Media: Fiber
Dr Markus Buchhorn: [email protected]
Optical Fibre Characteristics
• When copper just won’t do…
• Light (weight), very robust to oxidation, water, electrical interference, …
• Can go a long way
• But
– not very flexible (it breaks, cracks),
– not easy to join (need to melt it)
• Easy to make thin cables
– Copper down to 0.03mm diameter
– Standard Fibre down to 8micron diameter (0.008mm)
• And can go well below 1micron in the lab
• Not as cheap(?)
Quick change in language
• In copper we talk about frequency (f or ) [in Hz or s-1]
• In Fibre we talk about wavelength (λ) [in meters]
• High frequency ≡ Short wavelength
• c (speed) = f * λ (speed in that media)
2 λ
cvacuum = 300,000km/s
ccopper,glass ≈ 2/3 c
Huge change in performance
• Copper we use kHz (103) to MHz (106), Wireless to GHz (109)
• Optical we start at THz (1012) and up
• VDSL 12 MHz = 25 m
• Wifi 2.4GHz = 12.5 cm
• Yellow light = 600nm = 500 THz
How it works – Physics!
• Index of refraction = c/(speed of light in that material) = “n”
• Cross from one material to the other
= change of speed = change of direction
• Make a fibre cable:
– Take a glass fibre core,
– Wrap it in different glass
– Protect it with a plastic jacket
ncladding < ncore
How it works
• Get the angle and change of material right:
– Total internal reflection
• Within a ‘critical angle’
• Step-index fibre
• Graded-index fibre
Animations from http://www.thefoa.org/tech/ref/basic/fiber.html
Each ray = a “mode”
Modal distortion
Straight line is slowest
Reduces modal distortion
Multimode vs Singlemode
• These are “multi-mode” fibres
– With significant modal distortion
• Make core much narrower
– a few wavelengths?
– “single mode” fibre
• Performance goes way up
• So does Cost
Fibre standards
• For the cable
– Multimode fibres, OM1 (62/125μm), OM2-OM5 (50/125 μm)
– Singlemode fibres, ITU G.652-G.657 (9/125 μm)
– Performance expectations more than actual manufacturing
• For the connector
– So, so many standards
• Vary by sector
• 30+ on Wikipedia
• Do not mix up your cables
– MMF into SMF does mechanically work, but “performance suffers”.
Fibre connectors
• Need to be “perfect”!
• Glass face to Glass face
– Dust is a serious enemy
– Most use curved (or sometimes angled) faces, reduce reflections at the end
• Terminating copper cables involves scissors and pliers
• Terminating fibre involves melting and polishing glass/plastic
– Thinner than a human hair
• Splicing fibre directly via melting (good), or glueing (less good)
– But glue is cheaper…
Losses in fibre
• Attenuation 0.4-3 dB/km, due to
– Scattering (structures+materials in the fibre)
– Absorption (materials in the fibre)
– Distances of many km are trivial
• 8km still yields 75% of the original light.
• Attenuation depends on wavelength
• Fibres have multiple passbands
– Due to various materials
and manufacturing techniques
– Always improving, both absorption and range
Window Third
Other losses in fibre
• Chromatic dispersion
– Index of Refraction varies with wavelength
– A pure single wavelength is hard to do (even for a laser)
– Soliton pulses fix it
• Polarisation mode dispersion
– Core shape helps fix it
“Loss budgets”
• You have an energy budget
– How much you can send
– How much you need to receive (for a reasonably clear signal)
• Loss due to the fibre [0.4-1.0 db/km for SMF]
• Loss due to (mechanical) connectors [0.3dB each]
• Loss due to (physical) splicing [0.3dB each]
Going round corners
• Copper has tight bending radius
– N times diameter
• Fibre does not
– Fractures, causing attenuation and/or interference
– MMF better than SMF
• Fibre testing
– Send a pulse, look what gets through and what comes back!
THz channels
• A single wavelength of light at 500 THz can carry 1+Pb/s?
– In theory… a very faint signal!
• Electronics can’t (yet) keep up or see – better to be clearer and slower!
– Use ASK, PSK
• Can also do Polarisation Division Multiplexing
– Frequency-DM becomes Wavelength-DM – multiple wavelengths of light
• WDM can be coarse (CWDM) and dense (DWDM)
– Usually use each wavelength as a separate channel
Add-Drop Multiplexing - optically
• Given N wavelengths (channels) on a fibre
– How do I add one more, or take one out?
