Improved detection in correlated impulsive noise using wavelet denoising

It is widely acknowledged that the effect of impulsive noise is a major source of performance degradation within a wide range of communication systems. This is due to the fact that non-Gaussian interference is neglected within the system design philosophy for reasons of complexity and tractability. Performance degradation is exacerbated further when the impulsive component is no longer independent identically distributed, but correlated. When this situation arises, the impulsive component can be masked by the underlying background noise process, and reside over many data samples, which in turn renders classical suppression techniques ineffective. We directly address the problem of signal detection in correlated impulsive noise using a novel 'denoising' technique in which significant performance gains are achieved with low-complexity.