Calibration Of IQ Imbalance And PA Nonlinearity For OFDM Systems

Wireless transceiver which is based on direct conversion has the advantages of simple structure, multiple standards and multi-band transmitting/receiving, and is becoming the trend of wireless transceiver design. However, compared with super-heterodyne architecture, the IQ imbalance in the direct conversion architecture is more serious. IQ imbalance introduces image interference and reduces the performance of the system. Power amplifier is the key component in wireless transmitter. In order to increase the power efficiency of the power amplifier, we often make it work in the nonlinear area which is close to the saturation point and the power amplifier presents strong nonlinearity. The nonlinearity of the power amplifier introduces in-band distortion and out-band distortion at the same time. The in-band distortion distorts the magnitude and phase of the signal and limits the ability to resist symbol error of the system. The out-band distortion introduces the adjacent channel interference and decreases the system’s spectral efficiency.OFDM technology, which has the advantages of strong ability to resist multipath fading and high spectrum efficiency, is one of the key technologies in wireless communication systems. However, the OFDM system is very sensitive to IQ imbalance which introduces image interference and destroys the orthogonality of the sub-carriers of the OFDM system. This leads to a drop in the performance of the system. In addition, OFDM system requires the power amplifier to work linearly because its high PAPR make it more sensitive to nonlinear distortion. Therefore, it is of great importance to do IQ imbalance and power amplifier nonlinearity calibration in the OFDM system.This paper studies IQ imbalance and power amplifier nonlinearity in two parts. We first study the calibration algorithm for the OFDM system where only IQ imbalance is considered, then come up with another algorithm for the system where IQ imbalance and power amplifier nonlinearity exist in the transmitter at the same time.In this paper, we propose an algorithm to estimate and calibrate the transmitter and receiver IQ imbalances and channel in the OFDM system. When considering the frequency independent and frequency dependent IQ imbalances, which are introduced by both transmitter and receiver, most of the existing studies combine the channel, transmitter IQ imbalance and receiver IQ imbalance together. However, in this paper, a new preamble structure is proposed to estimate the parameters based on the time domain and frequency domain models of IQ imbalance. Based on the new preamble structure and the condition that the channel changes faster than the IQ imbalance does, a new parameter estimation algorithm is proposed to simultaneously and respectively estimate the three parameters, and iteration is also used to improve the estimation performance. In the end, we give two calibration methods to correct the IQ imbalance. With the separated parameters, pre-distortion and post-correction can be carried out in the transmitter and receiver, respectively. And simple symbol-by-symbol detection can be used to detect data symbols without significant performance loss. At the same time, with the separated ones, the combined effect can be calculated easily.For the OFDM system where both IQ imbalance and power amplifier nonlinearity exist, we come up with two joint calibration methods. The IQ imbalance which is introduced by the receiver is not considered in this part because it can be calibrated by the filtering operation or adopting the digital demodulation which doesn’t bring in the problem of IQ imbalance. So we only consider IQ imbalance and power amplifier nonlinearity of the transmitter in this part. In the designs, we treat the transmitter as a whole and only process the baseband input and power amplifier output. Without the need of intermediate signals, the system has the benefit of high integration. We propose two methods, one is the two-step design, where the IQ imbalance and power amplifier nonlinearity are pre-distorted in two-steps with a unified training structure to get the pre-distortion parameters. The other is the one-step design, where we first derive the model that can jointly characterize IQ imbalance and power amplifier nonlinearity, then one block is used to pre-distort the both. In the designs, both frequency independent and frequency dependent IQ imbalances are considered. We also show that decreasing the number of coefficients properly can decrease the computational complexity without significant loss in performance.