Time and Frequency Crosstalk in Pulse-Modulated Systems

Consider a transmission medium in which many simultaneous messages travel in one single direction. Each message, consisting of coded on-off RF pulses (PCM), has its own carrier and occupies a separate frequency channel. This occurs, for example, in the proposed long distance waveguide communication system. 1 The transmitter is considered as a filter through which the pulses of a message are fed to the transmission medium and the receiver as a filter that selectively couples the transmission medium to a detector. The problem is to design these filters in such a way that the communication medium handles information at the highest possible rate. This means that the channels must be close to each other, providing high frequency occupancy, and that each message must be made of pulses close to each other, providing high time occupancy. In other words, we want to minimize the product of channel spacing and pulse spacing. What prevents us from making this product arbitrarily small is that, in general, a reduction of pulse and channel spacings implies an increase of time and frequency crosstalk, and these values are fixed by other considerations: the signal-to-noise level and the probability of errors allowed in the system. We shall see how they enter the picture. The detector of each receiver reconstructs a message by deciding 951