Abstract
This article presents two novel neural network models based on recurrent neural network (RNN) for radio frequency power amplifiers (RF PAs): instant gated recurrent neural network (IGRNN) model and instant gated implict recurrent neural network (IGIRNN) model. In IGRNN model, two state control units are introduced to ensure the linear transmission of hidden state and solve the problem of vanishing gradients of RNN model. In contrast with conventional RNN model, IGRNN can better describe the long-term memory effect of power amplifier, more in line with the physical distortion characteristics of power amplifier. Furthermore the instantaneous gates are used to express the input information implicitly to reduce the redundancy of the input information, and a simpler IGIRNN model is proposed. The complexity analysis indicates that the proposed models have significantly lower complexity than other RNN-based variant structures. A wideband Doherty RF PA excited by 100MHz and 120MHz OFDM signals was employed to evaluate the performance. Extensive experimental results reveal that the proposed IGRNN and IGIRNN models can achieve better linearization performance compared with RNN model and traditional GMP model, and have comparable performance with lower computational complexity compared with the state-of-the-art RNN-based variant models, such as gated recurrent unit (GRU) model.
Highlights
With the arrival of the fifth-generation (5G) wireless communication system, the system capacity and communication rate are expected to increase significantly [1], [2]
In order to meet the requirements of high capacity and high rate of the system, the signal will have wider bandwidth and more complex modulation, which will lead to a higher peak to average ratio (PAPR) seriously affecting the linearity and efficiency of radio frequency power amplifiers (RF PAs) with the inherent nonlinear characteristics
Contrary to that of gated recurrent unit (GRU), the newly designed model is mainly based on the nonlinear physical characteristics of RF PAs, so that the new network structure is more in line with the characteristics of power amplifier, making it has excellent expression ability and still with lower computational complexity
Summary
With the arrival of the fifth-generation (5G) wireless communication system, the system capacity and communication rate are expected to increase significantly [1], [2]. Another vector decomposed time-delay neural network (VDTDNN) model based on MLP network is proposed [13], which is characterized by nonlinear operation only on the amplitude of the input signal, and by linear weighting to recover the phase information When it comes to the wideband communication scenarios, the memory effect and nonlinearities of RF PAs will be much more severe and complex, which leads to the relatively inferior linearization performance with MLP architecture. In order to solve the problem of gradient vanishing, a few RNN based variant models have been proposed, such as long short-term memory (LSTM) network [20], [21], gated recurrent unit (GRU) network [22], etc. A high performance and low complexity RNN-based model conforming with the nonlinear characteristics of RF PAs should be proposed
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have