Research On Digital Predistortion Of Radar-communication Integration System

As an important electronic equipment in the fields of aeronautics,astronautics and military, the traditional method that deal with information after radar reconnaissance with low speed in real time can not meet the requirements any more. With the development and progress of technology, the core technology of radar-communication integration system for the requirements of multiple terminal, miniaturization and multiple function is more and more urgent.The PA of radar has nonlinear effect on digital modulating signal, which is the main obstacle of high speed communication in radar-communication integrated system. To solve this problem, this article mainly research the nonlinear characteristic of radar power amplifier, the distortion of digital modulating signal which PA bring in and different digital predistortion algorithm base on polynomial and look up table. The radar power amplifier usually work in saturation region deeply to get high transmit power, which bring distortion in signal so serious that the error bit rate rise in receiver which even can not be demodulated. Different PA model and radar PA characteristic are simulated in the article, which proves the radar PA has obvious effect of memory. Memoryless nonlinear system model has bigger error to describe radar PA characteristic than memory polynomial model. And then the article simulate the nonlinear effect of PA model have on QPSK, pi/QPSK and 16 QAM modulating signal. The simulation result show clearly that 16 QAM signal is the most sensitive signal to nonlinear effect.The premise of radar-communication integrated system is adding communication signal processor on radar equipment without influence radar working. Therefore, this article mainly research the predistortion algorithm include memoryless polynomial, Hammerstein model, memory polynomial and look up table model. Analyzing the adaptive algorithms of predistortion performance and make improvement on the base of LMS to make it easy to be realized by hardware. Simulating the performance of different algorithm to correct 16 QAM signal through constellation diagram, eye diagram and error bit rate, which proves memoryless predistortion and Hammersteinmodel has limit effect on memory PA with nonlinear memory effect. Memory polynomial model compensate the PA efficiently at a cost of high order which means difficult to realize. Joint polynomial and look up table through store the coefficient of memory polynomial have the both advantage of two predistortions, which is simulated has ideal linearization performance in constellation diagram and eye diagram for radar PA, the symbol error rate is close to the linear PA in the same time.