Signal Encoding Techniques

You really should read chapter 4.
- You learn interesting things like
  - Fig 4.1 page 105 which shows the available frequencies for given media.
  - Table 4.1 shows attenuation rates at various frequencies for different media.
- You really should read the descriptions of fiber and twisted pair.
- It takes the information from chapter 3 and shows how it can be used.
Chapter 5, Signal Encoding Techniques
5.1 Digital Data Digital Signals.
- The receiver must
  - Know when one bit ends and another bit begins
  - Know what represents a 0 and 1
- The receiver then samples each bit in the middle of the interval to determine the value of the bit.
- Stallings observes
  - Increase in the data rate increases the bit error rate (BER)
  - This is the probability that a bit is received in error.
  - Increasing the SNR decreases the BER
  - Increasing the bandwidth allows for an increase in data rate.
- The encoding scheme can also lead to improved performance.
- Stallings tells us that we evaluate encoding schemes using the following criteria:
  - Signal spectrum:
    - DC components should be avoided if possible.
    - The power should be concentrated in the middle of the bandwidth
  - Clocking:
    - Allows the user to detect the beginning of a bit.
    - Expensive to provide a separate clock signal.
    - Nice to include the clock signal in the data.
  - Error detection
    - The earlier an error is detected, the less expensive.
    - If the physical layer can provide this it is good.
  - Signal Interference and noise immunity.
  - Cost and complexity
    - Some encoding rates require signal rates that is higher than the data rate.
- Encoding schemes
  - Return to zero
    - A 1 is a positive voltage
    - A 0 is a negative voltage
    - At some point within each bit, return to zero.
    - It has a built in clock signal.
    - But this requires it to use twice the data rate.
      - First half of the sine wave for the bit.
      - Second half for a return to zero.
  - Nonreturn to zero (NRZ)
    - Uses a high and low voltage. High = 1, low = 0
    - Stallings tells us it is common to switch this. (low = 1)
    - This is called nonreturn to zero level
    - There is no implicit clock, and long strings of 0s or 1s can lead to clock drift.
  - Nonreturn to zero invert
    - Invert the signal on a 1.
    - No change on a zero
    - This is a differential code.
    - Let t be the current signal and b be the next bit.
```
if b == 1 
   t_n+1 = -t_n
else
   t_n+1 = t_n
	      
```
    - Stallings tells us it may be easier to detect transitions than low and high voltages.
    - It is also better for avoid noise interference.
    - A long series of zeros can still cause clock sync problems.
  - All NRZ codes
    - Easy to engineer
    - Efficient use of bandwidth
    - Energy is centered nicely
    - Requires a DC component
    - No clock in signal.
  - Multilevel Binary
    - This set of encodings use multiple levels to correct some of the issues in NRZ.
  - Bipolar-AMI
    - Alternative mark inversion.
    - (Mark is a 1, space is a 0)
    - 1 represented by a change
    - 0 represented zero.
    - ```
    if (b == 0) {
       s  = 0;
    } else {
       t_old = t;
       t = -t_old;
       s = t;
    }
    	      
```
- The transitions allow for synchronization.
- There is no net dc component.
- It uses less bandwidth than NRZ.
- Basic error detection is possible.
- Multilevel vs NRZ
  - As above
  - Multilevel, because it must distinguish three levels, needs more power (~3dB) to achieve the same BER
  - Long strings of 0 are still a problem for clock.
  - ISDN apparently introduces a bit regularly to provide a clock signal.
- Manchester
  - This is a biphase encoding technique
  - There is a transition in the middle of each bit.
  - low to high represents 1 and high to low represents 0
- Differential manchester
  - Transition in the middle of every bit.
  - If zero, transition at the beginning of the bit.
  - If 1, no transition at the beginning.
- Biphase comments
  - Require twice the clock rate.
  - Or half the data rate
  - Contain a clock signal.
  - No C component
  - If noise causes the loss of a transition, we know it. (Error detection)
  - Manchester is used for ethernet, baseband coaxial cable
  - Differential manchester for token ring.
- Modulation Rate
  - This is the number of signal changes made per second.
  - This is not the same as the data rate.
  - In NRZ, a series of 0s will have a baud rate of 0
  - In Manchester, a series of 1s will be the data rate
  - In Manchester, a series of 0s will be 2 x data rate.
```
      R     R
  D = - = ------
      L    lg M
  	      
```
  - D = modulation rate (baud)
  - R = data rate (bps)
  - M = number of different signal elements.
  - L = number of bits per signal element.
  - For Manchester, we can see
```
  R = f
  L = 1/2
  D = f/(1/2) = 2f
  	      
```
  - Stallings tells us the way to measure baud rate is using the average number of transitions.
```
  1 = 1
  0 = 2
  Average 1.5
  D = 1.5f
  	      
```
  - Using multiple transitions per bit can overcome noise and even attempts to jam the signal.
  - Using multiple bits per symbol gives higher transmission rates.
  - 1000Base-T (gigabit ethernet over copper, IEEE 802.3ab)
    - See This white paper
    - It uses 4 pairs of UTP,
    - each pair carries up to 250 Mbps, full duplex
    - It apparently transmits at 125Mbaud.
    - Each signal is encoded as a 5 level PAM (Pulse Amplitude modulation) coding scheme.
    - This uses 4 symbols to encode 2 bits per symbol.
    - D = 125Mbaud, L = 2, R = 2x125 = 250Mbps
- Scrambling Techniques
  - Biphase are used in LANS
  - But not in long-distance
  - The high signaling rate to data rate is a problem.
  - Other approach: Replace long strings of 0 or 1 with a filling sequence.
  - It must be recognized by the other side and put back.
  - Bipolar with 8-zeros substitution (B8ZS)
    - Based on AMI (zero = 0, 1 shift from previous)
    - If 8 zeros occur in a row
    - If previous pulse was + -> 000+-0-+
    - If previous pulse was - -> 000-++-
    - In each case two AMI violations ere introduced, signaling filling sequence.
    - A picture from Stallings