Generally, Range FFT processing is done by the HWA hardware. Based on the RF parameters, the RangeProcDDMA DPU performs operations by configuring the FFT engine while setting up EDMA channels to control data input and output. A ping-pong mode is adopted to facilitate parallel processing with raw data acquisition, with three or five sets of HWA parameters used on each side of the ping-pong buffer. To increase the maximum velocity measurement range, users can sometimes enable fast processing mode in this DPU. This mode uses the direct-current (DC) statistics and interference threshold estimates from the previous Chirp instead of calculating these parameters in real time. This mechanism saves the execution time for the two parameter sets (DC estimation/DC subtraction and interference statistics).

The flow is as follows:

1. Data input

   The sample data is first transferred from the ADC buffer to the HWA memory via the EDMA. In this example, an non-configured parameter set is used, with no data processed but only the triggering method configured (DMA or direct trigger by user software). The latter, direct trigger by user software, is primarily used for fast processing mode.

2. Pre-processing and FFT stages

   For non-fast processing modes, three parameter sets are executed sequentially. Parameter set A performs DC estimation, followed by Parameter set B, which performs DC subtraction and interference statistics estimation. Parameter set C is finally invoked for interference mitigation and FFT computation. In fast processing mode, the DC and interference statistics obtained from the previous Chirp are used to manually update the threshold values, and DC subtraction, interference mitigation, and FFT computation are performed within the same parameter set.

3. Data compression

   The compression engine is configured to the BFP or EGE mode for data compression, and the EDMA write memory (radar cube) is initiated by an HWA completion event.