Physical Layer Rudra Dutta ECE/CSC Fall 2007, Section 001

Copyright Fall 2007, Rudra Dutta, NCSU2 Context Lowest of OSI layers Provides a bit pipe More communications than networking We want to understand: – General techniques – Theoretical results

Copyright Fall 2007, Rudra Dutta, NCSU3 Communication Links Various technologies for physical communication – Copper wire, coax, fiber, radio, satellites, … Single underlying phenomenon - EM waves One way to utilize the phenomenon - guide – Copper, glass Another way – no guide – Free space (“wireless cable”) – Radio, optical (Smoke signal? Semaphore? Magnetic tapes? Pigeons (supersonic or otherwise) ?)

Copyright Fall 2007, Rudra Dutta, NCSU4 EM Waves Energy can carry information More correctly, distribution of energy EM waves carry energy, hence information Amplitude, frequency, phase Modifications (“modulations”) of these carry information

Copyright Fall 2007, Rudra Dutta, NCSU5 Digital or Analog ? Digital – concept – Information can be analog or digital EM waves – analog by definition Analog EM signal can be made to transfer digital data Thus we could (and usually do) have: “Digital interpretation of analog signal representing digital representation of analog data” …

Copyright Fall 2007, Rudra Dutta, NCSU6 Propagation Media Guided – Twisted pair – Coax cable – Optical fiber Unguided – Radio (semi-guided follow curvature of earth) – Radio bounced off ionosphere – Fiberless optical (wireless optical) – Communication satellites

Copyright Fall 2007, Rudra Dutta, NCSU7 Communication in the EM Spectrum

Copyright Fall 2007, Rudra Dutta, NCSU8 Modulation A “carrier” wave exists on the medium – Own amplitude, frequency, phase – Base energy pattern – no information – Analog, of course A “signal” needs to be transmitted – Time varying; analog, or digital The value of the signal from instant to instant is used to change the energy pattern of the carrier

Copyright Fall 2007, Rudra Dutta, NCSU9 Injection Baseband – No carrier, modulation – State of the medium (voltage) is made to follow signal one-to-one – Uses “entire” medium Broadband – Modulation of a carrier – Carriers at different frequencies can carry different signals – Sinusoidal advantages – remember harmonic analysis – Natural frequencies of transmission

Copyright Fall 2007, Rudra Dutta, NCSU10 Synchronous vs. Asynchronous Various use of these terms Very multiply defined terms Can be used for traffic – ITU-T and CCITT have different definitions – Others such as FDDI possible In this transmission context – With clock - asynchronous Can fall out of step - long string of zeros – Synchronous - clock not needed

Copyright Fall 2007, Rudra Dutta, NCSU11 Synchronous Baseband Transmission “Self-synchronizing” codes – Provide guaranteed transitions in clock ticks – Rate suffers Manchester – Transitions, not states, indicate bits Many others – NRZI - Transitions indicate 1’s (needs line code) – MLT-3 - Alternate 1’s are high and low (needs line code)

Copyright Fall 2007, Rudra Dutta, NCSU12 Broadband injection Amplitude, Frequency, or Phase may be modulated “Shift keying”

Copyright Fall 2007, Rudra Dutta, NCSU13 BPSK modulation PSK has excellent protection against noise Information is contained within phase Noise mainly affects carrier amplitude

Copyright Fall 2007, Rudra Dutta, NCSU14 QPSK Modulation, QAM

Copyright Fall 2007, Rudra Dutta, NCSU15 Multiplexing Techniques to employ same medium for multiple transmissions Requirement: over same reasonably short time, each transmission should receive some share of medium capability Two main methods – Frequency division – Time division – Combinations thereof – Code division – new concept

Copyright Fall 2007, Rudra Dutta, NCSU16 Frequency Division Multiplexing (a) The original bandwidths (b) The bandwidths raised in frequency (b) The multiplexed channel

Copyright Fall 2007, Rudra Dutta, NCSU17 Wavelength Division Multiplexing

Copyright Fall 2007, Rudra Dutta, NCSU18 Time Division Multiplexing The T1 carrier (1.544 Mbps)

