Detection Using a Noise Estimation-Cancellation Technique

This paper introduces a novel class of detectors, which has a symbol-by-symbol, multisample, transversal filter-based structure. The essence of the proposed detectors is based on a noise estimationcancellation technique. By exploiting the noise correlation among the received samples, a linear minimum mean-squared estimate of the existing noise is made and subsequently cancelled. As an illustration of the detection principle, partial response signaling systems of the form
1 \pm D^{m}
are considered. The effects of the system bandwidth, the number of exploited samples and filtering allocation, on the system performance are thoroughly examined. It is shown that with optimum filtering allocation, the proposed detectors, while having simple structures, give substantial performance gains over conventional one-sample detectors.