Digital Modulation

In almost all of today's second- generation wireless communi cations systems, information is transformed into digital 1s and 0s and modulated on the transmitted carrier using complex digital modulation methods. These methods are substantially different from traditional first-generation analog systems, and testing methods are different as well.

Because digital modulation implies that the information to be transmitted already is in a digital form, the first function of the system is to convert the analog voice signals to digital. The basic conversion method is to use a coder/decoder (codec), which samples the voice signal's level and sends a bit stream to indicate either absolute level or differential steps up or down. The human voice, however, generally requires a data rate of at least 32kb/s, which would use too much bandwidth for most wireless systems.

Instead, systems use complex vocoders, which can represent particular human voice sounds with bit sequences, essentially using a "codebook." With this method, you can achieve data rates below 13kb/s, which are perfect for today's requirements. Each wireless system generally uses its own vocoder. CdmaOne, for example, uses the QSELP vocoder, and IS-136 first used VSELP, followed by ACELP, a higher-quality vocoder. Hence, the human voice is turned into 1s and 0s that are ready to be modulated on an RF carrier.

Shifting Methods In digital, because all of the information is stored in 1s and 0s, any method of shifting from one state to another can be used to signify a 1 or a 0. There are three components of an RF carrier that can be "keyed" this way: amplitude, frequency and phase.

Amplitude-shift keying (ASK) simply varies the amplitude of the carrier between two states, one representing a 1 and the other representing a 0. This method is susceptible to the same problems normal AM radios are. Speed also is a problem. Thus, ASK is not practical for advanced systems. Frequency-shift keying, where the frequency is shifted between two states to represent a 1 or 0, is used extensively in analog system signaling.

Phase modulation is the system best suited for today's wireless networks. The phase of the carrier is shifted depending on the data to be sent. The simplest form of this is binary phase-shift keying, where two phase states represent either a 1 or a 0.

Phase can be represented with a circular diagram. In this diagram, the phase of the signal is represented by the angle around the circle, and the amplitude of the signal is represented as the magnitude away from the origin (or center) of the circle, with the center of the circle representing the least amount of amplitude. Most systems use constant amplitude phase modulation, meaning the amplitude of the signal stays constant during the "decision" points, changing only during phase transitions. Thus, each decision point has twocomponents: phase and amplitude. Deviations from the ideal vector for each decision point is the error vector magnitude, which is a combination of two sub-measurements: phase error and magnitude error.

The three wireless systems use three types of phase modulation. GSM systems use Gaussian minimum-shift keying (GMSK); IS-136 systems use pi/4 differential quadrature phase-shift keying (pi/4 DQPSK); and CDMA systems use quadrature phase-shift keying (QPSK) and offset quadrature phase-shift keying (O-QPSK) for the forward and reverse links, respectively.

GSM GMSK frequency-shift keying results in carrier phase changes. The technique allows the carrier's magnitude to remain constant, with only the phase seeming to shift.

The constellation diagram for a GMSK signal resembles a circle at a constant amplitude around the origin. The key measurement of this type of modulation, therefore, is the phase error because there is no magnitude error. Many test instruments can plot this phase versus bits, as graphic representations often paint a picture much more clearly. In this way, the technician can see quickly whether the modulation is passing or not.

IS-136 There are eight possible phase positions, which are generated by four phase changes: 3Pi/4, Pi/4, -3Pi/4 and -Pi/4.

These four phase changes represent 2 bits each (e.g. 00, 11, 01, 10); thus, any sequence of 1s and 0s can be sent. The demodulator looks for these phase changes, as opposed to absolute phase position. The constellation diagram of a Pi/4 DQPSK signal has eight decision points, each four possible transitions that will represent one of four 2-bit symbols (00, 11, 01, 10).

CDMA CdmaOne uses QPSK on the forward channel and O-QPSK on the reverse channel. Like Pi/4 DQPSK, each symbol represents 2 bits; however, QPSK uses absolute phase position to represent the symbols. Thus, there are four phase decision points, and when transitioning from one state to another, it is possible to pass through the circle's origin, indicating minimum magnitude.

O-QPSK is used in the reverse channel to prevent transitions through the origin. Consider the components that make up any particular vector on the constellation diagram as X and Y components. Normally, both of these components would transition simultaneously, causing the vector to move through the origin. In O-QPSK, one component is delayed, so the vector will move down first, and then over, thus avoiding moving through the origin, and simplifying the radio's design. A constellation diagram will show the accuracy of the modulation.

If the problem is related to phase, the constellation will spread around the circle. If the problem is related to magnitude, the constellation will spread away from the origin. These graphical troubleshooting clues help isolate the problem much faster than by simply looking at numbers.