REPORT CHAPTER 2: FORMATTING AND BASEBAND MODULATION 2.

1 Baseband Systems

In Formatting, sources of information are converted into the sequences of binary digits. The sources of information consist of Digital information, Textual information and Analog information (Fig 2.2): Data already in digital format would bypass the formatting function. Data in text is transformed into binary digit Analog information is formatted using 3 processes: Sampling, Quantization, and Coding.

Then these digits are to be transmitted through the Baseband channel, such as a pair of wires or a coaxial cable, after they are transformed into the waveforms in a block labeled Pulse modulate that are compatible with the channel. For baseband channel, the waveforms are pulses. After transmission through the channel, pulses are demodulated, formatted to recover as the sources of information. 2.2 Formatting textual data (Character coding): Character coding is the step that transforms text into binary digits. Character can be encode with one of several standard formats: the American Standard Code for Information Interchange (ASCII), the Extended Binary Coded Decimal Interchange Code (EBCDIC), Baudot, and Hollerith. 2.3 Messages, Characters, and Symbols:

When digitally transmitted, the characters are first encoded into a sequence of bits, called a bit stream or baseband signal. Group of k bits forms new digits, or symbols, from a finite symbol set or alphabet of M = 2k such symbols. 2.3.1 Example of Messages, Characters, and Symbols. The figure below is an example of encoding the word THINK by 6 bit ASCII character coding.

In figure 2.5a, M is chosen to be 8 (k = 3). This resulting 10 numbers represent 10 octal symbols to be transmitted. The transmitter must have 8 types of waveform ( where i = 1, 2,...,8). This is similar to the way of coding in figure 2.5b but in this case, M = 32 (k = 5). 2.4 Formatting analog information. If the information is analog, it must first be transformed into a digital format. 2.4.1 The sampling theorem.

The most popular way implementing the sampling process is sample-to-hold operation. In this operation, a switch and storage mechanism form a sequence of the continuous input waveform. The ouput of the sampling process is called pulse amplitude modulation (PAM). The analog waveform can be approximately retrived from a PAM waveform by simple low-pass filtering. The sampling theorem (also known as the uniform sampling theorem): A bandlimitted signal having no spectral components above fm hertz can be determined uniquely by values sampled at uniform intervals of (2.1) (2.2) where fs = 1/Ts. The equation 2.2 is known as the Nyquist criterion. The sampling rate fs = 2fm is also called the Nyquist rate. The Nyquist criterion is a theorically sufficient condition to allow and analog signal to be reconstructed completely from a set of uniformly spaced discrete-time samples. 2.4.1.1 Impulse sampling. Within the original bandwidth, the spectrum Xs(f) of the sampled signal xs(t) is, to within a constant factor (1/Ts), exactly the same as that of x(t). In addition, the spectrum repeats itself periodically in frequency every fs (we choose fs = 2fm in this case so that the Nyquist is satisfied) hertz (as seen in fig 2.6b and 2.6f). Using the frequency convolution property of the Fourier transform, we can find the transform Xs(f) of the sampled waveform is expressed as: (2.8)

In this case fs = 2fm, each spectrum replicate from each of its neighbors seperately, and the analog waveform can theorically be completely recovered from the samples. If fs > 2fm, the replication moves farther apart in frequency. A typical low-pass filter characteristic used to separate the baseband spectrum from those at higher frequency, as shown in Fig 2.7a. If fs < 2fm, the replication will overlap, as shown in Fig 2.7b, and some information will lost. The phenomenon is called aliasing as the result of undersampling (sampling at two low a rate). Perfectly bandlimitted signals do not occur in nature; thus, realizable signals always contain some aliasing.

2.4.1.2 Natural sampling.

Using the frequency translation property of the Fourier transform, we solve for Xs(f) as follows: (2.14)

Similarly to the impulse sampling case, Xs(f) is a replication of X(f), periodically repeated in frequency every fs hertz (we choose fs = 2fm in this case so that the Nyquist is satisfied), as seen in fig 2.8b and 2.8f). When the pulse width, T, approaches zero, cn approaches 1/Ts for all n, and equation 2.14 converges to equation 2.8. 2.4.1.3 Sample and-Hold operator. The sample-and-hold operator can be described by the convolution of the sampled pulse train, [x(t) x(t)], shown in Fig 2.6e, with a unity amplitude rectangular pulse p(t) of pulse width Ts. This convolution results in the flattop (nh bng phng) sampled sequence:

[ [

] ]

(2.15)

The Fourier transform, Xs(f) is expressed as: [ ] (2.16)

Here P(f) is of the form Tssinc f Ts. The effect of this product operation results in a spectrum similar to the natural sampled example. The most obvious effect of hold operation is the significant attenuation of the higher-frequency spectral replicates (compare Fig 2.8f to Fig 2.6f). A secondary effect of the hold operation is the nonuniform spectral gain P(f) applied to the disired baseband spectrum shown in equation 2.16. 2.4.2 Aliasing. As mentioned in section 2.4.1.1, aliasing is the result of undersampling. The aliasing spectral components represent ambiguous data that appear in the frequency band between (fs - fm) and fm, as shown in Fig 2.9

Fig 2.11 and Fig 2.12 illustrate two ways of aliasing using antialiasing filters.

