View article

Received signal strength (RSS) is a major performance metric in vehicular communication system design. Experimental measurements of the RSS for vehicular communications are not cost effective. Therefore, offthe-shelf ray-tracing packages are deployed to substitute the costly measurements with RSS simulations. However, the simulation process is too time-consuming if the RSS is required over a long interval of time. We propose a new RSS estimation approach using neural network (NN) to reduce the computation time. First, ray-tracing is used for simulating the RSS in some instances of time and training the NN. Then, the NN is deployed to estimate the RSS afterward. We apply the new approach to an antenna placement problem in vehicle-to-vehicle (V2V) communications. The numerical results show that RSS computation time reduces significantly by using the proposed estimation approach, and the approach is as effective as ray-tracing in the RSS simulation for the antenna placement problem in vehicular communications.