Optical Fiber Communications

Chapter 8
Digital Links
 Overview – Chapter 8
8.1 Point-to-Point Links
8.2 Power Penalties
8.3 Error Control
8.4 Coherent Detection
8.5 Differential Quadrature Phase-Shift Keying

2
 Point-to-Point Links
Key system requirements needed to analyze optical fiber links:
1. The desired (or possible) transmission distance
2. The data rate or channel bandwidth
3. The bit-error rate (BER)
LED or laser MMF or SMF pin or APD

(a) Emission wavelength (a) Core size (a) Responsivity

(b) Spectral line width (b) Core index profile (b) Operating λ
(c) Output power (c) BW or dispersion (c) Speed
(d) Effective radiating area (d) Attenuation (d) Sensitivity
(e) Emission pattern (e) NA or MFD

3
 Link Power Budget
• The link loss budget is derived from the sequential loss
contributions of each element in the link.
• Each of these loss elements is expressed in decibels (dB) as
• The link loss budget considers the total optical power loss PT allowed
between the light source and the photodetector, and allocates this loss
to cable attenuation, connector loss, splice loss, and system margin:
PT = PS – PR = 2lc + αL + system margin

4
 Power Budget Example
• Specify a 20-Mb/s data rate and a BER = 10–9.
• With a Si pin photodiode at 850 nm, the required receiver input signal is –42 dBm.
• Select a GaAlAs LED that couples 50 mW into a 50-μm core diameter fiber flylead.
• Assume a 1-dB loss occurs at each cable interface and a 6-dB system margin.
• The possible transmission distance L = 6 km can be found from
PT = PS – PR = 29 dB = 2lc + αL + system margin = 2(1 dB) + αL + 6 dB
• The link power budget can be represented graphically (see the right-hand figure).

5
Example: Spreadsheet Power Budget

6
 Rise-Time Budget (1)
• A rise-time budget analysis determines the dispersion
limitation of an optical fiber link.
• The total rise time tsys is the root sum square of the rise times
from each contributor ti to the pulse rise-time degradation:
– The transmitter rise time ttx
– The group-velocity dispersion (GVD) rise time t GVD of the fiber
– The modal dispersion rise time tmod of the fiber
– The receiver rise time trx
Here Be and B0 are
given in MHz, so
all times are in ns.

7
Rise-Time Budget (2)

8
 Short-Wavelength Band
Attenuation and dispersion limits on the transmission distance vs. data rate
for a 770–910-nm LED/pin combination.
•The BER was 10–9 ; the fiber-coupled power was –13 dBm up to 200 Mb/s.
•The attenuation limit curve was derived by using a fiber loss of 3.5 dB/km
•The receiver sensitivities shown in the left figure (8.3)

9
 Attenuation-Limited Distances for Two Single-Mode Links
1. A DFB laser that has a fiber-coupled output of 0 dBm at 1550 nm.
2. At 1550 nm the single-mode fiber has a 0.20-dB/km attenuation.
3. The receiver has a load resistor RL = 200 Ω and T = 300°K.
4. At a 10–12 BER a value of Q = 7 is needed.
5. The InGaAs pin and APD photodiodes have a responsivity of 0.95 A/W.
6. The gain of the APD is M = 10 and the noise figure F(M ) = 5 dB.

10
 Power Penalties
• When any signal impairments are present, a lower optical power level arrives at
the receiver compared to the ideal reception case.
• This lower power results in a reduced SNR, which leads to a higher BER.
• The ratio of the reduced received signal power to the ideal received power is the
power penalty for that effect and is expressed in decibels.
• If Pideal and Pimpair are the received optical powers for the ideal and impaired cases,
respectively, then the power penalty PPx in decibels for impairment x is

