MODULE III. Foundations Of Data Communications And The Physical Layer

Transcription

1 MODULE III Founations Of Data Communications An The Physical Layer Computer Networks an Internets -- Moule 3 1 Spring, 2014 Copyright All rights reserve.

2 Topics Motivation an moel Information sources an signals Transmission meia Reliability an channel coing Transmission moes Moulation an emoulation Multiplexing an emultiplexing (channelization) Computer Networks an Internets -- Moule 3 2 Spring, 2014 Copyright All rights reserve.

3 Motivation An Moel

4 What Is Data Communications? Broa fiel of stuy Usually associate with the Physical Layer Touches on Physics Mathematics Engineering Inclues Transmission of signals Encoing ata Moulation an multiplexing Computer Networks an Internets -- Moule 3 4 Spring, 2014 Copyright All rights reserve.

5 Motivation Fin ways to transmit analog an igital information Using natural phenomena (e.g., electromagnetic raiation) Allow multiple seners to share a transmission meium Data communications provies A conceptual framework Mathematical basis Computer Networks an Internets -- Moule 3 5 Spring, 2014 Copyright All rights reserve.

6 Key Concept Although we ten to think of analog an igital communication separately, ultimately, all communication uses the same physical phenomena, usually electromagnetic energy. Differences lie in the way the physical phenomena are use Analog: use all values in a continuous range Digital: restrict use to a fixe set of values, usually two Data communications covers both analog an igital Computer Networks an Internets -- Moule 3 6 Spring, 2014 Copyright All rights reserve.

7 Conceptual Framework For Data Communications Information Source 1 Source Encoer Encryptor (Scrambler) Channel Encoer... Multiplexor Information Source N Source Encoer Encryptor (Scrambler) Channel Encoer Moulator Physical Channel (noise & interference) Demoulator Channel Decoer Decryptor (Unscrambler) Source Decoer Destination 1 Demultiplexor... Channel Decoer Decryptor (Unscrambler) Source Decoer Destination N Computer Networks an Internets -- Moule 3 7 Spring, 2014 Copyright All rights reserve.

8 Information Sources An Signals

9 Sources Of Information An input signal can arise from Transucer such as a microphone Receiver such as an Ethernet interface We use the term signal processing to escribe the recognition an transformation of signals Computer Networks an Internets -- Moule 3 9 Spring, 2014 Copyright All rights reserve.

10 Sine Waves Funamental because sine waves characterize many natural phenomena Examples Auible tones Raio waves Light energy Computer Networks an Internets -- Moule 3 10 Spring, 2014 Copyright All rights reserve.

11 Fourier Analysis Multiple sine waves can be ae together Result is known as a composite wave Correspons to combining multiple signals (e.g., playing two musical tones at the same time) Mathematician name Fourier iscovere how to ecompose an arbitrary composite wave into iniviual sine waves Fourier analysis provies the mathematical basis for signal processing Ba news: accoring to Fourier, a igital wave ecomposes into an infinite set of sine waves Computer Networks an Internets -- Moule 3 11 Spring, 2014 Copyright All rights reserve.

12 Sine Wave Characteristics Three important characteristics are use in networks: frequency, amplitue, an phase 1 1 sec sec 2 sec 0 t 0 2 sec t -1-1 (a) Original sine wave: sin(2πt) (b) Higher frequency: sin(2π2t) sec 2 sec 1 sec 0 t 0 2 sec t -1-1 (c) Lower amplitue: 0.4 sin(2πt) () New phase: sin(2πt+1.5π) Computer Networks an Internets -- Moule 3 12 Spring, 2014 Copyright All rights reserve.

13 Definition Of Analog Banwith Decompose a signal into a set of sine waves an take the ifference between the highest an lowest frequency Easy to compute from a frequency omain plot Example signal with banwith of 4 Kilohertz (KHz): amplitue 1 banwith frequency (in KHz) Computer Networks an Internets -- Moule 3 13 Spring, 2014 Copyright All rights reserve.

14 Digital Signals An Signal Levels A igital signal level can represent multiple bits Example amplitue amplitue time time bits sent 16 bits sent two levels with a single bit per level four levels with two bits per level Bau rate is number of times signal changes per secon; ata rate in bits per secon = bau log 2 ( levels ) Computer Networks an Internets -- Moule 3 14 Spring, 2014 Copyright All rights reserve.

15 Converting Digital To Analog Approximate igital signal with a composite of sine waves: t... (a) igital signal (b) sin(2πt/2) (c) sin(2πt/2)+α sin(2π3t/2) () sin(2πt/2)+α sin(2π3t/2)+β sin(2π5t/2) Mathematically, the banwith of a igital signal is infinite Computer Networks an Internets -- Moule 3 15 Spring, 2014 Copyright All rights reserve.

