This thesis investigates the behavior of two digital communication systems based on Moving-Average Matched Filters (MAMF). In general, matched filters are instrumental in detecting signals corrupted by noise as they are designed to maximize the probability of detection of the transmitted signals. The MAMF represents a subset of the class of matched filters.

The two communication systems under investigation are the classical MAMF system and one of its modifications, the proposed MAMF system. In the traditional system the N-dimensional signal vector, which encodes the bit to be communicated, remains fixed throughout the whole communication process ( transmission and reception). In the proposed system the encoding N-dimensional signal vector is composed of K linearly independent basis vectors spanning a signal vector subspace of dimension M (= N /K). By combining these basis vectors in the receiver, any vector in the signal vector subspace can be formed in order to maximize the Output Signal-to-Noise Ratio (OSNR).

The relative measure of comparison for the two systems is the Signal-to-Noise Ratio Improvement (SNRI). The SNRI is the ratio of the OSNR, which is measured at the output of the receiver, to the Input Signal-to-Noise Ratio (ISNR), which is measured at the input of the receiver. Since the ISNR is fixed for a particular transmitted signal vector and noise characteristics, an attempt is made to maximize the SNRI by maximizing the OSNR.