eduzhai > Applied Sciences > Engineering >

PR-NN RNN-based Detection for Coded Partial-Response Channels

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 30 pages

Abstract: In this paper, we investigate the use of recurrent neural network (RNN)-baseddetection of magnetic recording channels with inter-symbol interference (ISI).We refer to the proposed detection method, which is intended for recordingchannels with partial-response equalization, as Partial-Response Neural Network(PR-NN). We train bi-directional gated recurrent units (bi-GRUs) to recover theISI channel inputs from noisy channel output sequences and evaluate the networkperformance when applied to continuous, streaming data. The computationalcomplexity of PR-NN during the evaluation process is comparable to that of aViterbi detector. The recording system on which the experiments were conducteduses a rate-2 3, (1,7) runlength-limited (RLL) code with an E2PR4partial-response channel target. Experimental results with ideal PR signalsshow that the performance of PR-NN detection approaches that of Viterbidetection in additive white gaussian noise (AWGN). Moreover, the PR-NN detectoroutperforms Viterbi detection and achieves the performance of Noise-PredictiveMaximum Likelihood (NPML) detection in additive colored noise (ACN) atdifferent channel densities. A PR-NN detector trained with both AWGN and ACNmaintains the performance observed under separate training. Similarly, whentrained with ACN corresponding to two different channel densities, PR-NNmaintains its performance at both densities. Experiments confirm that thisrobustness is consistent over a wide range of signal-to-noise ratios (SNRs).Finally, PR-NN displays robust performance when applied to a more realisticmagnetic recording channel with MMSE-equalized Lorentzian signals.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×