16 Converting Analog To Digital Three steps taken uring conversion PCM encoer analog signal sampling quantization encoing igital ata Example sampling using eight levels quanta time Computer Networks an Internets -- Moule 3 16 Spring, 2014 Copyright All rights reserve.

17 Sampling Rate An Nyquist Theorem How many samples shoul be taken per secon? Mathematician name Nyquist iscovere the answer: sampling rate = 2 f max where f max is highest frequency in the composite signal Example: to capture auio frequencies up to 4000 Hertz, a igital telephone system samples at 8000 samples per secon Amount of ata generate by a single igitize voice call: ata rate = 8000 samples 8 bits = 64,000 bits secon sample secon Computer Networks an Internets -- Moule 3 17 Spring, 2014 Copyright All rights reserve.

18 Nonlinear Encoing Linear sampling oes not work well for voice Researchers create nonlinear sampling that moify ynamic range to reprouce souns to which the human ear is sensitive Mu-law (µ-law) Use in North America an Japan More ynamic range, but more sensitive to noise A-law Use in Europe Less sensitive to noise, but less ynamic range Computer Networks an Internets -- Moule 3 18 Spring, 2014 Copyright All rights reserve.

19 Synchronization Errors An Line Coing Synchronization error occurs when receiver an sener isagree about bit bounaries (clocks iffer) sent receive Line coing techniques prevent synchronization errors Computer Networks an Internets -- Moule 3 19 Spring, 2014 Copyright All rights reserve.

20 Example Line Coing: Manchester Encoing Use with Ethernet Synchronizes receiver with sener (transition represents bit) Example of (a) Manchester Encoing, an (b) ifferential Manchester Encoing: (a) (b) Computer Networks an Internets -- Moule 3 20 Spring, 2014 Copyright All rights reserve.

21 Transmission Meia

22 A Taxonomy Of Transmission Meia Electrical Twiste Pair Coaxial Cable Optical Fiber Energy Types Light InfraRe Laser Electromagnetic (Raio) Terrestrial Raio Satellite Is anything omitte? Computer Networks an Internets -- Moule 3 22 Spring, 2014 Copyright All rights reserve.

23 Some Really Ba News In the real worl, entropy rules Transmission is plague with problems Computer Networks an Internets -- Moule 3 23 Spring, 2014 Copyright All rights reserve.

24 Loss, Interference, An Electrical Noise Problems in the electrical an electromagnetic worls Resistance (leas to loss) Capacitance (leas to istortion) Inuctance (leas to interference) Ranom electromagnetic raiation is calle noise Can be generate by specific sources such as electric motor Backgroun raiation is an inescapable feature of the universe Computer Networks an Internets -- Moule 3 24 Spring, 2014 Copyright All rights reserve.

25 Examples When electrical signals propagate own a wire, electromagnetic energy is raiate (i.e., the wire acts like an antenna) When electromagnetic raiation encounters metal, a small electrical current is inuce that can interfere with signals being carrie on the wire When an electrical pulse is sent own an unterminate wire, reflection comes back When a signal passes across the connection between two wires, reflection an loss occur Note: a network iagnostic tool uses reflection to fin the istance to the point where a cable has been cut Computer Networks an Internets -- Moule 3 25 Spring, 2014 Copyright All rights reserve.

26 How Can We Reuce The Effect Of Noise On Copper Wiring Several techniques have been invente Unshiele Twiste Pair (UTP) Coaxial cable Shiele Twiste Pair (STP) All are use in computer networks Computer Networks an Internets -- Moule 3 26 Spring, 2014 Copyright All rights reserve.

27 How Twiste Pair Helps source of raiation ifference In an untwiste pair of wires, more current is generate in first wire the interference hits source of raiation ifference Twisting exposes each wire equally Computer Networks an Internets -- Moule 3 27 Spring, 2014 Copyright All rights reserve.

28 Coaxial Cable An Shieling Better protection: wrap a metal shiel aroun the wire outer plastic covering braie metal shiel plastic insulation inner wire for signal Shieling can be ae to twiste pair Aroun entire cable containing many pairs Aroun each pair as well as aroun cable Shieling etermines maximum ata rate Computer Networks an Internets -- Moule 3 28 Spring, 2014 Copyright All rights reserve.

