Recently, the official account mentioned the technologies that engineers may need to pay attention to such as FPGA, ASIC, and signal processing algorithms in the fields of mobile communications and satellite communications, including LMS algorithms, RLS algorithms, LCMV algorithms, and SAR image processing.
1 LMS Algorithm:
The LMS (Least Mean Square) algorithm can be considered to be the most basic and useful algorithm in machine learning. The implementation on the DSP is mainly in C.
The LMS algorithm is a search algorithm that assumes w to iterate from a given initial value, gradually changing J(W) in the direction of the minimum until a value is reached that makes J(w) converge.
2 RLS Algorithm:
RLS (Recursive Least Square) algorithm is a fast algorithm of least squares algorithm, which is the principle of adaptive filter and can be implemented with FPGA. There is a broad consensus that RLS algorithms are superior to LMS algorithms in terms of convergence speed and tracking performance, at the cost of more complex calculations.
Advantages: RLS adaptive filters provide faster convergence and tracking performance.
Disadvantages: Since RLS uses the recursion of the inverse matrix of the autocorrelation matrix, the convergence speed and tracking performance of RLS will deteriorate seriously once the autocorrelation matrix of the input signal is close to singularity.
3 LCMV Algorithm:
LCMV (Linearly constrained minimum variance) minimizes the variance of the array output under certain constraints when the desired signal form and signal direction are completely unknown. In order to eliminate the null trap in the expected signal of the array pattern, multiple linear constraints are adopted to force the reception of the expected signal.
There is a 2020 patent, an FPGA-based anti-jamming method and process for UAV GPS antennas
Document No.: 23099863 Release Date: 2020-11-27, used.
4 SAR Image Processing:
SAR是合成孔径雷达（synthetic aperture radar）的简称，。
。
Therefore, there are technologies such as SAR noise distribution characteristics, coherent spot noise suppression, image edge extraction, image segmentation and classification, and image target recognition.
The potential of automotive SAR is huge, but to reduce the cost of calculation.
A few years ago, there was a paper entitled ", which was used to solve the Doppler phase fixed-point, and the results were compared with the MATLAB double-precision theoretical data to verify the accuracy effectiveness.