Pruned volterra based behavioral modeling for power amplifier on outband spectra analysis

Behavioral modeling techniques provide a convenient way to predict nonlinear response without the computational complexity of full circuit analysis for power amplifier (PA) with memory effects. A pruning approach for Volterra behavioral model is presented by introducing two control factors. It combines the called near-diagnonality algorithm and dynamic deviation reduction algorithm, then use the “OR” relationship to select the unknown Volterra kernels which are more sensitive to the output error. The structure of Volterra model can be significantly simplified without incurring loss of model fidelity. Finally, a power amplifier is employed for validation and good agreement is obtained. The pruned Volterra model is effective and flexible on outband spectra analysis of PA with strong nonlinearity and longer-term memory.