ALTERA
Direction of Arrival Estimation with Neural Networks for Altera Agilex™ 5 E-Series
System Description:
Direction of Arrival (DOA) Estimation implemented with Altera DSPBuilder for MATLAB Simulink on the Altera Agilex™ 5 FPGA.
Engineering Need / Challenge:
Direction of Arrival (DOA) Estimation is a fundamental function in many phased array systems. It can be challenging to implement efficiently in a highly parallelized manner.
Engineering Approach:
DOA calculation is performed using a Neural Network-based approach implemented with DSPBuilder for MATLAB Simulink on the FPGA.
A Neural Network (NN) is trained to estimate DOA directly from the covariance matrix. Signal detection and estimation of the number of signals are also performed using separate NNs implemented on the FPGA.
Demo Description:
In this example, the DOA calculation is performed using a Neural Network-based approach implemented with Altera DSPBuilder for MATLAB Simulink on the FPGA.
The implemented system consists of an AD9081, a 4-channel ADC/DAC front-end connected to the Altera Agilex™ 5 FPGA via the JESD204C protocol.
Key Technologies:
Altera Agilex™ 5 FPGA | ADI AD9081 ADC/DAC |
JESD204C | Altera DSPBuilder for MATLAB Simulink |
Neural Networks
Engineering Highlight:
Agilex™ 5 FPGA system-level capabilities with high-end interfaces such as JESD204C, combined with AI Neural Networks implemented using the DSPBuilder tool for MATLAB Simulink.