The paper describes two new detection processes suitable for QAM signals that have 16 or more levels (possible data symbol values) and that have been subjected to linear distortion, causing severe intersymbol interference in the received signal. The detectors are developments of the pseudobinary systems previously described, and are designed to operate in the presence of severe intersymbol interference. Results of computer simulation tests are presented, comparing the tolerances to additive white Gaussian noise of various arrangements of each detector with those of more conventional detectors. The tests have been carried out on a model of a 19 200 bit/s synchronous serial data transmission system, operating with a 64-level QAM signal that is transmitted, in turn, over two different telephone circuits.

This paper studies various iterative detection processes for use in a synchronous serial data-transmission system where the signalelements are transmitted in orthogonal groups. The methods of operation of the different detection processes are first described, and then results of computer simulation tests are presented to compare the tolerances of the detection processes to additive white Gaussian noise, for various values of the sampled impulse response of the channel. It is shown that a good tolerance to noise can be achieved by detection processes that do not require an excessive number of sequential operations or unduly complex equipment.

A theoretical analysis is presented of an adaptive detector in a synchronous serial data-transmission system. A binary-coded signal is transmitted over a baseband channel, which introduces slowly time-varying signal distortion. The received signal is sampled and each signal element is detected by means of an iterative process that uses some or all sample values dependent on that element. The intersymbol interference of the signal element in the following elements is then canceled, before the detection of the next element, and so on. Immediately after the detection of each signal element, the channel impulse response is estimated and the detector is adjusted appropriately, so that it follows the variations in the channel transmission characteristics. No training signal is required except at the start of transmission. The tolerance to additive white Gaussian noise is evaluated for different arrangements of the basic system and for various values of the signal distortion in transmission, assuming the detector to be correctly matched to the channel.