Recovery of recorded data

We have been looking at the recovery of recorded data from bit patterned media (BPM) systems since 2002 and have published a number of journals papers outlining our observations. Key to this work was the development of a novel three-dimensional (3D) model of the readout process in magnetic storage systems, that is essential in order to accurately predict the replay signal in the case where the island geometry and read head geometry are comparable in size, as is the case in BPM.

3D Replay Model

In order to investigate the data recovery process in patterned media storage systems it is essential that a realistic replay waveform is used. However, as there is currently no practical implementation of a storage system incorporating patterned media we have to rely on simulations to generate the expected replay signals.

In the case of a patterned medium, the size and shape of the recorded domains are constrained by the geometry of the magnetic island. In this situation, a more accurate replay model is required which takes into account variation in magnetization in both the along-track (x) and across-track (z) directions; this requires the extension of the reciprocity integral over 3-D space. The replay model developed predicts the signal waveform that would be observed from a GMR read sensor as it scans along a track of islands in a magnetic storage system incorporating a patterned perpendicular magnetic storage medium. The replay model is based on an extension of the standard two-dimensional (2D) reciprocity approach to three-dimensional (3D) space and takes into account the geometrical aspects of both the patterned recording medium and the GMR read sensor used. The voltage signal from the GMR read head is proportional to the signal flux injected into the GMR sensor at the air-bearing surface (ABS). Hence, the signal flux at the GMR sensor along the scan direction is given by the 3D reciprocity integral.

The calculation of the voltage signal from the GMR read head is simply a Fourier transform process involving the magnetization distribution across the plane of the storage medium, and simulated potential distributions below the ABS of the GMR read sensor at the top and bottom surfaces of the storage medium. Once the recorded magnetization pattern in the storage medium has been generated, such as an isolated magnetic island, then the 3D reciprocity model can be used to predict the output response to that magnetic structure. More importantly, the model allows the generation of a readout waveform including the presence of inter-track interference (ITI) introduced as a result of the read head sensing islands along tracks adjacent to the main data track.

If data is recorded to a patterned medium such that each island is used to store a single recorded bit, then the replay waveform due to a train of random recorded data is produced by the superposition of an ideal step response, generated for an isolated island using the 3D reciprocity model described. This process involves the superposition of the step response at points in the replay waveform corresponding to the ideal physical location of the leading and lagging edges of each island as the GMR read head scans along a track of islands. As lithography jitter leads to variations in the position of the edges of each island, it is easily introduced into the simulated replay waveform by varying the position at which the step response is superposed. Here, the random shift in the edge position due to lithography jitter is defined to be a Gaussian distribution of zero mean and standard deviation specified in nanometers from the ideal position of the island edge. The Gaussian distribution in this case is truncated so that the edge shift introduced is no greater than the separation between islands. Jitter is added in the along-track direction only, as jitter across-track has little effect on the replay signal due to the relatively large read sensitivity function of the head compared with the track width.

The developed simulation takes into account:

  • finite GMR read width
  • island shape (square, round, BAR)
  • island distribution
  • media characteristics (thickness, presence of SUL etc)
  • ITI due to adjacent tracks
  • the effect of read head track misregistration (TMR)
  • lithography noise introduced as variations in island position and size.

We have used the model to help investigate the issue of data recovery in BPM systems, as well as design new approaches to help overcome some of the inherent problems associated with BPM storage.


  • P W Nutter, D M McKirdy, B K Middleton, D T Wilton, H A Shute, "Effect of island geometry on the replay signal in patterned media storage", IEEE Trans Mag, 40(6), 2004, 3551-3558. [pdf]
  • P W Nutter, I T Ntokas, B K Middleton, "An investigation of the effects of media characteristics on read channel performance for patterned media storage", IEEE Trans Mag, 41(11), 2005, 4327-4334. [pdf]
  • P W Nutter, I T Ntokas, B K Middleton, D T Wilton, "Effect of island distribution on error rate performance in patterned media", IEEE Trans Mag, 41(10), 2005, 3214-3216. [pdf]
  • I T Ntokas, P W Nutter, B K Middleton, "Evaluation of read channel performance for perpendicular patterned media", J Magn Magn Mater, 287, 2005, 437-441. [pdf]
  • I T Ntokas, P W Nutter, C J Tjhai, M Z Ahmed, "Improved data recovery from patterned media with inherent jitter noise using low-density parity-check codes", IEEE Trans Mag, 43, 2007, 3925-3929. [pdf]
  • P W Nutter, Y J Shi, B D Belle, J J Miles, "Understanding Sources of Errors in Bit-Patterned Media to Improve Read Channel Performance", IEEE Trans Mag, 44, 2008, 3797-3800. [pdf]
  • Y J Shi, P W Nutter, B D Belle, J J Miles, "Error Events Due to Island Size Variations in Bit Patterned Media", IEEE Trans Mag, 46(6), 2010, 1755-1758. [pdf]
▲ Up to the top