2018年6月24日から28日のVEHICULAR 2018において，東 峻太朗（M2），岩月 海人（M1），木村 健太（M1），西牧 佑哉（M1）の4名が以下のタイトルで発表を行いました．
Improvement of False Positives in Misbehavior Detection（東 峻太朗）
In recent years, research on autonomous driving and vehicle-to-vehicle (V2V) communication have been conducted in the Intelligent Transport Systems (ITS) field. In addition, vehicles have vehicle-to-cloud (V2C) communication with cloud servers using mobile lines. When vehicles are connected to various targets, malicious acts have enormous impact. This paper represents further work on our previous publication ”A Method of Detecting Camouflage Data with Mutual Position Monitoring”. In our previous research, we proposed how to detect malicious vehicles which sent masqueraded data of their positions. We evaluated the detection rates and received good results. We found that we could detect completely malicious vehicles by increasing the threshold value of our detecting method. However, we have some problems. We especially considered the false positives problem in our previous research. We thought that vehicle densities affect false positives, so we calculated them in high vehicle densities. We cloud find high vehicle densities help suppress false positives, but this countermeasure is effective in only this situation. We should address the false positives problem in low vehicle densities. In this paper, we will reveal our research’s target at first. Next, we will describe the operation of proposed method. Then, we will describe improvements of previous research, which are methods of weighting for each vehicle and dynamic determination, and then we will describe the evaluation of these methods.
Evaluation of WiFi Access Point Switching for Vehicular Communication Using SDN（岩月 海人）
In recent years, communication is often done through many mobile terminals including smartphones. Such cellular networks as LTE are mainly used for these mobile units. Since cellular networks are becoming congested by an increase in the number of mobile terminals, such applications require more data. To reduce the congestion of cellular networks, each carrier is offloading its cellular network traffic to unlicensed bands, such as WiFi systems. In fact, according to a Cisco survey, 58% of vehicular communication is offloaded traffic, which is expected to continue to rise in the future. Therefore, the importance of WiFi systems for vehicular communication will also increase in the future. We also expect to utilize WiFi systems in communication even in such fields as automobiles. However, as mentioned above, when a WiFi system is used for vehicular communication, a new problem arises: the time during which communication cannot be performed becomes longer than with a cellular system when the Access Point (AP) of the WiFi system is switched for a connection with a mobile terminal. Compared to cellular systems, existing WiFi system have a smaller coverage area that includes just one AP, and the APs belonging to a plurality of different networks cooperatively lack a function for assisting the AP ’s terminal switch, for example. In other words, vehicular communication using WiFi has a higher disconnection frequency and a longer disconnection time than cellular systems. When vehicular communication is done using a WiFi system with a narrow AP coverage area, if such connection procedures as authentication take too much time, the mobile terminal leaves the coverage area before the authentication is completed. Therefore, in this research, we shortened the disconnection time of communication during AP switching. Among APswitching operations when connecting to the AP, selecting the destination of the AP being switched and the authentication operation occupy most of the entire switching operation. In this research, we use the SDN (Software Defined Network) concept on WiFi networks to shorten these two disconnection times to solve the problem of using WiFi systems in vehicular communication.
Evaluation of Safety and Efficiency Simulation of Cooperative Automated Driving（木村 健太）
The research and development of automated driving have increased in recent years. A camera, laser radar, and milliwave radar are mounted on an autonomous automated vehicle for collecting peripheral information. Then the vehicle’s operation is controlled using the surrounding environment information. However, such in-vehicle sensors have drawbacks because detection is impossible outside the range of the viewing angles, and so avoiding collisions is difficult at intersections that suffer from poor visibility.
With Vehicle-to-Vehicle (V2V) communication, blind spot information can be acquired that the vehicle cannot see directly. To operate safely using this information, research on cooperative automated driving is being conducted. The recognition rate near 300 m increases by sharing the host vehicle ’s information and the sensor information using V2V communication instead of automated driving that relies solely on sensor information.
The level of automated driving techniques has already been defined by Society of Automotive Engineers (SAE) international. Level 2 vehicles, in which the automated vehicle partially controls the vehicle ’s operation, are beginning to appear on the market. For example, when a vehicle predicts an accident, it automatically brakes. This is not completely automated driving; it just illustrates the scope of driving support. In this research, we evaluate automated vehicles that can operate such vehicles whose popularity is expected to increase in the future. In this research, we determine the safety criterion for passing through intersections when using communication and compare cases with and without sharing the surrounding information of a vehicle. Based on our results, we evaluate the influence of shared communication on traffic efficiency and safety when passing through an intersection that suffers from poor visibility.
Evaluation of a Method for Improving Pedestrian Positioning Accuracy using Vehicle RSSI（西牧 佑哉）
As the penetration rate of smartphones and tablet-type devices continues to increase, various services using their location information are being used. For example, we can check our current location on a map, look up a route from it to a destination, and get surrounding shop information and coupons. However, if the positioning error is too large, we might get lost or fail to get the information we want. In recent years, Intelligent Transportation Systems (ITS) have also been investigated that improve traffic safety, efficiency and driving comfort. For example, through vehicle-to-vehicle and vehicle-to-pedestrian communication, related work makes efforts to prevent vehicles from colliding with vehicles and pedestrians through exchange of information such as position and speed. Since the large positioning error might lead to accidents, we need accurate position information of every involved pedestrians and vehicle to reduce traffic accidents and maintain safety. Among positioning systems that acquire position information which is important in such services and systems, Global Positioning System (GPS) is most frequently used. Its positioning accuracy ranges from several meters to several tens of meters. But in urban areas that are littered with high-rise buildings, GPS signals are blocked by buildings and influenced by multipaths, further increasing the positioning error. Since GPS accuracy is affected by the surrounding environment, achieving stable positioning is difficult.