Skip to Main Content
A well-known means of efficiently transmitting information over the continuous white Gaussian channel involves the encoding of successive blocks of data into sequences of binary digits (called code words). Efficient decoding of these sequences in turn necessitates a knowledge of the instants in time at which one code word ends and the succeeding word begins. This paper presents a method for obtaining this synchronization which neither decreases the channel capacity nor increases the complexity of the encoding equipment. The method is to select, from the many encodings which are equally good for purposes of synchronous operation, that encoding for which the maximum absolute value of the correlation po between any code word and any sequence formed from the overlap of two code words is a minimum. Thus a large correlation is observed only in the synchronous phase position. This technique is applied to an important class of block codes, the binary orthogonal codes. An algorithm for constructing these codes with the desired self-synchronizing properties is presented, and upper bounds on the value of po are thereby established.