A study of turbo decoding with embedded AR channel model for perpendicular recording

Performance improvement of partial-response maximum-likelihood (PRML) system in the perpendicular magnetic recording channel with jitter-like noise is studied. The performance of a Viterbi detector is improved by applying an autoregressive (AR) model in order to estimate the detector input sequence. The required signal-to-noise ratio to achieve a bit-error rate (BER) of 10/sup -4/ is obtained by computer simulation. In addition, a BER performance is also obtained by read/write experiments using a spin-stand. The results show that the proposed PRML system using an AR model exhibits better performance when compared with the normal PRML systems without an AR model.

In this paper, we study the effect of channel memory order on the information rate bounds in time-varying flat-fading (FF) channels. We model time variations of the FF channel with autoregressive (AR) processes with varying degrees of model order. We observe that in high SNR conditions (SNR ? 20 dB), the information rate penalty of not knowing the AR channel is a non-increasing function of the AR model order. This is expected, since the AR channel predictability cannot decrease with increasing its order. However, in low SNR conditions, the information rate penalty in low-order AR channels can be lower than those in high-order AR channels. Likewise, the intuitive and universal monotonic increase of the information rate bounds with the AR model order is only observed in almost noiseless conditions. In the low SNR regime, however, the achievable information rate bounds in low-order AR channels can be higher than those in high-order AR channels.

The BER performance of iterative decoding systems and PRML systems in a perpendicular magnetic recording channel is experimentally investigated. A performance evaluation system consisting of a spinstand, a recording waveform generator, a digital storage oscilloscope and a personal computer was used. PR1ML, PR2ML, EPR3ML, E/sup 2/PR3ML and ME/sup 2/PR4ML systems are discussed as a PRML channel. The 8/9 and 16/17 iterative decoding systems consists of the encoder, with RSC encoder serially concatenated with preceded PR channels and the iterative decoder with APP modules using Max-log-MAP algorithm. The results show that EPR3ML and E/sup 2/PR3ML provide good BER performance in a perpendicular magnetic recording channel and that the iterative decoding system is also an efficient way to improve the BER performance in the channel.

Speaker and language recognition in noisy and degraded channel conditions continue to be a challenging problem mainly due to the mismatch between clean training and noisy test conditions. In the presence of noise, the most reliable portions of the signal are the high energy regions which can be used for robust feature extraction. In this paper, we propose a front end processing scheme based on autoregressive (AR) models that represent the high energy regions with good accuracy followed by a modulation filtering process. The AR model of the spectrogram is derived using two separable time and frequency AR transforms. The first AR model (temporal AR model) of the sub-band Hilbert envelopes is derived using frequency domain linear prediction (FDLP). This is followed by a spectral AR model applied on the FDLP envelopes. The output 2-D AR model represents a low-pass modulation filtered spectrogram of the speech signal. The band-pass modulation filtered spectrograms can further be derived by dividing two AR models with different model orders (cut-off frequencies). The modulation filtered spectrograms are converted to cepstral coefficients and are used for a speaker recognition task in noisy and reverberant conditions. Various speaker recognition experiments are performed with clean and noisy versions of the NIST-2010 speaker recognition evaluation (SRE) database using the state-of-the-art speaker recognition system. In these experiments, the proposed front-end analysis provides substantial improvements (relative improvements of up to 25%) compared to baseline techniques. Furthermore, we also illustrate the generalizability of the proposed methods using language identification (LID) experiments on highly degraded high-frequency (HF) radio channels and speech recognition experiments on noisy data.

A new rate 1/2 turbo code with short memory 2 is developed in this paper. The proposed turbo code consists of two 4-state parallel concatenated non-punctured systematic rate-2/3 convolutional codes. Using a modified maximum a posteriori (MAP) decoding algorithm working on dual codes, its bit error rate (BER) performance is comparable to that of the classic 16-state turbo code at short block lengths, such as k=1024, while its decoding complexity is less than 1/8 that of the classic 16-state turbo codes; for information block length k=504, its BER performance is better than that of the (1008, 3, 6) LDPC codes with comparable decoding complexity. The dual code decoding algorithm is intrinsically partial parallel and a considerably higher throughput of the turbo decoder can be obtained compared with classic turbo codes consisting of punctured codes. The simulation results show that the proposed turbo code can be a competitive technique for short block length applications.

