensing techniques allow continuously gathering traffic measurements times series in different geospatial locations. The accuracy of the obtained raw measurements is often hindered by different factors related to the sensing environment and the sensing process itself and thus fail to capture the short-term traffic variations crucial for real-time traffic monitoring. In this paper, we propose the DGC-LSTM model for area-wide traffic estimation from error-prone measurements time series. The backbone of the DGC-LSTM model is a graph convolutional Long Short Term Memory model with a dynamic adjacency matrix. The adjacency matrix is learned and optimized during the model training. The adjacency matrix values are estimated from the set of contextual features that impact the dynamicity of the dependencies in both the spatial and temporal dimensions. Experiments on a realistic synthetic labelled Bluetooth counts dataset is used for model evaluation. Lastly, we highlight the importance of transfer learning methods to improve the model applicability by ensuring model adaptation to the new deployment site while avoiding the extensive data-labelling effort.

This paper presents an analytical model for the average Packet Error Rate (PER) of Bluetooth in Basic Rate (BR) mode. The effects of the physical layer and the MAC layer are taken into account. A Nakagami-m block fading is considered to take into account the effects of wireless channel fading. A Gaussian Frequency Shift Keying (GFSK) modulation scheme is used. Also, Forward Error Control (FEC) coding effects are taken into account. The interference between the different piconets is captured by a simple Medium Access Control (MAC) layer collision model. An approximation of instantaneous PER is obtained as a first result. Then, a closed-form expression is derived for the PER at the physical layer of Bluetooth. Finally, the overall average PER at the physical and MAC layer of Bluetooth is computed. Extensive simulations are performed to show the accuracy of the obtained results.

Due to the increasing number of car-centered connected services, making efficient use of limited radio resources is critical in vehicular communications. Hybrid vehicular networks dispose of multiple Radio Access Technologies (RATs) like cellular and vehicle-to-vehicle (V2V) networks, with complementary characteristics that allow for developing smarter network traffic distribution methods. This paper proposes a self-adaptive clustering system for ensuring a suitable trade-off between data aggregation (over the cellular network) and communication congestion due to cluster management (within the V2V network). The system's algorithms use a distributive justice approach for selecting cluster heads, to improve fairness among car drivers and hence help the social acceptability of self-adaptive clustering. Simulation results show that this approach significantly improves fairness over time without affecting network performance. This solution can thus optimize the usage of radio resources, reducing cellular access costs, without the need for uniformization among different mobile operators' access plans.

We propose a simulation model of Bluetooth (BT) passive scanning for vehicular traffic monitoring. The model is designed to simulate packet detection in the physical and Medium Access Control (MAC) layers for fixed BT sensors. At the physical layer, we consider the radio propagation effects of path loss, small- and large-scale fading. At the MAC layer, we model channel hopping and packet collisions. Associated to a vehicular traffic simulator, our proposal provides a cost-efficient way to generate large-scale BT traffic measurement datasets including both sensor and ground-truth data, avoiding deployment costs. We validate the model by comparing the simulation output to experimental data and we show a simulation use-case of vehicular traffic monitoring in a complex urban context.

Reinforcement learning is a commonly used technique in the field of traffic signal control. By iteratively testing actions given the current network state, a traffic signal agent gradually learns to optimally control traffic lights given the traffic situation at hand. While they usually outperform traditional control systems in the literature, these methods have to first go through an exploration phase where they test different actions in a trial-and-error fashion. The nature of reinforcement learning methods hence causes unstable performances and high computational costs. In order to limit the costs linked to this exploration phase, we propose a bootstrapping method that reduces the computation time needed for learning-based traffic control methods to reach acceptable performance levels. Our method models lanes of the road network as queues, and derives results on the average service time of vehicles at the intersection level in order to estimate the agent's policy. The performance of our bootstrapping method is then compared to a more traditional Q-Learning method using the SUMO simulator. Simulation results show that bootstrapping alleviates both of these issues by immediately reaching acceptable performance levels by quickly training the agent without any direct interaction with the simulation environment.

During the last decade, Bluetooth has become a widespread feature in the automobile industry, meaning that its signal activity can be correlated with road traffic. Using this technology as a means for assessing traffic is very cost-effective and has a low impact on the infrastructure. Nevertheless, unlike other techniques, it does not provide direct sensing of vehicles only. Statistical analysis methods need to be applied to the collected data. In this paper, we first propose a machine learning method for traffic flow estimation using Bluetooth sensors. We also propose a method for estimating the mean travel speed between two sensors. The performance of the proposed methods is evaluated through eight weeks of experimentation. Finally, we envision a potential solution for building real-time Origin- Destination matrices.