Abstract: Digital predistortion is a widely used linearization technique for power amplifiers. The decomposed vector rotation (DVR) digital predistortion model is widely used to improve the non-linearity of power amplifier due to its easy-to-implement hardware structure and good linearization performance. However, the computational complexity and cost of parameter extraction of the DVR model will increase sharply with the increase of the number of operator matrix entries and data length. To address this issue, a digital predistortion method with low operational complexity is proposed based on the DVR model. The proposed method consists of two key components including a low complexity decomposition vector rotation (LCDVR) digital predistortion model and a non-uniform selection sampling (NSS) algorithm. These components aim to reduce the operational cost of model parameter extraction. The proposed LCDVR model reduces the multiplication operation by increasing the number of zero terms in the operator matrix. Simultaneously, according to the characteristics of signal amplitude distribution, the NSS algorithm is employed to select data sampling points. This selective sampling helps reduce the data length required in parameter extraction, leading to a relatively uniform signal amplitude distribution after selection. Such uniform distribution facilitates the analysis of the selection of amplitude segment values of the LCDVR model. The experimental results show that when the length of the input signal data is 70 000, the θ_max of LCDVR model is 0.7 and θ_min is 0.3, and when the length of the data after the NSS algorithm is 10 849, the multiplication computation required for parameter extraction of the proposed method is only 2.24 % of that of the DVR model, while still maintaining a substantial linearization effect. Consequently, the proposed method significantly reduces the operational complexity of parameter extraction while preserving linearization accuracy, thereby enhancing its applicability and feasibility.