Oral
13 Jan 2025
Bit-patterned media recording (BPMR) is an advanced technology designed to surpass the superparamagnetic limit, enabling hard disk drives to achieve areal densities (ADs) beyond 1 Tb/in2. Increasing AD in BPMR systems by placing bit-islands closer together introduces significant inter-track interference (ITI) and inter-symbol interference (ISI), which degrade the bit-error rate (BER) during data detection [1-2]. Additionally, high AD exacerbates track misregistration (TMR), where the reader deviates from the desired track centerline, leading to severe BER degradation. This work addresses this issue by proposing a hybrid TMR mitigation technique based on K-means [3] and expectation-maximization (EM) algorithms [4] to enhance TMR prediction accuracy and improve BER in multi-head multi-track BPMR systems. The block diagram of the proposed system is shown in Fig. 1. For an AD of 3.0 Tb/in2, the track pitch (Tz) and bit period (Tx) are set to 14.5 nm. User bit sequences ak,l-1, ak,l, and ak,l+1 are encoded with a low-density parity check (LDPC) encoder before recording. Readback sequences rk,l-1, rk,l, and rk,l+1 are acquired by three readers and input into a hybrid K-means-EM-based TMR estimator. This estimator determines the TMR level by finding two sets of centroids for the readback signals using K-means and EM algorithms, dynamically selecting the estimated TMR output based on the predicted TMR range. Experiments show that K-means offers better accuracy at TMR levels of 0.5, 1.0, and 1.5 nm, while EM outperforms at 2.0, 2.5, and 3.0 nm. e.g., if estimated TMR values are 1.0 nm (K-means) and 1.5 nm (EM), the 1.0 nm value is chosen. Combining these algorithms achieves over 85% estimation accuracy across all TMR levels. For TMR mitigation, the appropriate 2D equalizer and 1D generalized partial response target coefficients for the SOVA detector are selected based on the estimated TMR level. The proposed system with TMR mitigation technique achieves BER performance close to an ideal system with 100% TMR estimation accuracy and significantly outperforms a conventional system without TMR mitigation with a performance gain up to 1.25 dB as shown in Fig. 2.References: [1] W. Busyatras, C. Warisarn, Y. Okamoto, et.al., AIP Advances., Vol. 7, No. 5, p. 056501-1, (2016) [2] K. Kankhunthod, C. Buajong, and C. Warisarn, IEEE Magn., Vol. 11, p. 1-5, (2020) [3] S. Jeong and J. Lee, IEEE Tran. Magn., Vol. 59, No. 3, p.1-5, (2023) [4] G. Kong, T. Kim, and M. Jung, IEEE Magn. Letters., Vol. 13, p. 1-5, (2022)
