Digital predistortion of power amplifiers is a concept of linearization. with the aim to improve performance and efficiency.
In the digital predistortion technique, a predistorter precedes the power amplifier, and the cascaded configuration of both exhibits linearity and constant signal gain. By including a predistorter, the power amplifier can be operated up to the saturation region, without compromising the linear characteristics.
Predistorters are modeled with and without memory. Memoryless predistortion uses the look-up table algorithm and the memoryless polynomial algorithm for modeling. The Wiener, Hammerstein, Wiener-Hammerstein, and memory polynomial algorithms are some models used for modeling predistortion with memory.
Wireless communication system engineers don’t want to have to choose between system efficiency and the linear operation of power amplifiers.
Choosing between efficiency or the linear operation of a power amplifier is a major issue for communication design engineers. The linear operation range of power amplifiers is not viable when the transmission formats have high peak-to-average power ratios (PAPR). It is sometimes necessary to sacrifice efficiency, reliability, operating expenses, and performance of the power amplifier when operating it at lower power (linear operating region).
Practically speaking, the efficiency of power amplifiers cannot be compromised, so the resulting nonlinearities must be handled effectively. If efficiency is required, then one must pull the power amplifier operation to the saturation range full of non-linearities. The non-linearities create signal degradation, in-band signal distortions, and out-of-band spectral regrowth. The most common non-linear problems in wireless communication systems are gain compression, phase distortion, intermodulation distortion, harmonic distortion, and adjacent channel interference.
Digital predistortion of power amplifiers is a concept of linearization, with the aim to improve its performance and efficiency. It is a remarkable technique, utilizing the technology advancements in analog to digital converters and digital signal processors.
The Concept of Digital Predistortion of Power Amplifiers
The digital predistortion of power amplifiers is a technique where the non-linear distortions of power amplifiers are reduced digitally. Improvements in linearity enhance power efficiency as well as the performance of power amplifiers.
In the digital predistortion technique, a predistorter precedes the power amplifier, and the cascaded configuration of both exhibits linearity and constant signal gain. By including a predistorter, the power amplifier can be operated up to the saturation region without compromising the linear characteristics. The digital predistorter expands the non-linearity in the baseband and compensates for the compressing characteristics of the power amplifier.
The communication baseband signal u(n) is pre-distorted to maintain the overall performance and efficiency of the power amplifier. The pre-distorted signal x(n) is given as input to the power amplifier. Generally, the transfer function of digital pre-distortion is designed so that it is the inverse transfer function of the power amplifier. In that case, the output signal y(n) is equal to u(n), without any non-linearities.
The challenging part of digital predistortion is in finding a model with the inverse transfer function of the power amplifier. The predistorters should also appropriately address any changes in the power amplifier characteristics due to temperature or component aging.
There is a need for predistorters with memory structures in wireless communication systems. Initially, predistorters were designed as memoryless structures, taking into account the power amplifier output nonlinearity due to its current input. However, this concept is not true for wideband power amplifiers. If we consider the example of a high power amplifier in base transmitting stations, the power amplifier output characteristics are causal in nature. They are dependent on the past and present values of the input. The non-linearity with memory is caused by the frequency-dependent behavior of the active devices and components in the biasing circuit of power amplifiers. These non-linear power amplifiers with memory effects cannot be linearized using predistorters without memory.
Implementation of Digital Predistortion
It is important to consider memory while designing digital predistortion of power amplifiers. Here, we will discuss memoryless predistorters as well as predistorters with memory structures.
Memoryless Digital Predistortion
Power amplifiers showing non-linear characteristics without memory effects only require memoryless pre-distortion. In memoryless power amplifiers, the output signal amplitude and phase deviation depend on the current input. Commonly used algorithms in memoryless predistortion models are the look-up table (LUT) algorithm and the memoryless polynomial algorithm. Memoryless predistortion can be classified as data predistortion and signal predistortion.
According to the location of the pulse shaping filter, data predistortion is bifurcated. In the first type of data predistortion, the pulse shaping filter succeeds the power amplifier and LUT can be utilized for mapping the input amplitudes to the desired locations. Usually, RF bandpass filters are used as pulse shaping filters, which make this scheme of data predistortion less significant. In the second type of data predistortion, the pulse shaping filter is placed next to the digital pre-distortion before the power amplifier.
The disadvantages of data predistortion—including the need for pulse shaping filters and the non-effectiveness in transmitting formats such as orthogonal frequency-division multiplexing (OFDM ) and Wideband Code Division Multiple Access (WCDMA)—can be overruled by signal predistortion. The signal predistortion can handle arbitrary input waveforms.
Digital Predistortion with Memory Structures
Memoryless predistortion limits the linearization of power amplifiers with memory. With wider signal bandwidth, the memory effects on power amplifiers become prominent and require digital predistortion with memory.
Generally, the Volterra series is used to model predistortion with memory, and this works well for amplifiers where the signal bandwidth is smaller than the carrier frequency. The number of coefficients in the Volterra series increases exponentially with the memory length and order of non-linearity. The count of Volterra series coefficients is a disadvantage to this memory model and derivatives are taken from the Volterra series to address this problem.
Special cases of the Volterra series include the Wiener model, the Hammerstein model, the Wiener-Hammerstein model, and the memory polynomial model. In the Wiener model and the Hammerstein model, there are two subsystems namely the linear time-invariant (LTI) system and the memoryless nonlinearity (NL) system.
Linear Time-Invariant Systems vs. Memoryless Non-Linearity Systems
In the Wiener model, LTI comes first, followed by NL and the reverse order for the Hammerstein model. In satellite communication systems, the Wiener-Hammerstein model is utilized, which follows the order LTI-NL-LTI. The memory polynomial model is a generalization of the Hammerstein model and uses the diagonal kernels from the Volterra series. Digital predistortion modeling can be approached through direct or indirect learning architecture. What matters most in predistortion design is the linearization of the power amplifier nonlinear characteristics with memory and ensuring high efficiency and performance.
Improve System Performance With Digital Predistortion of Power Amplifiers
With wideband bandwidth, different transmitting formats, temperature, and component effects, the non-linearity of power amplifiers is increasing. To improve signal quality, system performance, and overall efficiency, it is important to implement digital predistortion of power amplifiers in wireless communication systems. You need to do a detailed analysis of the power amplifier and its characteristics to identify the effect of past inputs on present output. Choose predistortion techniques wisely to prevent limited linearization.
If you’d like to keep up-to-date with our System Analysis content, sign-up for our newsletter curating resources on current trends and innovations. If you’re looking to learn more about how Cadence has the solution for you, talk to us and our team of experts.