29 Wiring Stanars An Data Rates Category Description Data Rate (in Mbps) CAT 1 Unshiele twiste pair use for telephones < 0.1 CAT 2 Unshiele twiste pair use for T1 ata 2 CAT 3 Improve CAT2 use for computer networks 10 CAT 4 Improve CAT3 use for Token Ring networks 20 CAT 5 Unshiele twiste pair use for networks 100 CAT 5E Extene CAT5 for more noise immunity 125 CAT 6 Unshiele twiste pair teste for 200 Mbps 200 CAT 7 Shiele twiste pair with a foil shiel 600 aroun the entire cable plus a shiel aroun each twiste pair What common ata rate is missing from the list? Computer Networks an Internets -- Moule 3 29 Spring, 2014 Copyright All rights reserve.

30 Meia Using Light Energy InfraRED transmission (short range an low ata rate) Point-to-point lasers (useful between builings) Optical fiber (high ata rate an long istance) Why light stays in a fiber: low ensity high ensity Refraction Absorption Reflection critical angle θ α α (a) (b) (c) Computer Networks an Internets -- Moule 3 30 Spring, 2014 Copyright All rights reserve.

31 Electromagnetic Spectrum An Properties Low frequencies Raio & TV Microwave InfraRe UV X ray Gamma ray 1 KHz 1 MHz 1 GHz 1 THz visible light Classification Range Type Of Propagation Low Wave follows earth s curvature, but < 2 MHz Frequency can be blocke by unlevel terrain Meium Wave can reflect from layers of the 2 to 30 MHz Frequency atmosphere, especially the ionosphere High Wave travels in a irect line, an will > 30 MHz Frequency be blocke by obstructions Computer Networks an Internets -- Moule 3 31 Spring, 2014 Copyright All rights reserve.

32 Satellite Communication Three types of communication satellites Orbit Type Description Low Has the avantage of low elay, but the isavantage Earth Orbit that from an observer s point of view on the earth, ( LEO ) the satellite appears to move across the sky Meium An elliptical (rather than circular) orbit primarily Earth Orbit use to provie communication at the North an ( MEO ) South Poles Geostationary Has the avantage that the satellite remains at a fixe Earth Orbit position with respect to a location on the earth s ( GEO ) surface, but the isavantage of being farther away Computer Networks an Internets -- Moule 3 32 Spring, 2014 Copyright All rights reserve.

33 GEO Satellites Figure below shows the earth s atmosphere rawn to scale Where woul a GEO satellite be in the figure? atmosphere Earth Computer Networks an Internets -- Moule 3 33 Spring, 2014 Copyright All rights reserve.

34 GEO Satellites (continue) Distance to GEO satellite is 35,785 km or 22,236 miles Approximately 3 times earth s iameter or one-tenth of the istance to the moon In other wors: the satellite is far off the page A consequence for networking: a long roun-trip time, even at the spee of light: Roun trip time = meters = sec meters/sec Computer Networks an Internets -- Moule 3 34 Spring, 2014 Copyright All rights reserve.

35 Measures Of Transmission Meia Propagation elay - time require for a signal to traverse a meium Channel capacity - maximum ata rate Computer Networks an Internets -- Moule 3 35 Spring, 2014 Copyright All rights reserve.

36 Channel Capacity Nyquist s Theorem gives theoretical boun on maximum ata rate for harware banwith B an K signal levels D = 2 B log 2 K Mathematical result known as Shannon s Theorem gives the maximum channel capacity, C, in the presence of noise C = B log 2 ( 1 + S/N) Quantity S / N is known as the signal-to-noise ratio Computer Networks an Internets -- Moule 3 36 Spring, 2014 Copyright All rights reserve.

37 Assessment Nyquist s Theorem gives us hope: using more signal levels can increase the ata rate Shannon s Theorem is sobering: electrical noise in the universe limits the effective channel capacity of any practical communication system Computer Networks an Internets -- Moule 3 37 Spring, 2014 Copyright All rights reserve.

38 Reliability An Channel Coing

39 Sources Of Errors An Types Error sources: interference, istortion, an attenuation Resulting error types: Type Of Error Description Single Bit Error A single bit in a block of bits is change an all other bits in the block are unchange (often results from very short-uration interference) Burst Error Multiple bits in a block of bits are change (often results from longer-uration interference) Erasure (Ambiguity) The signal that arrives at a receiver is ambiguous (oes not clearly correspon to either a logical 1 or a logical 0; can result from istortion or interference) Channel coing use to etect an correct errors Computer Networks an Internets -- Moule 3 39 Spring, 2014 Copyright All rights reserve.

