A data-derived reference signal technique for adaptive arrays

Abstract

LMS (least mean square) adaptive arrays [I] are nseful for protecting conlmunication systems from interference. An LMS array uses a reference signal to discriminate hetween signals. It forms a beam on any signal correlated with the reference signal and nulls any signal uncorrelated with it. If a reference signal can be obtained that is correlated with a desired communication signal and uncorrelated with interference, the LMS array will track the desired signal and null the interference. The difficolt part of using an LMS array in a communication system is of conrse the problem of obtaining a suitable reference signal. In general, there is no way to obtain a reference signal correlated with a desired signal in real time for an arbitrary type of desired signal modulation. However, if the desired signal modulation can be chosen in an special way, then it may he possihle to introduce sufficient structure into the desired signal that reference signal.generation is possihle. One such technique was described by Compton [2] for binary phase-shift keyed (BPSK) signals, and a similar hut more rohust techniqne was presented by Winters [3] for quadrature phase-shift keying (QPSK). In both of these methods, psendonoise (PN) codes were included in the transmitted modnlation, in addition to the data. The presence of the PN codes makes it possible to obtain a reference signal. In this paper we describe a new method for generating a reference signal for an LMS array. This method differs fundamentally from those in [Z] and [3] in that no PN codes are used in the transmitted signal. Instead, the reference signal is based on a predicted hit stream obtained from the data modulation itself. In this paper we describe the we of this method with noncoherent binary frequency shift keyed (FSK) modulation, hut it appears that the same method can also he used with other types of digital modulation. The method presented here was first studied with computer sinlulations by Hudson [4] and then was experimentally verified by Ganz [5]. Ganz has also evaluated the theoretical performance of adaptive arrays with FSK modulation [6].