It is essential to accurately evaluate the performance of behavioral models and digital predistorters. This is useful for the proper selection of their structure, especially with the abundance of models that are available in the literature, as will be discussed in Chapters 4–7. Moreover, performance evaluation metrics can be adopted to decide on the model's parameters and its dimensions. These metrics can be defined either in the time or frequency domain.
This chapter is organized as follows. First, the focus will be on clearly distinguishing between the behavioral modeling and digital predistortion (DPD) applications and describing the specifics of each. Then, a variety of performance quantification metrics that have been reported in the literature for power amplifier (PA) behavioral models and digital predistorters will be thoroughly described. These are mainly categorized into two classes: time domain metrics and frequency domain metrics. Finally, the impact of memory effects on the performance assessment metrics is discussed and static nonlinearity cancelation techniques are introduced along with their relevance to behavior models and predistorter performance evaluation.
3.2 Behavioral Modeling versus Digital Predistortion
Behavioral modeling and predistortion are quite similar in various ways since most of the steps needed to derive a behavioral model or a digital predistorter are identical and most of the model ...