cascade along the line and at the receiver, as illustrated in Fig. 3-7. These repeaters amplify and “clean up” the signal periodically. If the signal were analog instead of digital, only linear amplifiers with appropriate filters could be used, since relative amplitude values would need to be preserved. In this case, in-band distortion would accumulate from linear repeater to linear repeater. This is one of the disadvantages of analog signaling. However, with digital signaling, nonlinear processing can be used to regenerate a “noise-free” digital signal. This type of nonlinear processing is called a regenerative repeater. A simplified block diagram of a regenerative repeater for unipolar NRZ signaling is shown in Fig. 3-19. The amplifying filter increases the amplitude of the low-level input signal to a level that is compatible with the remaining circuitry and filters the signal in such a way as to minimize the effects of channel noise and ISI. (The filter that reduces ISI is called an equalizing filter and is discussed in Sec. 3-6.) The bit synchronizer generates a clocking signal at the bit rate that is synchronized, so that the amplified distorted signal is sampled at a point where the eye opening is a maxi¬mum. (Bit synchronizers are discussed in detail in the next section.) For each clock pulse, the sample-and-hold circuit produces the sample value that is held (for Tb, a 1-bit interval) until the next clock pulse occurs. The comparator produces a high-level output only when the sample value is larger than the threshold level VT. The latter is usually selected to be one-half the expected peak-to-peak variation of the sample values.'1' If the input noise is small and there is negligible ISI, the comparator output will be high only when there is a binary 1 (i.e., a high level) on the corrupted unipolar NRZ line code at the input to the repeater. The comparator— a threshold apparatus—acts as a decision-making device. Thus, the unipolar NRZ line code is regenerated “noise free,” except for bit errors that are caused when the input noise and ISI alter the sample values sufficiently so that the sample values occur on the wrong side of VT. Chapter 7 shows how the probability of bit error is influenced by the SNR at the input to the repeater, by the filter that is used, and by the value of VT that is selected.*
In long-distance digital communication systems, many repeaters may be used in cascade, as shown in Fig. 3-7. Of course, the spacing between the repeaters is governed by the path loss of the transmission medium and the amount of noise that is added. A repeater is required