The Research Of Human Dynamics In Vehicular Networks

Every day, vehicles transit in a city and along their trajectories, they encounter other vehicles. The frequency of these encounters is influenced by many factors, such as: vehicle type, speed, destinations, traffic conditions, weather conditions, and the period of the day. However, these factors are justified by the public roads limits and the driver's behavior. The drivers present daily routines and similar behaviors that have a great impact in the daily traffic evolution. We can find that the drivers play a decisive role in the driving process from a social analysis of vehicular network generated by the vehicles. That is, for the promotion of the human dynamics can reflect from the vehicular network properties.Vehicle ad hoc network(VANET) is a special type of Ad hoc network, formed by vehicles with processing and wireless communication with each other, or road infrastructures. The goal of VANETs is to construct a self-organizing, easy to deploy, low cost, open vehicle communication network. It has no center, selforganization, support multi hop data transmission capability, to achieve the accident early warning, auxiliary driving, road traffic information inquiry, vehicle inter communication, and Internet access services. VANETs has the characteristics of dynamic topology, strict time delay, nodes move quickly, predictable trajectory, unlimited energy, and accurate location and so on. In addition, the application of VANETs prospects are bright and broad, the research range across the field of intelligent transportation systems, computer networks and wireless communications three traditional research areas. These all make the research of VANETs attracted many attention from academic and industrial.The traditional vehicle mobility is restricted to the driver habits and the line-of-sight. The new-emerging connected vehicles enable information exchange with each other at vicinity, which will undoubtedly bring a greatly positive effect on the vehicle mobility pattern. In this paper, we propose to model and analyze the vehicle mobility patterns respectively driven by traditional drivers and connected drivers. We perform the extensive simulations to explore the statistical differences between two mobility patterns and the underlying reasons through comparing to the real vehicle trace datasets.Vehicular network has short connection time, frequently changed topology and other unique properties compared with traditional mobile networks, conventional communication technology does not transplant well. In this paper, we present two human dynamics based algorithms of network optimize: the first one is driver social relation based clustering, which is in view of vehicles' social relationship and combined with the instantaneous position and speed of the vehicle node; the other one is a novel cross-layer optimization method based on Partially Observed Markov Games(POMG), which is to improve optimization decision against the inaccurate observed context caused by high-speed movement, sensor errors, and other unavoidable reasons. POMG extends Markov Decision Process(MDP) and Partially Observed Markov Decision Process(POMDP) to dynamically adjust the concerned actions(e.g. transmission range, contention window, and bit rate) according to the observed traffic density, and thus can attain optimization in several dimensions, e.g. throughput, channel utilization, delay, and total number of neighbor nodes.It is undoubtedly envisioned that autonomous cars will massively appear to move on road in the near future. Autonomous cars are capable of perceiving the surrounding environment and navigating without driver's input. Various autonomous cars are envisioned to move on road in the near future, in which the driver model can not only guarantee driving safety against collisions, but also customize driver's own preferences. The action decisions(e.g. acceleration and overtake) during the driving process are heavily dependent on the current human dynamics(e.g. age, gender, and emotion). For examples, male drivers generally drive faster than female drivers, and young drivers tend to overtake frequently compared to the old ones. Therefore, it is essential to introduce the human dynamics into the driver model for autonomous cars. In this paper, we present a new driver model based on human-behavior dynamics for autonomous cars, which allows driverless cars to move appropriately in accordance to the behavioral features of driver owners. This model is established through analyzing drivers' various properties, e.g. gender, age, driving experience, personality and emotion. These attributes collectively determine all the actions occurred during the driving process.