Lattice结构无限冲击响应（IIR）数字滤波器，不仅具有高选频、低延迟和低运算次数等IIR滤波器共有特性，还对有限字长（FWL）效应具有强鲁棒性。本项目针对Lattice结构IIR滤波器优化设计面临的优化模型高度非凸和无法实现精确线性相位两大挑战，利用lattice结构对FWL效应强鲁棒性和部分优化方法可降低优化模型非凸性之优点，提出一种序贯部分优化方法，将原高度非凸的Lattice结构IIR滤波器设计问题转化为一系列非凸性程度较低的子问题，每个子问题中仅更新部分滤波器系数。使用这种方法，研究基于频率相应误差最小化和相位误差最小化的lattice结构IIR滤波器约束最小二乘和约束minimax设计，提出lattice结构IIR滤波器设计的序贯部分优化理论与算法，设计出比现有算法性能更优且对FWL效应影响有强鲁棒性的滤波器，使其可广泛应用于通信、生物医学和实时音/视频等数字信号处理领域。

Infinite impulse response (IIR) digital filters with lattice structure not only have the common characteristics of IIR filters such as high frequency selectivity, low group delay, and low computational complexity, but also have strong robustness to the finite word length (FWL) effect. In this project，for the two main challenges in optimal design of IIR filter with lattice structure that optimization model is highly non-convex and exact linear phase cannot be achieved, we combine the advantage of partial optimization methods that can reduce the non-convexity of the optimization model with the strong robustness of the lattice structure to the FWL effect, and propose a sequential partial optimization method to convert the original highly non-convex design problem into a series of low non-convex sub-problems, in each of which only partial of the filter coefficients are updated. Using this method, constrained least-square and constrained minimax designs of IIR filter with lattice structure are researched based on minimizing frequency response error and phase error, for the non-convex and non-linear phase problems of the optimization model. And sequential partial optimization theory and algorithm for the design of IIR digital filter with lattice structure are proposed to design filters with better performance than existing algorithms and strong robustness to FWL effect. The designed filters can be widely used in many digital signal processing fields such as communications, biomedicine, and real-time audio / video processing.