The Simulation Study Of Cognitive Radar Waveform Optimization Based On Environment Perception

In the complex and changeable environment,the performance of traditional radar will be greatly reduced due to its relatively unchanged working mode and single transmitting and receiving process,which is a problem that must be solved for the further develop-ment of radar.Cognitive radar,as a new concept radar,which can solve the problems of target detection,tracking and identification in complex environments.Moreover,it can obtain information by interacting with its surroundings and the other objects and carry out adaptive processing with assistance of knowledge,thus realizing the intelligent transfor-mation of radar.However,how to collect and use environmental information to filter the appropriate receiving algorithm,transmitting waveform and working mode is a challenge the development of radar technology faced.Contemporary research on cognitive radar is usually divided into two aspects.one is to enhance the function and performance of radar,the other is to achieve spectrum sharing to avoid interference.This paper focuses on these two directions and its main contributions are as follows:(1)In this paper,the performance bottleneck of traditional architecture in today’s complex environment and the characteristics of the existing cognitive radar framework has been analyzed.According to the technical characteristics and components that cognitive radar has,radar system becomes more practical due to the architecture of cognitive radar system improved.Furthermore,the cognitive system of cognitive radar has been given and its cognitive behavior has already been analyzed.(2)A multi-input network based on environmental clutter modeling and classification method has been proposed.In this method,geographic information and meteorological information are respectively used as static information and dynamic information to represent the environment,and the clutter distribution models has been mapped to different clutter distribution models through this multi-input networks to realize clutter modeling and classification.Compared with the traditional classification method which only has one kind of information,multiple information fusion has been made used in this network,making the classification result been more consistent with the actual complex environment.(3)The detection performance of classical constant false alarm detection algorithms(mean class and ordered class algorithms)under uniform background,multi-target interference background and clutter edge background has been studied.A knowledge-assisted constant false alarm algorithm has been proposed for the scene where clutter envelope distribution is Weibull,logarithmic Weibull and K distribution.The advantages of the knowledge-assisted algorithm has been verified by comparison with the simulation using the traditional CA/GO/SO/OS algorithm in uniform clutter background and clutter edge background.(4)The study modeling the environment under the coexistence of radar signals and communication signals(or other interference signals)by using reinforcement learning has been made.We set the goal of obtaining a higher signal-to-interference and noise ratio and occupying more bandwidth of our desired signal,The tracking performance of cognitive radar using Markov Process(MDP)algorithm and Deep Reinforcement Learning(DQN)algorithm,which are used for continuous interference,intermittent interference,triangular wave frequency hopping interference and sawtooth frequency hopping interference,has been analyzed in this paper.Ultimately,the adaptation of reinforcement learning in solving such problems has been revealed by the simulation results.