Comsys Communication & Signal Processing Ltd.

Comsys Communications Signal Processing Ltd.

Ariel Zaltsman

Alexander Kleinerman

Ophir Shabtay

Alon Eran

Edo Navon

Yuval Bachar

An apparatus for and method of generating normalized soft decision information output from an inner decoder (i.e. equalizer) in a communications receiver. The invention is operative to normalize the soft decision information before it enters a soft outer decoder. The normalization is performed using a noise power estimate that is dynamically calculated in response to changing noise statistics on the channel. The normalized soft decision output is then applied to the soft outer decoder thus realizing maximum performance therefrom. The noise power estimate is derived from the training sequence and/or the data portion of the received signal. Both types of estimates are calculated. A binary or smoothly weighted average is calculated using both types of estimates. The weighting factor is determined based on one or more performance metrics, such as the Signal to Noise Ratio (SNR).

Normalization of equalizer soft output for channels with varying noise power

A novel and useful whitening matched filter (WMF) for use in a communications receiver. The WMF is constructed by cascading a matched filter and a noise-whitening filter. The response of the matched filter is derived from the time reversed complex conjugate of the channel impulse response. The whitening filter is derived by extracting the minimum phase portion of the mixed phase channel impulse response using homomorphic deconvolution. The whitening filter is implemented using either an FIR or IIR filter adapted to process the data received before and after the training sequence using a minimum phase system in a direction in time opposite to that of the direction of corresponding data sample processing performed by the equalizer.

Whitening matched filter for use in a communications receiver

A novel and useful apparatus for and method of interference reduction in a communications receiver. The invention first finds the channel estimate and the noise vector from the receive signal. An interference detector is then used to choose an appropriate compensation filter and select between adaptive and deterministic moving average models for noise prediction. After compensation filtering, a new channel estimate and noise vector is calculated. The new noise vector is used to determine noise whitening coefficients if the adaptive model is used. In the deterministic model case, the coefficients do not need to be calculated since they are already known, having been calculated a priori. The noise whitening coefficients are then used in an equalizer such as a Viterbi algorithm based equalizer.

Growing community of inventors

Raanana, Israel

Ariel Zaltsman