Abstract:
Autonomous vehicles are becoming prevalent day-by-day, and with the advent of such technology comes its various safety risks and challenges. The primary objective is to improve road safety and mobility for all users. The key enabling technology that has shown viable promise is millimeter wave communication, but has its limitations: Communication links are line-of-sight and directional. They experience very high propagation loss due to atmospheric absorption. A possible solution would be to use a Joint radar-communication transceiver where radar waveforms are embedded within the communication frame. Utilizing advanced machine learning framework to characterize the features of the JRC, such as range, doppler, angle of arrival of targets from corresponding radar echoes. The positives of such an approach is shared hardware, common waveform generation, no channel estimation necessary. The advantages we would receive are no additional cost for hardware, no requirement for separate radar spectrum leading to no interference, lower latency as channel estimation is not done. A hardware implementation for multiple target detection is discussed, the details of the implementation of the algorithms such as Matched-filtering, MUSIC and CLEAN are presented. A complete end-to-end system for multiple target detection using a Python based GUI is presented. Analysis and design details are thoroughly discussed, using RMSE plots for various fixed-point word-lengths along with Hardware Software Co-Design(HSCD). Keywords:
