MSIT 413: Wireless Technologies Week 4

MSIT 413: Wireless Technologies Week 4 Michael L. Honig Department of EECS Northwestern University February 2014 1

Outline Finish radio propagation Applications: location tracking (radar), handoffs Digital modulation 2

Radio Channels Troposcatter Microwave LOS T T Mobile radio Indoor radio 3

Power Attenuation distance d reference distance d 0 =1 Reference power at reference distance d 0 Path loss exponent In db: P r = P 0 (db) 10 n log (d) P 0 slope (n=2) = -20 db per decade P r (db) slope = -40 (n=4) log (d) 0 4

Link Budget How much transmit power is required to achieve a target received power? dbs add: Target received power (dbm) + path loss (db) + other losses (components) (db) - antenna gains (db) Total power needed at transmitter (dbm) 5

Example Transmitter What is the required Transmit power? wireless channel 40 db attenuation Receiver Received power must be > -30 dbm Recall that dbm measures the signal power relative to 1 mw (milliwatt) = 0.001 Watt. To convert from S Watts to dbm, use S (dbm) = 10 log (S / 0.001) Transmitted power (dbm) = -30 + 40 = 10 dbm = 10 mw What if the received signal-to-noise ratio must be 5 db, and the noise power is -45 dbm? 6

Shadow Fading Random variations in path loss as mobile moves around buildings, trees, etc. Modeled as an additional random variable: normal (Gaussian) probability distribution P r = P 0 10 n log d + X standard deviation log-normal random variable -σ σ received power in db For cellular: σ is about 8 db 7

Large-Scale Path Loss (Scatter Plot) Most points are less than σ db from the mean 8

Urban Multipath No direct Line of Sight between mobile and base Radio wave scatters off of buildings, cars, etc. Severe multipath 9

Narrowband Fading Received signal r(t) = h 1 s(t - τ 1 ) + h 2 s(t - τ 2 ) + h 3 s(t - τ 3 ) + attenuation for path 1 (random) delay for path 1 (random) If the transmitted signal is sinusoidal (narrowband), s(t) = sin 2πf t, then the received signal is also sinusoidal, but with a different (random) amplitude and (random) phase: r(t) = A sin (2πf t + θ) Transmitted s(t) Received r(t) A, θ depend on environment, location of transmitter/receiver 10

Rayleigh Fading Can show: A has a Rayleigh distribution θ has a uniform distribution (all phase shifts are equally likely) Probability (A < a) = 1 e -a2 /P0 where P 0 is the reference power (averaged over different locations) 1 Prob(A < a) 1-e -a2 /P 0 Ex: P 0 =1, a=1: Pr(A<1) = 1 e -1 = 0.63 (probability that signal is faded) P 0 = 1, a=0.1: Pr(A<0.1) = 1 e -1/100 0.01 (prob that signal is severely faded) 11 a

Small-Scale Fading Fade rate depends on Mobile speed Speed of surrounding objects Frequency 12

Short- vs. Long-Term Fading Short-term fading Signal Strength (db) T T Long-term fading Time (t) Long-term (large-scale) fading: Distance attenuation Shadowing (blocked Line of Sight (LOS)) Variations of signal strength over distances on the order of many wavelengths

Combined Fading and Attenuation Received power P r (db) distance attenuation Time (mobile is moving away from base)

Combined Fading and Attenuation Received power P r (db) distance attenuation shadowing Large-scale effects Time (mobile is moving away from base) 15

Combined Fading and Attenuation Received power P r (db) distance attenuation shadowing Rayleigh fading Small-scale effect Time (mobile is moving away from base) 16

Example Diagnostic Measurements: 1XEV-DO drive test measurements drive path

Time Variations: Doppler Shift Audio clip (train station) 18

Time Variations: Doppler Shift velocity v distance d = v t Propagation delay = distance d / speed of light c = vt/c transmitted signal s(t) delay increases received signal r(t) propagation delay Received signal r(t) = sin 2πf (t- vt/c) = sin 2π(f fv/c) t Doppler shift f d = -fv/c received frequency 19