• Active electronic separation vs Passive optical separation of channels
Transmitting over light
• Convert digital data to optical signalling (modulation…)
• You can pulse an led
• Fast-ish, and very cheap
• Don’t get very bright or nicely shaped pulses
• Too broad a colour range
• Used in MMF
• You can chop a laser
• Semiconductor lasers are now 100micron in size
• And wavelength tunable on the fly
• And use an “Optical Mach-Zehnder” modulator
• Used in SMF
Multi-core cable design
• Individual fibres are fragile
• Fibres with sub-fibres or hollow channels
• Cable bundles up to 1024 fibres
• Costs as much to bury/hang as a single fibre
• Connect buildings, cities, countries
• People often still want their own fibre path
– Security
– Guaranteed performance
– Avoid interference
– New technologies
– Concept of ‘dark’ and ‘grey’ fibre
• An empty glass you can fill how you want, or
• A slot alongside others to add your wavelength
How fast can you go?
• Depends on “fast”
• Highest absolute speed (2012): 1Pb/s over 50km (that’s 1,000,000Gb/s…)
– And (2017)10.2Pb/s over 11km
• High speed over long distances (2009): 100Gb/s over 7000km times 155λ
– And (2016) 65Tb/s over 6600km (?)
• Various techniques, with multiple wavelengths, multiple cores, pre-distortion, …
• Better measure: bandwidth*distance
– 2016: 4*105 (Tb/s).km
– 2012: 5*104 (Tb/s).km
– 2012 used very specialised fibre, 2009 was commodity fibre
– 2020 Internet traffic estimate: 375,000PBytes/month = 1.4Pb/s…
– And these measures don’t include energy costs!
How far can you go?
• Without effort, 1-2 km over MMF, 50-100 km over SMF
• Want more? Brighter lasers – gets difficult
• Regenerate/Repeat every 50-100 km
– Expensive optics and electronics, reconstruct signal and retransmit – perfectly
– Optical-electronic-optical “OEO” interfaces
• Amplify every 50-100 km
– Cheap electronics, and can even be done optically (erbium doping)
– Amplifies signal and noise.
– Record: 2015 - Melbourne to Melbourne, via Sydney&Perth = 10,358km
Undersea cabling
Underwater cables are safe!
Doing more with less
• Unlike copper, fibre is not (easily) a shared medium
– Point-to-point
• Can use a single fibre for RX and TX at the same time
– Optical splitters at both ends
– Can get crosstalk, in connectors, and within fibre
• Still more common to have a fibre pair though
The Last mile and fibre
• Lots of discussion about ‘costs’
– Compare capital costs of installing fibre vs copper
– Compare running costs of fibre vs copper
– Compare performance of fibre vs copper
• General approach – mix it up:
– Push fibre as close as you “can” (afford, achieve)
• To the home? To the driveway? To some nearby node?
– then use copper for the last bit (with DSL)
– While keeping an eye on the electronics
Exchange Range
Copper links = 4km circle
Fibre links = 40+km circle
Radius2 # of houses
Exchange Range
Some FTTx models
• Leverage what is already in the ground/on the poles
– Most homes have landline phones = capital investment
• Reduce the average distance of copper
– Push fibre as deeply as “affordable” (by who?)
Copper (POTS - DSL)
FTT Home
FTT Kerb
< 300m
< 4km
FTT Node
What needs power?
Who is sharing?
How many houses per ‘box’
What needs managing?
Compare FTTx with HFC
• Fibre To The … (FTTx):
– FTTPremises/FTTHome,
– FTTBasement,
– FTTCurb/Kerb/dp
– FTTNode
• Hybrid Fibre Coax (HFC)
– “FTTN” to get you close,
– Then shared (coax) copper
• To a lot of houses, 50-100’s in NBN
• And again try to make it cheap to run…
– What is active, smart, powered, etc.
– And what isn’t.
• (Gigabit) Passive Optical Networking
• Using the flexibility of optics
– And the allure of cheap broadband
– With minimal fibre usage…
• TDM out to the active Optical Network
Units or Terminals (ONU/ONT)
• TDM out to the (passive) splitters
– From the Optical Line Terminals (OLT)
• WDM: RX and TX on a single fibre,
– SDM: Make more money out of the cable
51作业君 51作业君



添加客服微信: ITCSdaixie