Characterization and correction of data loss in a high bandwidth passive radar system

Passive radar receivers use transmitters of opportunity such as digital television (DTV) broadcast to detect targets. The next generation Manastash Ridge Radar (MRR) is designed without the use of an analog downconverter to observe transmitters up to 1.5 GHz. In addition to fast sampling, the receiver is built around a Xilinx Virtex-5 field programmable gate array (FPGA) for software-defined, flexible, and real-time, low latency processing. The FPGA channelizes data from up to four antennas and streams 8-bit IQ data through a 10 GbE link to a data recorder. Challenging the capacity of this link is extremely desirable, as it will allow the user to save a wide RF spectrum to disk for experimental processing. In the fastest use of this link, we observe up to 10 frequency-adjacent DTV stations simultaneously, however packet loss occurs. We present here characterization of this loss in real data, simulations of how this loss propagates through the processing chain and affects the final data product, suggestions for correcting this loss, and apply these strategies to real detections of aircraft on each of four antennas. The results of the simulations suggest that the radar system can absorb even 50% data loss while losing only 3 dB in detectability of targets and completely recover accurate range and Doppler velocity estimates. The detection of an aircraft with our system in the presence of 12% data loss follows the trends observed in simulation. This encouraging result shows that systems with high processing gain are incredibly robust to noise and the sacrifice of lost data in the face of observing more RF spectrum (more transmitters) is truly not a sacrifice at all.