一种优化的gOMP稀疏OFDM信道估计方法

发布时间：2018-01-14 18:24

本文关键词：一种优化的gOMP稀疏OFDM信道估计方法　出处：《工程科学与技术》2017年05期 　论文类型：期刊论文

【摘要】：无线多径信道多呈现稀疏特性,即信道时延扩展大,但是路径的个数少,利用信道先验稀疏信息的稀疏信道估计方法可以提高稀疏信道的估计准确性。针对贪婪算法选择字典原子在残余误差最小意义下的非最优性,及广义正交匹配追踪算法gOMP利用含噪信号估计稀疏信道过程中过多的选择字典原子导致gOMP算法重建性能下降的问题,提出优化的广义正交匹配追踪算法(optimized generalized orthogonal matching pursuit,OgOMP)。在OgOMP算法原子选择阶段,采用使残余误差最小化的原子选择标准代替残差与字典内积绝对值最大化的原子选择标准以选择原子。为删除多余的误选原子,添加原子精炼步骤对每一步迭代后选择的字典原子进行二次选择,选择对应最大信道衰落系数的原子,选择的原子数与信道稀疏度相同,删除错选原子以保证重建信号与原始信号的稀疏性一致。本文仿真对比了gOMP和OgOMP算法的信道估计均方误差、误码率、残差收敛速度以及不同导频数、不同原子选择数对算法的影响。仿真结果表明:相同的误码率下,OgOMP算法比gOMP算法在估计稀疏信道时最大可以节省4 dB的信噪比,信噪比为20 dB时均方误差最大可以减小5 dB;两种算法的残差收敛速度均优于MP算法;导频数的增加可以减小两种算法的信道估计均方误差,相同信道估计性能下OgOMP算法具有更小的导频开销;每步迭代选择的原子数目不同时,相比于gOMP算法,OgOMP算法性能基本不变,具有更好的稳定性,仿真结果验证了改进算法的有效性。
[Abstract]:Many wireless multipath channels show sparse characteristics, that is, the channel delay spread is large, but the number of paths is small. The estimation accuracy of sparse channel can be improved by using the sparse channel estimation method with prior sparse channel information. The non-optimality of dictionary atom in the sense of minimum residual error is selected for greedy algorithm. And generalized orthogonal matching tracking algorithm gOMP uses noisy signals to estimate sparse channel. Too many choice dictionary atoms lead to poor reconstruction performance of gOMP algorithm. An optimized generalized orthogonal matching pursuit is proposed. OgOMP. In the atomic selection phase of the OgOMP algorithm. The atomic selection criterion which minimizes the residual error is used instead of the atomic selection standard which maximizes the absolute value of the residual error and the dictionary product to select the atom. The dictionary atoms selected after each step iteration are selected by adding the atomic refining step, and the atoms corresponding to the maximum channel fading coefficient are selected, and the number of atoms selected is the same as the channel sparsity. Error-selective atoms are deleted to ensure the sparsity of the reconstructed signal and the original signal. The channel estimation mean square error and bit error rate of the gOMP and OgOMP algorithms are simulated and compared in this paper. The convergence rate of residual error and the influence of different pilot numbers and different atomic selections on the algorithm. The simulation results show that: under the same bit error rate. The maximum SNR of the OgOMP algorithm is 4 dB less than that of the gOMP algorithm, and the maximum mean square error can be reduced by 5 dB when the SNR is 20 dB. The convergence speed of the two algorithms is better than that of MP algorithm. The increase of pilot number can reduce the channel estimation mean square error of the two algorithms. Under the same channel estimation performance, the OgOMP algorithm has less pilot overhead. Compared with gOMP algorithm, the number of atoms selected by each iteration is different, and the performance of OgOMP algorithm is basically invariant. The simulation results show that the improved algorithm is effective.
【作者单位】：天津大学电气自动化与信息工程学院;
【基金】：国家自然科学基金资助项目(61571318) 海南省重点研发计划资助项目(ZDYF2016153) 青海省自然科学基金重点项目资助(2015-ZJ-904)
【分类号】：TN929.53
【正文快照】： 随着数字电子技术和大规模集成电路的发展,正交频分复用(orthogonal frequency division multi-plexing,OFDM)因其具有较高的频率利用率和抗多径干扰的能力得到了迅猛发展[1 2]。OFDM系统把频率选择性衰落信道转换为并行的平坦衰落信道,仅使用简单的均衡器即可实现信道的均衡

【相似文献】