Data-dependent Noise Research Articles

Information rates of multidimensional front ends for digital storage channels with data-dependent transition noise

This paper presents a simulation-based method for the computation of the information rates of recently proposed multidimensional front ends for digital storage channels with transition noise. First, we propose an algorithm, based on linear prediction and state reduction techniques, that extends recent work on the information rates of magnetic recording channels affected by colored Gaussian thermal noise, intersymbol interference, and signal-dependent transition noise. Following a previous study on statistical sufficiency, we apply this algorithm to compute the information rates of digital storage systems featuring a multidimensional detection front end. The resulting information rates indicate that significant gains may be achievable by multidimensional signal processing techniques in transition-noise-limited digital storage channels

Multiuser Detection for Square-Law Receiver Under Gaussian Channel Noise With Applications to Fiber-Optic Communications

In fiber-optic wavelength-division multiplexing (WDM) communications systems, the nonlinearity of the channel and the limitations of optical filters result in received channel signals that are correlated. Furthermore, the square-law nature of the detector converts the additive fiber amplifier noise to a data-dependent asymmetric and non-Gaussian noise. The optimal multiuser detector for a Gaussian approximation is compared to the actual statistics and found to give a close performance estimate. Through bounds on the error probability and the asymptotic multiuser efficiency, the optimal multiuser detector based on a Gaussian approximation is found to improve significantly performance in highly correlated data cases. This result is applied to a practical optical WDM system.

Robustness evaluation of a minimal RBF neural network for nonlinear-data-storage-channel equalisation

The authors present a performance-robustness evaluation of the recently developed minimal resource allocation network (MRAN) for equalisation in highly nonlinear magnetic recording channels in disc storage systems. Unlike communication systems, equalisation of signals in these channels is a difficult problem, as they are corrupted by data-dependent noise and highly nonlinear distortions. Nair and Moon (1997) have proposed a maximum signal to distortion ratio (MSDR) equaliser for data storage channels, which uses a specially designed neural network, where all the parameters of the neural network are determined theoretically, based on the exact knowledge of the channel model parameters. In the present paper, the performance of the MSDR equaliser is compared with that of the MRAN equaliser using a magnetic recording channel model, under conditions that include variations in partial erasure, jitter, width and noise power, as well as model mismatch. Results from the study indicate that the less complex MRAN equaliser gives consistently better performance robustness than the MSDR equaliser in terms of signal to distortion ratios (SDRs).

Performance analysis of extended OQPSK carrier phase recovery algorithm in satellite channel

This paper proposes a new extended decision-directed decision-estimated phase recovery algorithm based on maximum likelihood parameter estimation for OQPSK (offset QPSK). In this scheme, compared to the conventional one, the data dependent noise of the phase recovery loop is reduced by inserting a filter with 2 taps to the in-phase and quadrature channel respectively before the phase detector. The proposed scheme is compared to the conventional QPSK scheme with respect to BER (bit error rate) and phase error for roll-off factors of the baseband filter, output backoffs of the nonlinear satellite channel and loop bandwidth.

Baud sampling bit synchroniser for channels with data dependent noise

A bit synchronisation algorithm for channels with data dependent noise which operates with one sample per symbol is presented. The algorithm uses the same information as the Mueller and Muller (M&M) algorithm, and is optimised for operation with data dependent noise. The performance is derived and it is shown that significant improvements over the M&M algorithm can be obtained in practical optical channels.

Data storage channel equalization using neural networks

Unlike in many communication channels, the read signals in thin-film magnetic recording channels are corrupted by non-Gaussian, data-dependent noise and nonlinear distortions. In this work we use feedforward neural networks-a multilayer perceptron and its simplified variations-to equalize these signals. We demonstrate that they improve the performance of data recovery schemes in comparison with conventional equalizers. The variations of the MLP equalizer are suitable for the low complexity VLSI implementation required in data storage systems. We also present a novel training criterion designed to reduce the probability of error for the recovered digital data. The results were obtained both from experimental data and from a software recording channel simulator using thin-film disk and magnetoresistive head models.

Modeling thin-film storage channels

A straightforward method for modeling nonlinearity and data-dependent (transition-dependent) media noise in a thin-film storage channel is presented. The least-squares approach for measuring the channel linear pulse response is described, and this FIR (finite impulse response) channel model is contrasted with a nonlinear moving average modeling technique described in terms of a random-access memory used as a finite state machine (FSM). It is shown that the channel nonlinearity is largely trailing in time, and performance figures of modeling accuracy for the channel FSM are presented. The nonlinear distortion increased dramatically at higher data rates for the thin-film disk examined. In addition, the variation of the channel noise power according to the flux transition density is described. A jitter noise effect is identified. A FSM modeling technique for identifying this noise power variation as a function of the channel input data pattern is given. Finally, it is shown how to combine these two modeling techniques to create arbitrarily long, representative read signals for the thin-film disk. By using these compact channel models, one can easily test new coding or equalization techniques. >