1. Home
  2. RIP
  3. View Record
https://trb-annual-meeting.nationalacademies.org/registration

A Machine Learning-Assisted Framework for Determination of Performance Degradation Causes and Selection of Channel Switching Strategy in Vehicular Networks

As all three major US mobile carriers have launched their own 5G networks and are working hard to expand their coverage nationwide, 5G has come into everyone’s daily life. 5G networks use millimeter-wave (mm-Wave) for higher speeds, while 4G long-term evolution (LTE) networks favor lower-band spectrum for better coverage. Vehicle-to-vehicle (V2V) communication enables wireless communication between cars and exchanges their speed, location, and acceleration information. 5G mm-Wave and 4G LTE bands are used in V2V sidelink transmissions. These two wireless channels are affected by different weather conditions, such as rain, snow, dust, and sand. Compared with 4G networks, 5G networks are designed to accommodate the increasing number of devices with higher transfer speed, lower latency, and improved security. However, our study shows that severe weather degrades the 5G performance more significantly than 4G. In this paper, we use NS-3 as a simulator to study the effect of harsh weather of dust or sand on the propagating loss of 5G mm-Wave and 4G LTE signal. We investigate their performance degradation and use a time-series machine learning technique, long short-term memory (LSTM), to predict future signal strength for 5G and 4G. Our simulation results show that LSTM performs well in forecasting signal strength, and we plan to design a system that can dynamically choose the better wireless channel in the future.