40 Concept Of Forwar Error Correction (FEC) ORIGINAL MESSAGE ORIGINAL MESSAGE encoer ecoer a extra bits for protection check an optionally correct Discar output coewor receive coewor transmission over channel Examples: Single parity bit Row An Column (RAC) Cyclic Reunancy Check (CRC) Computer Networks an Internets -- Moule 3 40 Spring, 2014 Copyright All rights reserve.

41 Example: Row An Column Coe To sen 12 bits, arrange the bits in a matrix, compute a parity for each row an column, an sen 20 bits bits from atawor parity for each row parity for each column Receiver computes same parity for the 12 bits an compares to the parity bits receive single bit change uring transmission locations where calculate parity bits isagree, inicating the row an column of the error Computer Networks an Internets -- Moule 3 41 Spring, 2014 Copyright All rights reserve.

42 Hamming Distance Use to assess coe s resistance to errors Define to be number of bit changes to transform bit string S 1 into bit string S 2 Can be compute as number of 1 bits in the exclusive or of S 1 an S 2 To assess coe s strength, compute Hamming istance among all possible pairs of coewors, an take the minimum If minimum Hamming istance is n, an error that changes fewer than n bits will be etecte Computer Networks an Internets -- Moule 3 42 Spring, 2014 Copyright All rights reserve.

43 Internet Checksum Computation Given: A message, M, of arbitrary length Compute: A 16-bit 1s complement checksum, C Metho: Pa M to an exact multiple of 16 bits; Set a 32-bit checksum integer, C, to zero; for ( each 16-bit group in M ) { Treat the 16 bits as an integer an a to C; } Extract high-orer 16 bits of C an a to C; Checksum is inverse of the low-orer 16 bits; If the checksum is zero, substitute all 1s; Computer Networks an Internets -- Moule 3 43 Spring, 2014 Copyright All rights reserve.

44 Cyclic Reunancy Coe (CRC) Use with Ethernet an other high-spee networks Properties: Arbitrary Length Message As with a checksum, the size of a atawor is not fixe, which means a CRC can be applie to an arbitrary length message Excellent Error Detection Because the value compute epens on the sequence of bits in a message, a CRC provies excellent error etection capability Fast Harware Implementation Despite its sophisticate mathematical basis, a CRC computation can be carrie out extremely fast by harware Computer Networks an Internets -- Moule 3 44 Spring, 2014 Copyright All rights reserve.

45 Explanation Of CRC Mathematicians explain CRC computation as the remainer from polynomial ivision Theoretical computer scientists explain CRC as the remainer from a ivision of binary numbers Cryptographers explain CRC as an operation in a Galois fiel of orer 2 Computer programmers explain CRC as an algorithm that iterates through a message an uses table lookup Harware architects explain CRC computation as a small harware pipeline unit that uses exclusive or Computer Networks an Internets -- Moule 3 45 Spring, 2014 Copyright All rights reserve.

46 Question Can you explain the following? Fact 1: it is possible to write a function that computes the 32-bit CRC use with Ethernet Fact 2: commercial Ethernet proucts use harware instea of software to compute a CRC Computer Networks an Internets -- Moule 3 46 Spring, 2014 Copyright All rights reserve.

47 Transmission Moes

48 Terminology Serial - one bit at a time Parallel - multiple bits at a time Taxonomy of transmission methos: Transmission Moe Parallel Serial Asynchronous Synchronous Isochronous Computer Networks an Internets -- Moule 3 48 Spring, 2014 Copyright All rights reserve.

49 Serial Orering Of Bits An Bytes Both sies must agree on orer in which bits are transmitte Two approaches known as big-enian an little-enian Example: Ethernet uses byte big-enian an bit little-enian orer byte 1 byte 2 byte 3 byte 4 x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x Computer Networks an Internets -- Moule 3 49 Spring, 2014 Copyright All rights reserve.

50 Asynchronous An Synchronous Transmission Asynchronous: line ile when not in use; ata starts at arbitrary time voltage arbitrary arbitrary time -15 ile start stop ile Synchronous: each bit slot use voltage receiver must know how to group bits into bytes... 0 time Computer Networks an Internets -- Moule 3 50 Spring, 2014 Copyright All rights reserve.

51 Illustration Of Simplex An Duplex Moes sen receive (a) simplex sen receive receive (b) full-uplex sen sen receive receive (c) half-uplex sen Computer Networks an Internets -- Moule 3 51 Spring, 2014 Copyright All rights reserve.

52 Moulation An Demoulation

53 Illustration Of Amplitue Moulation carrier signal moulate carrier Computer Networks an Internets -- Moule 3 53 Spring, 2014 Copyright All rights reserve.

54 Illustration Of Frequency Moulation signal moulate carrier Computer Networks an Internets -- Moule 3 54 Spring, 2014 Copyright All rights reserve.

