Research On Calibration Method For Amplitude And Phase Errors Of Broadband Array

Adaptive array antenna is an important branch of array antenna, which is widely used in radar, sonar, communication, radio astronomy and other national economic and military applications. Wideband adaptive array antenna has good performance on the condition that array antenna has small errors. However, the array channel amplitude and phase frequency characteristics are not consistent with each other in practical application, which is caused by channel amplitude and phase errors, namely channel mismatch. Channel mismatch will affect the performance of wideband array systems, which can’t work normally when channel mismatch reaches to a certain level. Therefore, channel mismatch must be calibrated before beam- forming.This thesis focuses on the study o f adaptive channel equalization to overcome the channel mismatch. The dominant research contents are as follow:① The signal model of wideband array is illustrated firstly. Then we introduce the methodology of broadband adaptive beam- forming based on space-time two-dimensional structure and the direct sampling matrix inversion algorithm. Finally simulations and analysis are carried out for wideband adaptive beam- forming with the algorithm discussed above.② The performance weakness of wideband adaptive array caused by channel mismatch is investigated. Two channel mismatch models are derived at the same time. Then, computer simulations are illustrated through different array parameters, such as beam pattern and array output SINR.③ The research discusses the adaptive channel equalization technology. The thesis elaborates the basic principle of time domain channel equalization and frequency domain channel equalization respectively. Two kinds of adaptive equalization algorithms in time domain and one adaptive equalization algorithm in frequency domain are investigated deeply. For time domain algorithm, adaptive equalization filter employs least mean square algorithm and recursive least squares algorithm. On the other hand, the least squares fitting strategy is utilized in frequency domain equalization. The above three algorithms are verified and analyzed by computer simulations. Finally, the relationship between some factors and channel equalization is analyzed by simulations at the same time.④ If the SNR of the reference signal is very small, the performance of equalization will also be very poor since the noise takes the prime role in the least squares fitting algorithm in frequency domain. A new approach is proposed to reduce the effect of noise on equalization. The new algorithm only focuses attention on the frequency range of the frequency where the reference signal exits and ignores the other frequency region in the first N yquist domain. However, the proposed algorithm will enlarge the amplitude response outside the signal frequency region. We use a band-pass filter to remedy the problem which is located in front of the equalization. Finally, the simulation results prove the feasibility of modified algorithm.