744 IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 19, NO. 4, APRIL 2001 Dropout-Tolerant Read Channels Fatih Sarigoz, Hongwei Song, B. V. K. Vijaya Kumar, Senior Member, IEEE, and James A. Bain Abstract—Dropouts are intermittent losses of signal commonly seen in magnetic tape recording readout. The main reason for such losses is the increased spacing between the head and the medium due to media defects or debris particles. The resulting signal is not only degraded by the apparent amplitude loss, but the characteris- tics of the pulses due to transitions are also changed. Moreover, in many cases, the locations of the pulses are altered, causing excessive amounts of peakshift. In this paper, a model, linking the liftoff to these effects, is presented. Experimental verification of the model using actual signals from a test tape drive is also given. Artificial dropout waveforms generated using this model are used to test two different read channel strategies. The first approach is an equaliza- tion-based correction scheme that attempts to undo the dropout. Results indicate that the dropout effects can be almost completely eliminated if an appropriate equalization procedure is applied. As an alternative approach, it is shown that the use of turbo coding in the presence of dropouts appears to be promising. Index Terms—Adaptive equalization, defects, dropouts, dropout correction, dropout peakshift, envelope detection, spacing loss, turbo codes. I. INTRODUCTION M ANY of the errors in magnetic tape recording are re- ported to come from events of sporadic loss of ampli- tude, called “dropouts” [1], [2]. The signal exhibits an envelope, as shown in Fig. 1(a), during a dropout event, which appears as if it is amplitude modulated. The result may be as serious as a complete signal loss, and the dropout event can span hundreds of bits. The main sources of dropouts are known to be media defects (asperities or lumps of magnetic or nonmagnetic mate- rial) [3], [4] and loose particles in the head media interface [5]. The random nature of the event and the variety in the number of sources make dropouts difficult to characterize. However, if spacing loss is ascribed as the sole mechanism causing the dropouts, models can be developed to characterize the phenom- enon [4], [6]. A representative model is presented in Fig. 1(b), in which the head-to-media distance is shown to be increased due to liftoff by a particle or a protrusion from the media. The effect of spacing in a magnetic recording system on the readout signal is very significant. In fact, high densities are typically accomplished by decreasing the head-to-media spacing and scaling other system parameters accordingly. A smaller value of spacing means writing sharper magnetic Manuscript received March 14, 2000; revised November 10, 2000. This work was supported in part by the National Science Foundation, under Grant ECD- 8907068. F. Sarigoz is with Agere Systems (formerly Lucent Technologies), Allentown, PA 18109-2139 USA (e-mail: sarigoz@agere.com) H. Song, B. V. K. V. Kumar, and J. A. Bain are with the Data Storage Systems Center, Department of Electrical and Computer Engineering, Carnegie Mellon University, Pittsburgh, PA 15123 USA. Publisher Item Identifier S 0733-8716(01)01766-8. Fig. 1. (a) Readback signal from a tape drive with a dropout. (b) A possible model of dropout formation (adapted from [24]). transitions and sensing stronger and sharper readout pulses. In an increased spacing event that may happen during both writing and reading, however, signal quality is degraded, transitions are broadened, the signal is weakened, and readout pulses are widened. In [7], it is reported that a systematic shift of transitions also takes place during such an event and, therefore, should be added to this list. Dropouts have been studied from various perspectives. Sev- eral studies have looked at the problem from a tribology point of view, examining head and media wear and the conditions that affect it [5], [8]–[10]. Several researchers have attempted to model the tape tenting effect due to a particle [11], [12]. Others have performed statistical analysis of some of the characteris- tics of dropouts, such as their duration, depth, amplitude statis- tics [13]–[17], and error distributions [1], [4], [18], [19]. De- velopments of more robust timing recovery schemes have been reported in [20] and [21]. Dropouts are usually handled by error correction (ECC) schemes (Reed–Solomon codes) that are de- signed to correct burst errors as well as random errors caused by noise [16], [22]. Except the use and design of ECC schemes, none of these approach the problem from a read channel per- spective. In [23], temporal variations in the channel character- istics of a digital video recording channel in a dropout are mea- sured and adaptive equalization is suggested. In this paper, we focus on the changes in the signal during a dropout with the goal of designing a scheme to correct those changes. The effects of a dropout on the signal are quantified with two models [24]; one that links a spacing profile to changes in pulse shape and amplitude, and the other to peakshift. Based on these models, a method has been developed that attempts to correct all three effects by appropriate equalization [24]. The dropout correction scheme is a front-end approach applying cor- rection to the readback signal and does not assume a partic- ular equalization/detection strategy. An alternative approach is 0733–8716/01$10.00 © 2001 IEEE