Dynamic magnetic information storage or retrieval – General processing of a digital signal – Head amplifier circuit
Reexamination Certificate
1999-07-30
2002-04-30
Neal, Regina Y. (Department: 2651)
Dynamic magnetic information storage or retrieval
General processing of a digital signal
Head amplifier circuit
C360S065000, C706S022000, C706S031000
Reexamination Certificate
active
06381083
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to signal recording and playback equalization, and in particular to recording and equalization of signals of a magnetic recorder.
2. Description Relative to the Prior Art
In order to attain higher transfer densities in the recording of digital magnetically recorded data, the trend has been towards ever increasing pulse packing on the medium. Typically the digital recording uses some form of binary encoding, such as NRZI, where binary 1's are represented as transitions from one state of magnetization of the medium to another state, and 0's are represented by no change in magnetization at specific clock times. The two states are usually the plus and minus saturation states of the medium, and the time of magnetization reversal during a transition is immaterial. Accordingly, the hysteresis of the medium does not directly affect the recording or recovery of the data in straight binary recording.
However, increasing the pulse packing of the medium leads to signal distortion on playback of the signal. The crowding together of the transitions causes peak shifts in the recovered signals, and adjacent transitions are no longer readily identifiable on playback., This is due to non-linearities arising from the mutual interference of sequentially recorded pulses on the magnetic medium, and becomes more severe as the recording density is increased.
It has been known in the prior art to use a neural network to represent nonlinear maps as part of a signal equalizer in a binary recording/playback system at high packing densities. U.S. Pat. No. 5,361,327 issued in the name of Takahashi discloses a magnetic disk drive system in which a computer simulated neural network is used to derive the parameter values of an equivalent “hard wired” neural network employed as a waveform equalizer. Similarly, Padden in U.S. Pat. No. 5,594,597 teaches the use of a neural network in a magnetic playback system to equalize a playback signal subject to playback non-linearities due to interference between crowded pulses. Padden applies his disclosure to a magnetic disk recording system. Such systems require varying the recording density as a function of the radius to maintain constant density recording, and also requires control of a phase locked loop to maintain constant recorded wavelength as a function of the radial location of the recorded zone. It will be appreciated that the above patents disclose magnetic recording systems in which only the two signals levels characteristic of saturation binary recording are recorded on the magnetic medium. The information content of such binary recording is limited by the use of two recording levels, and, as previously stated, by the non-linearities introduced by pulse crowding. In the binary recording systems of the prior art, the only way to increase information content is by higher density recording with a concurrent severe loss of signal-to-noise ratio and increase in the problem of playback equalization due to pulse crowding.
SUMMARY OF THE INVENTION
The present invention teaches the use of multilevel, high density magnetic recording to allow storage of quantized data with increased information content. The use of 4 level recording increases the information content by a factor of 2 compared to the binary recording of the prior art. The invention utilizes biased magnetic recording to eliminate the hysteresis present in the magnetic medium, thereby providing a single valued input/output transfer function necessary in a multilevel recording system. While the transfer function is single valued, it is linear only over the range from 0 magnetization to about −10 db below the saturation level. If recording is restricted solely to the linear range, then the signal to noise ratio (SNR) is correspondingly decreased. To utilize the maximum SNR available for recording, recording at levels into the non-linear region of the transfer function is permitted, and a nonlinear map having the ability to approximate a wide variety of multidimensional non-linear transformations is implemented to provide the waveform equalization necessary to recover the recorded data. In the preferred embodiment of this invention, realization of this map, representing the distortions of the communication channel, is realized using a neural network. The invention is disclosed using both 4 level amplitude modulation for an improvement by a factor of 2 in information content over straight binary recording, and by a combination of 4 level amplitude modulation and phase modulation in a quadrature amplitude modulation (QAM) configuration which provides a corresponding improvement by a factor of 4 over straight binary recording. More levels of recording both in amplitude and in phase are possible depending upon the signal-to-noise ratio attainable in the recorder system.
While the invention has been described in terms of magnetic recording, it will be appreciated that the teachings of the invention may be applied to recording and playback of quantized information from other media capable of supporting multiple level analog signal recording, such as or thermoplastic tape recording.
REFERENCES:
patent: 5542054 (1996-07-01), Batten, Jr.
patent: 5594597 (1997-01-01), Padden
patent: 5818653 (1998-10-01), Park et al.
Abarbanel Henry D. I.
Bryant Paul H.
Korzinov Lev N.
Lemke James U.
Rulkov Nikolai F.
Applied Nonlinear Sciences, LLC
Neal Regina Y.
Robbins Daniel
LandOfFree
Multilevel signal equalization utilizing nonlinear maps does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Multilevel signal equalization utilizing nonlinear maps, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Multilevel signal equalization utilizing nonlinear maps will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2872061