• In some cases, increasing the received optical power can reduce the power
penalty. For other cases (some nonlinear effects) increasing the received
power level will have no effect on the power penalty.
• Power penalties may be due to chromatic and polarization-mode dispersions,
modal (speckle) noise, mode-partition noise, the extinction ratio, chirp,
timing jitter, reflection noise, and nonlinear effects arising from high optical
power level in a fiber link.
11
 Chromatic Dispersion Penalty
• Chromatic dispersion arises since each wavelength has a slightly
different velocity, and thus they arrive at different times at the fiber end.
• Therefore, the range of arrival times at the fiber end of the spectrum of
wavelengths will lead to pulse spreading.
• The accumulated chromatic dispersion increases with distance.
• A basic estimate of what limitation chromatic dispersion imposes on link
performance can be made by specifying that the accumulated dispersion
should be less than a fraction ε of the bit period Tb = 1/B, where B is the
bit rate:

The ITU-T Rec. G.957 for SDH and the Telcordia Generic Requirement
GR-253 for SONET:
•For a 1-dB power penalty the accumulated dispersion should be less
than 0.306 of a bit period (ε ≤ 0.306).
•For a 2-dB power penalty the requirement is ε ≤ 0.491.
12
 Polarization-Mode Dispersion Penalty
• Polarization-mode dispersion (PMD) arises sincet light-signal energy
at a given wavelength in a single-mode fiber occupies two
orthogonal polarization states or modes.
• PMD arises because the two fundamental orthogonal polarization
modes travel at slightly different speeds owing to fiber birefringence.
• This PMD effect cannot be mitigated easily and can be for links
operating at 10 Gb/s and higher.
• To have a power penalty < 1.0 dB, pulse spreading from PMD must
on the average be less than 10% of a bit period Tb:

13
 Extinction Ratio Power Penalty
• The extinction ratio re in a laser is the ratio of the optical power level P1
for a logic 1 to the power level P0 for a logic 0, that is, re = P1 / P0.
• Letting P1-ER and P0-ER be the 1 and 0 power levels, respectively, with a
nonzero extinction ratio, the average power is

• When receiver thermal noise dominates, the 1 and 0 noise powers are
equal and independent of the signal level. In this case, letting P0 = 0 and
P1 = 2Pave, the extinction ratio power penalty becomes

• Optical transmitters usually have minimum re of 7 to 10 (8.5 to 10 dB).

• The power penalties range from 1.25 to 0.87 dB.
• A minimum re = 18 is needed for an ER power penalty < 0.5 dB.

14
 Chirping Power Penalty
• A single-mode laser may experience dynamic line broadening when the
injection current is directly modulated.
• This line broadening is a frequency “chirp” associated with modulation-
induced changes in the carrier density.
• Chirping can produce significant dispersion when the laser wavelength is
displaced from the zero-dispersion wavelength of the fiber.
• When the effect of laser chirp is small, the eye closure Δ can be
approximated by

• The power penalty for an APD can be estimated from the SNR
degradation (in dB) due to the signal amplitude decrease as

15
 ER and Chirping Power Penalties
Extinction-ratio, chirping,
and total-system power
penalties at 1550 nm for a
4-Gb/s 100-km single-
mode G.652 link having a
dispersion D = 17 ps/(nm-
km) and a DFB laser with
an active layer width of
1.75 mm. (From Corvini
and Koch, Ref. 53.)

16
 8.3 Error Control
• An error in a data stream can be a single-bit error or a burst error.
• In a burst error more than a single bit in a data unit has changed.
• This type of error happens most often because the duration of a noise
burst lasts over several bit periods.
• The length of an error burst is measured from the first corrupted bit to
the last corrupted bit. Not all the bits in this segment were damaged.

17
 Concept of Error Detection
The basic concept of error detection:
• Prior to being inserted into a transmission channel, the
information bit stream coming from a communication
device is encoded so that it satisfies a certain pattern or a
specific set of code words.
• At the destination the receiver checks the arriving
information stream to verify that the pattern is satisfied.
• If the data stream contains segments (that is, invalid code
words) that do not conform to the intended pattern, then an
error has occurred in that segment.