55 Shift Keying Like moulation except signal is igital carrier igital signal carrier with amplitue shift keying Computer Networks an Internets -- Moule 3 55 Spring, 2014 Copyright All rights reserve.

56 A Challenge Write a computer program that takes as input a series of points efining a signal an prouces plots of sine waves that show amplitue an frequency moulation as in the previous iagrams Computer Networks an Internets -- Moule 3 56 Spring, 2014 Copyright All rights reserve.

57 Other Moulation Topics Phase shift moulation Increasing bits per secon by combining amplitue an phase shift (QAM techniques) Constellation iagrams to represent combinations Moems (moulator / emoulator) Computer Networks an Internets -- Moule 3 57 Spring, 2014 Copyright All rights reserve.

58 Multiplexing An Demultiplexing (Channelization)

59 Concept Of Multiplexing An Types multiplexor emultiplexor sener 1 share meium receiver 1 sener 2. sener N receiver 2. receiver N Types: Frequency ivision multiplexing Wavelength ivision multiplexing Time ivision multiplexing Coe ivision multiplexing Computer Networks an Internets -- Moule 3 59 Spring, 2014 Copyright All rights reserve.

60 Frequency Division Multiplexing (FDM) Use in broacast raio an cable TV multiplexor emultiplexor sener 1 sener 2. sener N channel 1 channel 2. channel N receiver 1 receiver 2. receiver N Demultiplexing implemente with sets of filters frequencies for all channels emultiplexor filter 1 each output has frequencies for one channel filter 2. filter N Computer Networks an Internets -- Moule 3 60 Spring, 2014 Copyright All rights reserve.

61 FDM In Practice Each channel assigne a range of frequencies Channel Frequencies Use KHz KHz KHz KHz KHz KHz KHz KHz KHz KHz KHz KHz A guar ban separates ajacent channels guar ban KHz Computer Networks an Internets -- Moule 3 61 Spring, 2014 Copyright All rights reserve.

62 Wavelength Division Multiplexing (WDM) Form of FDM use with light (i.e., on an optical fiber) Separate frequencies calle colors or lambas Prisms use to separate frequencies prism optical fiber carrying a beam of light λ 1 λ 2 λ 1 λ 2 λ k λ k Current technology is Dense WDM (DWDM); an iniviual channel can provie 10 Gbps Computer Networks an Internets -- Moule 3 62 Spring, 2014 Copyright All rights reserve.

63 Time Division Multiplexing Seners take turns transmitting multiplexor emultiplexor sener 1 ata flow receiver 1 sener N receiver 2.. sener N receiver N Synchronous TDM Each sener assigne a slot (typically roun-robin) Use by the telephone company Statistical TDM Sener only transmits when reay (e.g., Ethernet) Computer Networks an Internets -- Moule 3 63 Spring, 2014 Copyright All rights reserve.

64 Coe Division Multiplexing Mathematical form of multiplexing use with cell phones Algorithm Each sener/receiver pair is assigne a unique number calle a chip sequence Seners multiply the ata value by their chip sequence (orthogonal vector spaces) Transmitte value is a sum of all seners Each receiver multiplies incoming value by its chip sequence to extract ata Avantage over statistical TDM: lower elay when network loae Computer Networks an Internets -- Moule 3 64 Spring, 2014 Copyright All rights reserve.

65 Hierarchical Multiplexing Hierarchies use with FDM an TDM to combine multiple lower-capacity channels Example of TDM hierarchy use by the phone system 7 DS-2 igital phone channels (6.312 Mbps each) 6 DS-3 igital phone channels ( Mbps each) 4 DS-1 igital phone channels (1.544 Mbps each) 24 DS-0 igital phone channels (64 Kbps each) 1 DS-4 igital phone channel ( Mbps total) Computer Networks an Internets -- Moule 3 65 Spring, 2014 Copyright All rights reserve.

66 Inverse Multiplexing Divies ata from a single channel into several lower-spee channels Use when high-spee channel is unavailable or too expensive Some ISPs use inverse multiplexing to combine several 10 Gbps channels into a higher-spee channel single high-spee input single high-spee output multiple low-spee connections Computer Networks an Internets -- Moule 3 66 Spring, 2014 Copyright All rights reserve.

67 Summary Data communications eals with the Physical Layer an ata transmission Concepts inclue Signals an conversion between igital an analog Transmission meia Reliability an channel coing Moulation an emoulation Multiplexing an emultiplexing Computer Networks an Internets -- Moule 3 67 Spring, 2014 Copyright All rights reserve.