硬盘读写通道的FIR噪声预测算法与仿真模型

FIR Noise Prediction Algorithm and Simulation Model of Read/Write Channel in Hard Disk

本文提出了一种嵌入FIR噪声预测的部分响应最大似然算法(NPML-Noise-prediction Partial-response Maximum Likelihood),该算法改进了传统部分响应最大似然算法(PRML-Partial Response Maximum Likelihood)中Viterbi译码过程的分支度量计算方法,能白化信号中有色噪声,减少误码率,提高硬盘读写通道的性能。本文还建立了读写通道的图形化仿真模型,提出了FIR滤波器长系数的快速逼近方法,并构造了检验模型工作正确性的方法。通过仿真模型对传统的Viterbi译码算法与嵌入FIR预测的NPML算法进行了分析。结果表明,在相同的存储密度下,NPML能获得比PRML高xxdb的信噪比和低xxdb的位错误率。

Because of the improvement of the storage density in the hard disk, and the decreased space between bits to bits, the noises in the read back signals of magnetic head have been nonstationary and correlative colored noises, which can not adapt to Viterbi detect algorithm. By embedding FIR Noise Prediction Maximum Likelihood （NPML） algorithm to the computing process of the branch metric in Viterbi, the colored noises can be effectively whiten. So it can be used to reduce the bit error rate （BER） and improve the system performance. In this paper the comparing analysis of the ordinary Viterbi detector, the noise prediction with FIR is given respectively. Furthermore a SIMULINK simulation model of read/write channel, a fast approaching way to acquire the FIR taps, and a method to verify the correction of the model have been constructed to analyze the different noise prediction algorithms. Simulation results show that under the same storage density NPML detector can attain SNR which is xxdb higher and BER which is xxdb lower than PRML detector.