18
 Linear Error Detection Codes
• The single parity check code is a simple error detection method.
• A code word combines k information bits and a single added check bit.
• If the k information bits contain an odd number of 1 bits, then the check
bit is set to 1; otherwise it is set to 0.
• This procedure ensures that the code word has an even number of ones,
which is called having an even parity. The check bit is called a parity bit.
The single parity check code can detect when an odd number of errors
has occurred in a code word.
• This is a linear code because the parity bit bk+1 is calculated as the
modulo 2 sum of the k information bits:

b1, b2, … , bk are the information bits.

19
 Binary Linear Code
• A more general linear code with stronger error detection
capabilities is called a binary linear code.
• This code adds n–k check bits to a group of k information bits
to form an n-bit code word designated by the notation (n, k).
• One example is the (7, 4) linear Hamming code
• The first four bits of a code word are the information bits b 1,
b2, b3, b4 and the next three bits b5, b6, b7 are check bits.
• This particular one can detect all single and double bit errors,
but fails to detect some triple errors.

20
 Polynomial (CRC) Codes
• The cyclic redundancy check (CRC) technique uses a binary division
process on the data portion of a packet and a set of redundant bits.
• Step 1. At the sender end a string of n zeros is added to the data unit on
which error detection will be performed.
• Step 2. The enlarged data unit is divided by an n+1 predetermined
divisor. The remainder that is called the CRC.
• Step 3. The n-bit CRC replaces the n zeros that were added in step 1.

Step 4. The receiver divides

the composite unit by the
n+1 divisor.

Step 5. A remainder
indicates that some bits
have become corrupted
during the transmission
process.
21
 Common CRC Codes
• Popular CRC codes are designated as CRC-8, CRC-10, CRC-16, and CRC-32.
• The numbers 8,10,16, and 32 refer to the size of the CRC remainder.
• The CRC divisors for these polynomials are 9,11,17, and 33 bits.

22
 Forward Error Correction
• Forward error correction (FEC) may be done with redundancy
in the data stream.
• FEC codes can be cyclic codes (e.g., Reed-Solomon [RS] codes)
• RS codes add a redundant set of r symbols to blocks of k data
symbols, with each symbol being s bits long, e.g., s = 8.
• The codes are designated by the notation (n, k) where n = k + r
• For a symbol size s, the maximum RS code word is n = 2s - 1.
• RS codes can correct up to t symbol errors, where 2t = n - k.
• RS codes also allow transmission at a lower power level to
achieve the same BER as without encoding.
• This is a coding gain. The (255,239) RS code gives a 6-dB gain.

23
 8.4 Coherent Detection
• Coherent detection provides gain to the incoming optical
signal by combining or mixing it with a locally generated
continuous-wave (CW) optical field.
• The result of the mixing is that the dominant noise in the
receiver is the shot noise coming from the local oscillator.
• Thus the receiver can have a shot noise limited sensitivity.

24
 Coherent Receiver Performance
• Let the electric field of the transmitted optical signal to be a
plane wave having the form Es = As cos[ωst + φs(t)]
– As is the amplitude of the optical signal field, ωs is the optical signal
carrier frequency, and φs(t) is the phase of the optical signal
• Let the local-oscillator field be ELO = ALO cos[ωLOt + φLO (t)]
– ALO is the amplitude of the local oscillator field, and ωLO and φLO (t) are
the local oscillator frequency and phase, respectively
• Then letting
– φ = φs - φLO be the phase difference between the signal and LO fields
– cos θ(t) be the polarization misalignment between the signal and LO
• The optical power P(t) at the photodetector is

25
BER Comparisons

26
Comparison of Number of
Required Photons per Bit

27
 Differential Quadrature Phase-Shift Keying
• Multilevel modulation formats are of interest for 10 and 40 Gb/s speeds
• This modulation format transmits more than one bit per symbol.
• In the DQPSK method, information is encoded by means of the four phase
shifts {0, + π/2, - π /2, π}.
• The set of bit pairs {00, 10, 01, 11} is assigned to individual phase shifts.
– For example, a phase shift of π means that the bit pair 11 was sent.
• Thus DQPSK transmits at a symbol rate of half the aggregate bit rate.
• A DQPSK transmitter typically uses two nested Mach-Zehnder modulators

28