Characterization of Fading in Vehicle-to-Vehicle Visible Light Communication Channels

In the years to come, communication between vehicles is the future trend. With Light Emitting Diode (LED) becoming very common in today''s automotive lights, Vehicular Visible Light Communications (V2LC) will be a promising solution to the intelligent transportation system. Moreover, LED has many advantages, such as long life expectancy, low power consumption, and minimal heat generation lighting. Modulating these existing LED lights to transmit optical signals, and utilizing either a photodiode module or a camera as the receiving component to demodulate the transmitted optical signal, such a typical V2LC system is a intelligent transportation system and requires minimum additional cost. While many existing studies have investigated V2V RF channels, there is very little work characterizing V2V VLC channels in driving scenarios. The path loss of a V2LC channel is dictated by the transmitter-receiver distance, the irradiance angle, and the incidence angle. In this paper, (1) we collected extremely accurate vehicle mobility traces using a 2-dimensional LIDAR mounted on the front bumper of the rear vehicle. From these traces, we can get above parameters of the front vehicle (the receiving vehicle) with respect to the rear vehicle (the transmitting vehicle). (2) we collected optical signal by photodiode mounted on the front vehicle and used OEM LED headlamps as the transmitter mounted on the rear vehicle. Combining the above two sets of data, we found (1) there are two localized factors causing significant variation of the received power. One is vehicle vibration, providing variation up to 5 dBm; and the other is turning at intersections, the variation of which can up to 10 dBm. (2) for both 90\% and 50\% coherence time, the freeway has larger coherence time than urban area in V2LC channels and V2LC has mush slower time variation; the values of the V2LC channel are at least an order of magnitude larger than that of the DSRC channel. (3) the path loss exponent is contributed by the distance rather than the area and the empirical path loss exponent increases at shorter distance, but converges to 4 at larger distance.