Data transmission at 19.2 kbit/s over telephone circuits

Detection of coded and distorted QAM signals

“…For data transmission systems operating at data-symbol rates greater than 2400 symbol/s and with telephone circuits which introduce typical levels of signal distortion (such as telephone circuit 2), the use of the equipment filters-l result in excessive signal distortion [42]. This can be verified by inspecting the attenuation and group-delay characteristics of the equipment filters-l and telephone circuit 2.…”

“…now determines the appropriate data symbol s, according to Fig. 8.2 [42]. As in the case of the 16-1evel QAM signal in Chapter 3, the fIrst two binary digits in any binary coded number determine the quadrant containing s, • The remaining four digits determine the position of s, in the quadrant, the latter digits in any quadrant are the same as those in the all-positive quadrant, if this is rotated to coincide with the given quadrant [14,42,85].…”

“…This can be verified by inspecting the attenuation and group-delay characteristics of the equipment filters-l and telephone circuit 2. Thus, a reduction in signal distortion, accompanied by the corresponding improvement in tolerance to noise, can be achieved by increasing the bandwidth of these filters [39,42]. Thus, wider-band equipment filters (called here equipment filters-3) are used for the transmission systems operating at rates of 16000 and 19200 bit/so These systems operate with a data-symbol rate of 3200 symboVs.…”

“…When the received signal is sampled at the Nyquist rate a noise-whitening matched filter is no longer required. It is now sufficient to use in its place, a low pass filter with a rectangular (or nearly rectangular) transfer function and a cut-off frequency equal to the half of the sampling rate followed by the sampler and then an adaptive linear feedforward transversal filter similar to that employed in the optimum (conventional) nonlinear equalizer (see the previous section) [26,41,42). …”

Detection process of a 9600 bit/s serial modem for HF radio links

“…An alternative and more attractive approach is to use an adaptive noise-whitening matched filter [6][7][8] or its equivalent [18], together with a near-maximum-likelihood detector developed from the Viterbi algorithm. The amount of storage required here and the number of operations per received data symbol are greatly reduced relative to the Viterbi algorithm, but without unduly degrading the tolerance of the detector to noise [4,[17][18][19][20][21][22][23][24]. With the ideal arrangement of the noise-whitened matched filter [18], a wide range of simple near-maximumlikelihood detection processes can now be used successfully [17][18][19][20][21][22][23][24].…”