数字预失真（DPD）技术是目前射频前端的主流线性化技术。随着未来无线通信技术的发展，频谱资源的利用率越来越高，带宽也不断提高，这对功放线性化技术提出了更高的要求。传统的DPD技术受限于模数转换器即ADC采样速率及系统成本，用于宽带无线通信中将受到严重限制。因此，降低采样速率是宽带DPD急需解决的问题。压缩感知（CS）近年来发展十分迅速，已经在包括信号采集在内的众多领域中取得了突破性成果。本项目将压缩感知理论引入DPD研究中，从常用的宽带OFDM信号入手，分析信号本身特性、先验知识及其与DPD系统之间的内在联系，研究欠采样DPD系统结构的理论与技术。通过建立宽带自适应稀疏DPD模型，并设计出相应的CS算法，使得在欠采样下能直接或间接恢复出稀疏DPD模型系数，进而实现DPD。CS算法作为纽带，将宽带自适应稀疏DPD模型与欠采样DPD结构连接起来，构成一个完整的欠采样DPD系统。

Digital PreDistortion (DPD) is the most attractive linearization technique used in radio frequency front-ends. Along with the development of next generation wireless communication system, efficient use is made of the spectrum resource and of the wider bandwidth,which requires a more challenging linearization technique. Conventional DPD could be seriously limited in dealing with the wideband wireless communication systems, due to the high speed Analog-to-Digital converter requirements and the high system cost. Therefore, reducing the sampling rate for wideband DPD will be an urgent problem. Compressive Sensing (CS) has a rapid development recently, and significant breakthroughs have been made in many areas including signal acquisition. This proposal introduces CS theory to DPD. We will begin with the study of the priori knowledges of the commonly used wideband OFDM signals, then analyze the inherent relations between the signal and DPD system, and study the theory and techniques of the sub-sampling DPD structure and system. Then, we will develop a wideband adaptive sparse DPD model, and work out a corresponding CS algorithm so that the parameters of the sparse DPD model can be recovered from the sub-sampled signal directly or indirectly and that the DPD can be realized. The CS algorithm links the wideband adaptive sparse DPD model and the sub-sampling structure, and makes up a complete sub-sampling DPD system.