• What does the physical layer do?
• How do you transmit and handle data in an optical infrastructure?

Physical Layer (L1)

Bit/Signal transmission: convert 0s and 1s into electrical, optical or radio signals to transport across the medium
Physical medium: defines the hardware, i.e. copper, fibre, radio waves
Signal standards: specifies voltage, timing, data rates, modulation, ...

Modem: convert digital data to analog data (or vice versa)
Conversion from analog to digital is done through a codec

Signalling

How do we represent 1s and 0s
E.g. varying amplitude, varying frequency, phase
All need clocks (devices need to be synced)
Needed to recognise the stars of bits

Modem Speeds:

Bit rate: number of bits processed per unit of time
Bauds / signal / modulation rate: how many signalling events per unit of time
Example:
4 symbols per second = 4 Bd
8 bits per second = 8 b/s
Baud rate does not tell you bit rate: you need to know how many bits fit in a symbol
Symbols can be (for example) 0/1, or 00/01/10/11, encoding 1 or 2 bits per symbol

Encoding

How many bits can be grouped into a symbol
Groups come with advantages:
Reduce bit level error
Help distinguish data bits from control bits
Better media error detection

Non-Return to Zero Encoding (NRZ)

Used in optical transmission
High voltage = 1
Low voltage = 0
Voltage does not return to 0 between bits (stay until bit changes)
Simple, but does suffer from clock sync issues with long runs of the same bit

Manchester Encoding

Each bit is encoded as a transition, rather than a value
High-to-low mid-bit is a 1
Low-to-high mid-bit is a 0

Solves the long run of the same bit issues

Xb/Yb encoding

Encode $X$ bits of data into $Y$ bit symbols
Extra bits are used for error correction, sync, ...
8b/10b used in:
Gigabit ethernet
InfiniBand
USB 3.0
PCIe (<3.0)
64b/66b used in 10/40/100GE

Encodes bit patterns based on amplitude

Recognising Frame Signals

Specific physical signals denote start and end of frame
There may be random signals in-between due to e.g. interference
These are not decoded nor passed up to L2

Transmission Media

Electrical transmission
Electrical pulses through a wire
E.g. UTP/STP, COAX
Optical
Photons through a glass fibre
Higher bandwidth, longer distances
200KM / ms
$\lambda =c/v$
Multi-mode fiber:
Cheaper
Orange cable
Thicker glass, more refractions
Cheaper
850/1300nm
Single-mode fibre
Long distance
Less refraction

Optical Networks

First generation optics:
Optics used for transmission
Switching and logic handled in electronics
Second generations optics:
Fully optical

Basic Components

Couplers and splitters used to combine and split signals
Taps take a small portion of transmission power from a light stream

(De-)Multiplexer send multiple wavelengths over the same fibre, split them at the end

Micro Electronic Mirror (MEMs) devices
Use tiny mirrors to determine the pathing of light into fibres
Basically L0 switching

Transmitters

Lasers are the light sourced
Usually fixed wavelength
Both sides should have the same wavelength
Tunable lasers can alter wavelength
Enables reconfigurable or optical packet switched networks

Signal Degradations

3 types of degradation:

1. Loss of energy (attenuation)
2. Shape distortion (dispersion)
3. Loss of timing (jitter)

Fixed by the 3 R's:

1. Re-amplify (in the middle of transmission)
2. Re-shape (at the end)
3. Re-time (needs to convert to electric and back, O->E->O, conversion to electric is undesirable)

Attenuation

Specified as loss per KM (dB/KM)
-0.40 dB/KM at 1310 nm
-0.25 dB/KM at 1550 nm

Decibels (dB) is a relative unit to express differences in signal strength

Decibels milliwatt (dBm) is the unit to express the power of an interface, and the receiver sensitivity
This is an absolute value
$$dBm=10Log_{10}(\frac{Power}{1mW})$$

Power Budget

In order to get the correct power for your receiver you may need to:
Lower the power: use an attenuator
Increase the power: use an amplifier

Dispersion

Modal dispersion
For multi-mode fibres, different wavelengths bounce at different angles, changing their speed through the core
Chromatic dispersion:
Diff. wavelengths travel at different speeds
Polarisation mode dispersion:
Real fibre has tiny asymmetries causing light to split into two polarisation modes

Can be calculated based on the quality and length of the fibre, and the wavelength/data you use

Jitter

The precise timing can shift slightly

Multiplexing Schemes

Time-Division Multiplexing

Physically take turns on the channel
Divide time domain into several rotating time-slots of fixed length, one for each incoming stream

SONET/SDH

SONET: Synchronous Optical Networking
SDH: Synchronous Digital Hierarchy

Put a bit from every stream into a section of your frame

Limitations:
Designed for voice, i.e. fixed-bandwidth traffic
Less suited for IP, as it is variable bandwidth and frame size
End up with:
Byte boundary ambiguity: where does a packet end and the next begin in the payload
Idle fill: if there are no packets, what kind of idle pattern do you use?
Packets spanning frames: how do we re-assemble packets across frames

So people invented the Generic Framing Protocol (GFP) to make packets more suited for SONET

Wave-Division Multiplexing

Multiplex multiple signals onto a single fibre using multiple wavelengths of light
Now all (or multiple) signals can send at the same time

Transponders convert incoming optical signals into ITU-standard wavelengths

Similar to GFP, Optical Transport Network (OTN) is a similar method of creating payloads to send onto the optical channel

Coarse versus Dense WDM

Dense WDM (DWDM)
Channel spacing of 0.8nm
Up to 96 channels
Thousands of KMs of distance (w/ amplifiers)
Coarse WDM
Channel spacing of 20nm
Up to 18 channels
40-80KM range, un-amplified
Suitable for metro access, enterprise, data centre interconnect
(Cheaper)