In figure 2.11, the analog signal is prefiltered (filter before sampling) that the new maximum frequency, , is reduced to fs/2 or less. Thus there are no aliased components seen in figure 2.11b. Eliminating the aliasing terms prior to sampling (before sampling) is good engineering practice. In figure 2.12, the aliased components are removed by postfiltering after sampling; the filter cutoff frequency, , need to be less than (fs - fm). We use this technique for signals whose structure is well known. Notice that both technique used above will result in a loss of some of the signal information. Thus, the sample rate, cutoff bandwidth, and filters type selected for a particular signal bandwidth are all interrelated (lin h qua li). 2.4.3 Why oversample? Oversampling is the most economic solution for the task of transforming an analog signal to a digital one, and vice versa.

(Em khng hiu phn ny)

2.5 Sources of corruption 2.5.1.1 Quantization noise (n lng t) The distortion inherent (vn c) in quantization is a round-off or truncation (ct bt, gim bt) error. This distortion is referred as quantization noise. These amount of such noise is inversely propotional to the number of levels employed (c dng) in the quantization process. 2.5.1.2 Quantizer saturation (bo ha lng t) The quantizer (or analog-to-digital converter) have the task of approximating the continuous range of inputs for which the difference between the input and output is small is called the operating range of the converter. If the input exceeds this range and the difference becomes large, we say that the converter is operating in saturation. Saturation errors, being large, are more objectionable (gy kh chu) than quantizing noise. Generall, saturation is avoided by the use of automatic gain control (AGC), which effectively extends the operating range of the converter. 2.5.1.3 Timing Jitter. (C ging cho em phn ny) Jitter is the undesired deviation ( lch) from true periodicity of an assumed periodic signal in electronics and telecommunications, often in relation to a reference clock source. Jitter may be observed in characteristics such as the frequency of successive pulses, the signal amplitude, or phase of periodic signals. Jitter is a significant, and usually undesired, factor in the design of almost all communications links (e.g., USB, PCI-e, SATA, OC-48). In clock recovery applictions it is called timing jitter (adapted from Wikipedia).

Tri pha v rung pha (Wander v Jitter) l nhng bin i v pha ca tn hiu s thu c so vi v tr l tng ca chng: - Nhng bin i pha c tn s hn hn hoc bng 10Hz gi l rung pha. - Nhng bin i pha c tn s nh hn 10Hz gi l tri pha If there is slightly jitter in the position of the sample, the sampling is no longer uniform; therefore, reconstruction of the signal using the sampling theorem is not precise. 2.5.2 Channel effects 2.5.2.1 Channel noise Thermal noise, interference from other users and interference from circuit switching transients can cause errors the pulses carrying the digitized samples. Channel-induced errors can degrade (gim) the reconstructed signal quality quite quickly. This rapid degradation is called a threshold effect. If the channel noise is small, there will be no problem detecting the presence of the waveform. In this case, only quantization noise presents in the reconstruction. 2.5.2.2 Intersymbol Interference The channel is always bandlimitted. A band limitted chanel disperses (phn tn) or spreads a pulse waveform passing through it. When the channel bandwidth is close to the signal bandwidth, the spreading will exceed a symbol duration and cause signal pulses to overlap. This overlapping is called intersymbol interference (ISI). ISI causes system degradation (higher error rates); it is a particularly insidious (m thm, khng nhn thy c) form of interference because raising the signal power to overcome the interference will not always improve the error performance. 2.5.3 Signal-to-noise ratio for Quantized pulses. Fig 2.15 illustrates an L-level linear quantizer for an analog signal; the step size between quantization levels, called the quantile interval, is denoted q volts. When the quantization levels are uniformly distributed over the full range, the quantizer is called a uniform or linear quantizer. The degradation of the signal due to quantization is limmited to half a quantile interval q/2 volts. The quantizer error variance is found to be (2.18a)

(2.18b)

Where p(e) = 1/q is the uniform probability density function of the quantization. The variance 2 corresponds to the average quantization noise power.