Under appropriate conditions, stochastic processes are described by the ARMA model, however, the AR model is popularly used in reactor noise analysis. Hence, the properties of AR model as an approximate representation of the ARMA model should be made clear. Here, convergence of AR-parameters and PSD of AR model were studied through numerical analysis on specific examples such as the neutron noise in subcritical reactors, and it was found that: 1. The convergence of AR-parameters and AR model PSD is governed by the "zero nearest to the unit circle in the complex plane" (μ−1,|μ|<1) of the ARMA model transfer function.2. The AR-parameters of AR(M) model have biases from those for the infinite model order, and these biases decrease approximately in proportion to |μ|M.3. The AR model of the neutron noise of subcritical reactors needs a large model order because of an ARMA-zero very close to unity corresponding to the decay constant of the 6-th group of delayed neutron precursors.4. In applying AR model for system identification, much attention has to be paid to a priori unknown error as an approximate representation of the ARMA model in addition to the statistical errors.

Very large scale integration (VLSI) design methodology and implementation complexities of high-speed, low-power soft-input soft-output (SISO) a posteriori probability (APP) decoders are considered. These decoders are used in iterative algorithms based on turbo codes and related concatenated codes and have shown significant advantage in error correction capability compared to conventional maximum likelihood decoders. This advantage, however, comes at the expense of increased computational complexity, decoding delay, and substantial memory overhead, all of which hinge primarily on the well-known recursion bottleneck of the SISO-APP algorithm. This paper provides a rigorous analysis of the requirements for computational hardware and memory at the architectural level based on a tile-graph approach that models the resource-time scheduling of the recursions of the algorithm. The problem of constructing the decoder architecture and optimizing it for high speed and low power is formulated in terms of the individual recursion patterns which together form a tile graph according to a tiling scheme. Using the tile-graph approach, optimized architectures are derived for the various forms of the sliding-window and parallel-window algorithms known in the literature. A proposed tiling scheme of the recursion patterns, called hybrid tiling, is shown to be particularly effective in reducing memory overhead of high-speed SISO-APP architectures. Simulations demonstrate that the proposed approach achieves savings in area and power in the range of 4.2%-53.1% over state of the art.

The long-term bit error rate (BER) performance of partial response maximum likelihood (PRML) system in a perpendicular magnetic recording (PMR) channel with thermal decay is studied. A thermal decay model based on the experimental data for three CoPtCr-SiO2 PMR media with different thermal stability is obtained. Then, the performance of PR1ML channel for the 16/17(0,6/6) and 128/130(0,16/8) codes as a run-length-limited (RLL) code is evaluated by computer simulation using the model. Moreover, the relationships between the elapsed time and the BER are obtained for the combinations of two RLL codes and three PMR media and their performances are compared. The results show that for the combination of the 128/130(0,16/8) code and the medium with the poorest thermal stability, the degradation with the elapsed time is noticeable compared with the other combinations

We define super codes as the serial concatenation of punctured convolutional codes. Puncturing pattern and interleaver design techniques are presented so as to maximise the free distance of the code and thus the performance at very low bit error rates (BER). The performance of super codes is compared to turbo codes. For BERs around 10/sup -5/, turbo codes perform 0.1, 0.4 and 0.6 dB better than super codes for rates 3/4, 1/2 and 1/3, respectively. However, due to the use of serial concatenation, super codes have an interleaver gain of N/sup -2/, compared to only N/sup -1/ for turbo codes and interleaver size N. Thus, super codes are expected to perform significantly better at BERs around 10/sup -10/.

Autoregressive (AR) model is commonly used in many areas of signal processing and is particularly significant in Electroencephalogram (EEG) related studies owing to its enhanced resolution, smoother spectra and its potential to be used on short segments of a signal. When fitting an AR model to an EEG multivariate system, selection of model order ( p) is of critical importance. While the lower model orders provide inadequate representation of the signal, higher orders drastically increase noise. Therefore, identification of optimum AR model order is an open challenge. Conventional methods for estimating model orders include Akaike Information Criterion (AIC), Final Prediction Error (FPE), Bayesian Information Criterion (BIC), and Hannan Quinn (HQ). In this paper, we show how these criteria fail to determine optimal order of multivariate EEG signals that undergo mandatory filtration to separate bands and noise. Consequently, we present a novel, yet simple and effective technique to find the AR optimum order to model such systems. Extensive application of proposed method on different EEG dataset indicates that the new method gives better signal estimation with minimum possible order; thereby, improving the accuracy of reconstructed signal at reduced computational cost.

