2025年03期 v.46 17-22页
刘雅宁 李江林
（北京中电华大电子设计有限责任公司，北京 102209）
摘要：针对信道传输特性不理想导致的码间串扰问题，提出一种基于核支持向量机的盲均衡算法。该算法将恒模算法的误差函数引入支持向量机的代价函数中，改善了恒模算法在非线性信道下易产生相位偏移的问题；利用小波函数构造支持向量机的核函数，降低了求解非线性问题的运算复杂度，进一步提升了盲均衡算法的运算效率。仿真测试结果表明，与恒模算法、采用Sigmoid核函数的核支持向量机相比，采用小波函数作为核函数的核支持向量机收敛速度更快：且基于此种核支持向量机的盲均衡算法分类准确度更高。
关键词：盲均衡；核支持向量机；恒模算法；小波函数；码间干扰
中图分类号：TN911.5 文献标志码：A 文章编号：1674-2605(2025)03-0003-06
DOI：10.12475/aie.20250303 开放获取
A Blind Equalization Algorithm Based on Kernel Support Vector Machine
LIU Yaning LI Jianglin
(CEC Huada Electronic Design Co., Ltd., Beijing, Beijing 102209, China)
Abstract: To address the inter-symbol interference (ISI) caused by non-ideal channel transmission characteristics, a blind equalization algorithm based on kernel support vector machines (SVM) is proposed. This algorithm introduces the error function of the constant modulus algorithm (CMA) into the cost function of the support vector machine, mitigating the phase offset issue of CMA in nonlinear channels. By constructing the SVM kernel function using wavelet functions, the computational complexity of solving nonlinear problems is reduced, further improving the computational efficiency of the blind equalization algorithm. Simulation results demonstrate that compared to the constant modulus algorithm and SVM with Sigmoid kernel functions, the SVM with wavelet kernel functions exhibits faster convergence speed. Moreover, the blind equalization algorithm based on this kernel SVM achieves higher classification accuracy.
Keywords: blind equalization; kernel support vector machine; constant modulus algorithm; wavelet function; inter-symbol interference
