All communications are degraded by naturally occurring noise in both the environment and hardware. This noise interferes with the transmitted bit energy, and if it is strong enough, it may cause the receiver to make an error in deciding which bit (0 or 1) was transmitted. Because noise is random, a digital receiver's performance can only be described in terms of the probability that it will make a bit error, given the transmitted bit energy and the noise environment.
Bit errors in digital communication systems can be caused by unwanted noise in the hardware and environment. In this figure, the top graph represents the bit stream and the middle graph represents the noise. The bottom graph combines the two, and shows how noise can cause an error in the bit stream. |
By plotting the bit-error probability against the bit-signal-to-noise ratio, designers can get an idea of what kind of error performance to expect for a given transmitted bit energy and noise environment. These so-called bit-error-rate curves are commonly used to compare different digital communication techniques. For example, for transmitting digitized voice, a bit-error probability of 10-3 (1 error per 1000 bits, on average) is generally considered sufficient. When choosing between two possible transmission techniques, a designer might use bit-error-rate curves to determine which technique would require the smallest transmitted bit energy to achieve the desired bit-error probability of 10-3.
Power spectral density is another common characteristic used to compare the bandwidth of communication signals. Power spectral density curves show the distribution of signal power in the frequency domain. Often, a system requires that multiple user signals coexist within a given bandwidth. If the system uses frequency-division multiplex techniques, each signal, or channel, is assigned a unique center-transmission frequency, and channels are placed adjacent to each other.
As the spacing of these signals gets closer, their power spectra start to overlap, and power from one signal spills into another. This phenomenon, known as adjacent channel interference, increases the bit-error rate (which, in practical terms, means a more-complex receiver or more bit-signal-to-noise power is needed to accommodate the interference). A limited number of signals can be packed within a specified frequency band, given a maximum allowable signal-to-noise degradation; this number is known as the channel packing efficiency. When packing signals, communication system designers need to trade channel spacing (in frequency) with performance degradation caused by adjacent channel interference.