The peak power of the analog signal (normalized to 1) can be expressed as ( ) ( ) (2.19)

Equation 2.18 and equantion 2.19 combined yield the ratio of peak signal power to average quantization noise power, assuming that there are no errors due to ISI or channel noise: ( ) (2.20)

In the limit, as L , the signal approches the PAM format (without quantization), and the signal-to-quantization noise ratio is infinite. 2.6 Pulse code modulation. Pulse code modulation (PCM) is the name given to the class of baseband signals obtained from the quantized PAM signals by encoding each quantized sample into a digital word. Assume that an analog signal x(t) is limited in its excursions to the range -4 to +4V. The step size between quantization levels is 1V. Thus, eight quantization levels are employed; these are located at -3.5, -2.5, . . . , +3.5V. We assign the code number 0 to the level at -3.5V, 1 at -2.5V, and so on until 7 is assigned to the level at 3.5V. Each code number has its representation in binary arithmetic, ranging from 000 to 111 (Fig 2.16 for L = 8). The choice of voltage levels is guided by two constraints: The quantile interval between the levels should be equal

Convenient for the levels to be symmetrical about zero.

If we increase the number of levels, L, the quantization noise will reduced . In a real-time communication system, the messages must not be delayed. Hence, the transmission time for each sample must be the samle, regardless (cho d) how many bits represent the sample. Hence when there are more bits per sample, the bits must move faste and data rate is that increased. The difference between PCM and a PCM waveform is that the former represents a bit sequence, and the latter represents a particular waveform coveyance of that sequence. 2.7 Uniform and nonuniform quantization. From the equantion 2.18b, we see that the quantization noise quantization depends on the step size (size of the quantile interval). When the step are uniform in size, the quantization is known as uniforrm quantization. Such a system is watseful for weak signal (such as speech signal) because many of the the quantizing steps would rarely be used. In the system using equally spaced quantization levels, the quantization noise (SNR) is worse for low-level signals than for high-level signals. Nonuniform quantization can provide fine quantization of the weak signal and coarse (th, khng mn) quantization of the strong signals. Thus, in the case of nonuniform quantization, quantization noise can be made proportional to signal size. Nonuniform quantization can be used to make the SNR a constant for all signals within the input range. Fig 2.18 below compares the quantization of a strong versus a weak signal for uniform and nonuniform quantization:

2.7.1 Nonuniform Quantization.

We achive nonuniform quantization by using a nonuniform quantizer characteristic (Fig 2.19a) or by distorting the original signal with a logarithmic compression characteristic (Fig 2.19b), and then using a uniform quantizer. After compression, the distorted signal is used as the input to a uniform quantizer characteristic, as shown in Fig 2.19c. At the receiver, an inverse compression,

called expansion, is applied so that the overall transmission is not distorted. The processing pair (compression and expansion) is usually referred to as companding (p dn). 2.7.2 Companding characteristic. Today, most PCM systems use a linear approximation to the logarithmic compression characteristic. In North America, a -law compression characteristic is used [ ( | | )] (2.22)

Where { And where is a positive constant, x and y represent input and output voltages, and xmax and ymax are the maximum positive excursions of the input and output voltages, respectively. The compression characteristic is shown in Fig 2.20a. The standard value for is 255, and = 0 corresponds to linear amplification (uniform quantization).

In Europe, the A-law is mainly used. Its characteristic is defined as | | [ | | ] | | | | (2.22)

{ Where A is a positive constant. Its characteristic is shown in Fig 2.20b. The standard value for A is 87.6.

2.8 Baseband transmission. 2.8.1 Waveform representation of binary digits. We will represent the binary digits with electrical pulse in order to transmit them through a baseband channel (Fig 2.21).

Fig 2.21a shows the 4-bit codeword representation of each quantized sample. In Fig 2.21b, each binary one is represented by a pulse and each binary zero is represented by the absence of pulse. At the receiver, a determination must be made as to the presence or absence of a pulse in each bit time slot. Detecting the presence of a pulse seems to be a function of the receiver pulse energy. Thus there is an advantage in making the pulse width T in Fig 2.21b as wide as possible. When the pulse width equal to the bit time T, we have the waveform as shown in Fig 2.21c with two levels (upper voltage level represents binary one whereas lower voltage level represents binary zero). 2.8.2 PCM waveform types

(phn 2.8.2 em vn cha hiu v my dng sng trang 87 v cch din t bng li cho nhng dng sng trong trang 88) 2.8.3 Spectral Attributes (thuc tnh) of PCM waveforms.

Fig 2.23 shows the spectral characteristics of some of the most popular PCM waveforms. The figure plots power spectral density in watts/herzt versus normalized bandwidth, WT (often referred to as the time-bandwidth product of the signal). The units of normalized bandwidth are herzt/(pulse/s) or herzt/(symbol/s) (on sau em khng hiu) 2.9 Corelative coding.

Sifting property Attenuation Aliasing Postfiltering Ambiguous Fidelity Construe Quantization noise Quantizer saturation Round-off Coarse

S nh bt, s yu i Khng r rng, m h S chnh xc St, gn ging S lm trn Th, khng mn

Adj N V

N Adj