Performance analysis of convolutional codes over non-stationary-noise channel

This paper is focused on the performance of convolutional codes with bit interleaving in presence of non-stationary noise due to time-varying interference scenario. Two different decoding schemes are considered based on the knowledge of the instantaneous or the average noise power at the receiver. Analytical performance on average error probability has been derived under the assumption that interference can be modelled as Gaussian noise with random power. More specifically, the closed-form is for central 𝒳E2 noise power with any degree of freedom, while upper bounds are derived for either static or fading channels. Simulations validate the analytical relationships and show that there is a definite advantage in exploiting the instantaneous noise power when the non-stationary interference can be modelled as time-varying power with heavily skewed distributions (e.g., lognormal).