Robust turbo decoding in impulse noise channels

Abstract

Powerline communications, an enabling technology for data networking alternative, susceptible to impulse noise such that it is likely to miss an assured quality of service. Turbo coding has long served as an essential tool to overcome this obstacle. However, this comes at the cost of estimating the statistics of impulse noise, which is generally not time-invariant and may change rapidly over time, further increasing receiver complexity. This study proposes clipping on trellis to avoid this problem. A corresponding decoder metric dictated by the clipping threshold, which is a function of the probability of guessing for the unknown impulse arrival probability, was properly formulated to leverage the coding gain. Regardless of the impulse noise model assumed, the simulation results in this study, in terms of bit error probability, show that the proposed decoding scheme is robust to impulse and performs fairly close to its sophisticated counterpart, which however fully exploits the statistics of the impulse noise.