Research on Ultrahigh-Density Perpendicular Magnetic Recording Using Adaptive PRML Systems

根据CoPtCr-SiO_2三种垂直磁记录(PMR)介质的实验数据，得到了热稳定性因子KUV/KT分别为94、60和45的热衰减模型。利用热衰减模型对128/130(0，16/8)RLL码的PRML、GPRML和GPRML-AR系统的长期误码性能进行了计算机模拟，结果表明，在t_E=10.8秒的所有PRML系统中，在K_UV/KT=94的介质中，误码性能没有下降。而在K_UV/KT=60的介质中，BERS有一个数量级左右的降解，而在K_UV/KT=45的介质中，BERS的降解有三个数量级左右。然而，GPR1ML系统在误码率=10^&lt；-4&gt；时，信噪比分别提高了0.4、0.4和4.9dB.对于K_UV/k_T分别为94、60和45的PR1ML系统，GPR1ML-AR系统在t_E=10^6s时的性能改善分别为1.5、1.8和6.6分贝。当采用神经网络作为均衡器，并使用反向传播算法和遗传算法相结合的混合遗传算法对网络进行简化和优化时，当K_UV/k_T=60时，直到t_E=10^5时，PRML系统的性能没有下降。因此，当采用神经网络作为均衡器，并使用BP算法和遗传算法相结合的混合遗传算法对网络进行简化和优化时，在K_UV/k_T=60和t_E=10^5时，没有发现PRML系统的性能下降。因此，当采用神经网络作为均衡器，并使用反向传播算法和遗传算法相结合的混合遗传算法对网络进行简化和优化时，当K_UV/K_T=60时，直到t_E=10^5时，PR1ML-AR系统的性能没有下降。因此，当采用神经网络作为均衡器，并使用这表明神经网络均衡与PRML系统的结合对热衰减具有良好的适应性。此后，将研究使用神经网络均衡器的自适应GPRML和PRML-AR系统。

A thermal decay model based on the experimental data for three CoPtCr-SiO_2 perpendicular magnetic recording (PMR) media with thermal stability factors K_uV/kT of 94, 60 and 45 is obtained. The long-term BER performance of PRML, GPRML and GPRML-AR systems for the 128/130(0,16/8) RLL code is evaluated by computer simulation using the thermal decay model at a linear recording density of 1000 kBPI, which is difficult to evaluate experimentally.It is clarified that, in all PRML systems at t_E = 10^8 seconds, there is no degradation in BER performance for the medium of K_uV/kT =94, while an order of magnitude or so degradation in BERs is found for the medium of K_uV/kT =60 and three orders of magnitude or so degradation in BERs can be seen for the medium of K_uV/kT =45. Nevertheless, GPR1ML system provides the SNR improvements of 0.4, 0.4 and 4.9 dB at BER=10^<-4> over PR1ML system for the media of K_uV/k_T =94, 60 and 45, respectively, and GPR1ML-AR system offers the improvements of 1.5, 1.8 and 6.6 dB at t_E=10^6 is seconds.When a neural network is employed as a equalizer and the network is simplified and optimized by using a hybrid genetic algorithm which is the combination of a back propagation algorithm and a genetic algorithm, the performance degradation of PRML system is not found for K_uV/kT =60 and till t_E = 10^5. Therefore, this shows that the combination of neural network equalization and PRML system has the good adaptability for the thermal decay. Hereafter, the adaptive GPRML and PRML-AR systems using a neural network equalizer will be studied.