Copyright Fall 2007, Rudra Dutta, NCSU19 Time Division Multiplexing (2) TDM can be hierarchically performed

Copyright Fall 2007, Rudra Dutta, NCSU20 GSM – A Combined Approach GSM uses 124 frequency channels, each of which uses an eight-slot TDM system

Copyright Fall 2007, Rudra Dutta, NCSU21 CDMA – A New Approach Combines multiplexing and collision issues – New approach lies in treating collisions - may extract some data – Multiplexing is more like FDM Binary “chip” sequences assigned to stations May appear that bit rate increase should not result - in fact does – Power control an essential part – We discuss later (in MAC)

Copyright Fall 2007, Rudra Dutta, NCSU22 The Issue of Bitrate Consider simple AM (ASK) – Transmit one of two distinct amplitudes (voltages)  transmission of one bit How soon after can we transmit another bit? – How fast can transmitter change its state? – How fast can receiver recognize line state? – Appears to limit bit rate, but - Does not have to be just two states – Why not transmit one of four distinct amplitudes? – Why not more?  No limit to bit rate

Copyright Fall 2007, Rudra Dutta, NCSU23 Channel Characteristics Channel modifies the EM wave

Copyright Fall 2007, Rudra Dutta, NCSU24 Phenomena Hindering Transmission Interference with energy (pattern) – Attenuation (entropy loss) – Distortion (variable delay of different energy packets) – Dispersion – Noise (unpredictable) All but noise can be guarded against – With the ideal infinite data transfer rate

Copyright Fall 2007, Rudra Dutta, NCSU25 A Little Communication Theory The road to EM transmission: – Fourier: Harmonic analysis – Nyquist: Sampling theorem – bit rate – Shannon: Bit rate in presence of noise Briefly, – Most signals can be represented by sinusoid combinations – Discrete time sampling can reconstruct signals – Noiseless channel has limited maximum bit rate – Noise reduces maximum bit rate

Copyright Fall 2007, Rudra Dutta, NCSU26 Bandwidth-Limited Signals (a) A binary signal and its root-mean-square Fourier amplitudes, (b-c) successive approximations

Copyright Fall 2007, Rudra Dutta, NCSU27 Bandwidth-Limited Signals (2) (d) – (e) Successive approximations to the original signal

Copyright Fall 2007, Rudra Dutta, NCSU28 Sampling – Nyquist’s Theorem Twice the highest frequency  no reconstruction loss

Copyright Fall 2007, Rudra Dutta, NCSU29 Nyquist’s Result – Intuitive View Fitting a sinusoid – Low-rate sampling  wrong sinusoid – Half-rate sampling  wrong sinusoid – Full-rate sampling  still could be wrong – Double rate  no possibility of wrong sinusoid “Highest frequency” – Naturally introduced by device characteristics – Medium carries all frequencies between a lowest and highest frequencies (“frequency band”) – Hence “band” “width”

Copyright Fall 2007, Rudra Dutta, NCSU30 Bandwidth limited Bit rate Nyquist’s theorem – Maximum bit rate = 2H log 2 V bits/sec – H = bandwidth – V = number of discrete states Shannon’s theorem – Maximum bit rate = H log 2 (1 + S/N) bits/sec – Introduces signal-noise ratio – Insight: random characteristic limits bit rate – Note on application SNR in Shannon’s theorem - ratio of power content (P S /P N ) Usual unit of SNR - dB, a logarithmic unit dB = 10 log 10 (P S /P N )

Copyright Fall 2007, Rudra Dutta, NCSU31 Comparing Results Both results give bitrates, but – With different assumptions and input Nyquist’s theorem – Bit rate IF exactly V states are successfully used Mo-Dem equipment already decided – Noise must allow V states Often stated as perfect channel assumption Shannon’s result – Estimation of what value of V will be successful Noise level decides, so need noise level as input Either might be larger, depending on input

Copyright Fall 2007, Rudra Dutta, NCSU32 What Have We Learned? EM communication links provide bit pipes – lowest layer of networking Various transmission methodologies Theoretical results providing channel bit rates At higher layers, bit rate is what we are primarily interested in Validation of layering concept