Contact-type magnetoresistive (MR) inductive head designs and recording characteristics have been discussed in perpendicular magnetic recording. The shielded-type MR head offered a high D50 of 100–135 kFRPI, and the unshielded MR head had the largest output in a low density region. However when the recorded bit length became shorter than the shield-to-shield spacing, the shield had little effect and the output of the shielded head was no longer almost equal to that of the unshielded MR head. In result, the shielded-type and even unshielded-type MR heads have good high density recording characteristics in the head-to-medium contact state.

The perpendicularly oriented crystal structure conditions of electroless plated film for high density perpendicular recording media were investigated mainly by (002) plane rocking curves and cross sectional SEM observation. A highly oriented narrow dispersion of c‐axis orientation normal to the surface plane was consequently obtained, and the columnar structure of hcp crystal with ca. 0.05 μm diam was confirmed. As a result of preparing disks A and B, which were chemically plated from an optimum bath and a more improved bath, respectively, the recording characteristics with a combination of ring head were obtained as recording densities of for disk A and 25,000 FRPI for disk B, in spite of a relatively thick magnetic medium. The di‐pulse waveforms, which are characterized with perpendicular recording, were observed at low bit recording density. It was indicated by the above results that the chemically plated Co alloy films are applicable as a perpendicular recording medium.

The demagnetizing field near the top surface of a medium for perpendicular recording is compared with that of a medium for longitudinal recording. On the basis of both the result of this comparison and the output dependence on the medium thickness, a perpendicular medium with a large anisotropy field, H/sub k/spl perp// or (H/sub k/spl perp//-4/spl pi/M/sub s/), of over 10 or 5 kOe, respectively, is needed to obtain large output and high resolution. Exploiting this large anisotropy field, a composite perpendicular medium with a (Co/sub 72/Cr/sub 28/)/sub 78/Pt/sub 22/ layer over a Cr-rich Co/sub 72/Cr/sub 28/ layer was fabricated. By taking advantage of both the large anisotropy in the upper layer and the strong particle interaction in the lower layer, the composite medium with a merged MR head obtained an S/N of 24.0 dB at 200 kFCI and a D/sub 50/ of 205 kFCI, with a 50-nm magnetic spacing.

Recently, an efficient error performance scheme named as labeling diversity has been proposed in the literature for wireless local area network (WLAN). This paper proposes labeling diversity for 8-phase shift keying (8-PSK) modulation named as boosted space-time scheme. A pair of labeling map is opted for this scheme by maximizing the utility function in asymptotic coding gain under the supposition of error-free feedback. Furthermore, labeling diversity is employed with the space-time forward error correction (FEC) codes like convolution and turbo codes for point to point communication scenarios. Since non-line of sight (NLOS) point to point communication scenario is assumed therefore Rayleigh distributed channel model is selected which remains constant for single space-time block code (STBC). The STBC used in boosted space-time scheme is non-orthogonal therefore maximum likelihood (ML) soft de-mapper is employed at the receiver. The convergence of iterative decoding between soft de-mapper and decoder is analyzed by the extrinsic information transfer (EXIT) chart. The validity of the convergence of iterative decoding is also verified by Monte Carlo simulation. It is evident from the simulation results that bit error rate (BER) performance of boosted space-time scheme is superior to the conventional spacetime scheme under identical conditions. Moreover, the boosted space-time scheme with convolution code performs better for low signal-to-noise ratio (SNR) regime in comparison to boosted space-time scheme with turbo code. However, for high SNR regime, turbo coding scheme outperforms the convolution scheme. In addition, turbo coding scheme provides better BER performance to the convolution scheme under non-iterative decoding.

In this contribution, we consider the performance of multi-user transmitter pre-processing (MUTP) assisted interleaved division multiple access (IDMA) scheme for downlink (DL) communications. We realize MUTP using Zero Forcing (ZF)algorithm. In this paper, we consider turbo coded system to achieve high coding gain. We construct the turbo code using parallel concatenated convolutional encoder and certain bits are punctured periodically to achieve code rate ½. We construct the pre-processing matrix using channel state information of all the users. We evaluate the performance of such coded ZF-MUTP assisted IDMA system for realistic channel model based on SUI channel model specifications. From our simulations, we observe that the ZF-MUTP assisted multi-user IDMA system provides better performance in terms of bit error rate (BER) with low signal to noise ratio (SNR).