Doppler Shift (Ex) Mobile moving away from base è v > 0, Doppler shift < 0 Mobile moving towards base è v < 0, Doppler shift > 0 Carrier frequency f = 900 MHz, v = 60 miles/hour = 26.82 meters/sec Mobile à Base: f d = fv/c = (900 10 6 ) 26.82 / (3 10 8 ) 80 Hz meters/sec 20

Doppler (Frequency) Shift ½ Doppler cycle in phase out of phase Frequency= 1/50 Frequency= 1/45 21

Application of Doppler Shift: Astronomy Doppler shift determines relative velocity of distant objects (e.g., stars, galaxies ) red shift : object is moving away Observed spectral lines (radiation is emitted at discrete frequencies) blue shift object is moving closer sun light spectrum spectrum of galaxy supercluster

Application of Doppler Shift: Police Radar Doppler shift can be used to compute relative speed. 23

Scattering: Doppler Spectrum distance d = v t transmitted signal s(t) received signal?? power Received signal is the sum of all scattered waves freq. Doppler shift for each path depends on angle (vf cos θ/c ) frequency of s(t) Typically assume that the received energy is the same from all directions (uniform scattering) 24

Scattering: Doppler Spectrum distance d = v t transmitted signal s(t) power Doppler shift f d Doppler Spectrum (shows relative strengths of Doppler shifts) power 2f d frequency of s(t) frequency frequency frequency of s(t) + Doppler shift f d 25

Scattering: Doppler Spectrum transmitted signal s(t) distance d = v t power frequency of s(t) frequency power Doppler spectrum 2f d frequency of s(t) + Doppler shift f d

Rayleigh Fading phase shift deep fade Received waveform Amplitude (db) 27

Channel Coherence Time relative amplitude (db) Coherence Time: Amplitude and phase are nearly constant. Rate of time variations depends on Doppler shift: (velocity x carrier frequency)/(speed of light) Coherence Time varies as 1/(Doppler shift). time 28

Fast vs. Slow Fading received amplitude transmitted bits coherence time time Fast fading: channel changes every few symbols. Coherence time is less than roughly 100 symbols. time Slow fading: Coherence time lasts more than a few 100 symbols. 29

Fast vs. Slow Fading received amplitude transmitted bits coherence time time time What is important is the coherence time (1/Doppler) relative to the Data rate. 30

Fade Rate (Ex) f c = 900 MHz, v = 60 miles/hour è Doppler shift 80 Hz. Coherence time is roughly 1/80, or 10 msec Data rate (voice): 10 kbps or 0.1 msec/bit à 100 bits within a coherence time (fast fading) GSM data rate: 270 kbps à about 3000 bits within a coherence time (slow fading) 31

Channel Characterizations: Narrowband vs. Wideband Narrowband signal (sinusoid) infinite duration, zero bandwidth Multipath channel Amplitude attenuation, Delay (phase shift) delay spread Wideband signal (impulse) s(t) time t zero duration, infinite bandwidth Multipath channel r(t) multipath components time t 32

Pulse Width vs. Bandwidth signal pulse Narrowband Power bandwidth = 1/T T time frequency signal pulse Wideband Power bandwidth = 1/T T time frequency 33

Power-Delay Profile Received power vs. time in response to a transmitted short pulse. delay spread τ For cellular systems (outdoors), the delay spread is typically a few microseconds. 34

Two-Ray Impulse Response reflection (path 2) direct path (path 1) s(t) r(t) reflection is attenuated τ time t τ = [(length of path 2) (length of path 1)]/c time t 35

Urban Multipath s(t) r(t) time t r(t) different location for receiver Spacing and attenuation of multipath components depend on location and environment. 36 time t time t

Delay Spread and Intersymbol Interference s(t) r(t) time t Multipath channel time t Time between pulses is >> delay spread, therefore the received pulses do not interfere. r(t) s(t) Multipath channel time t Time between pulses is < delay spread, which causes intersymbol interference. The rate at which symbols can be transmitted without intersymbol interference is 1 / delay spread. 37

Coherence Bandwidth channel gain coherence bandwidth B c Frequencies far outside the coherence bandwidth are affected differently by multipath. f 1 f 2 frequency The channel gain is approximately constant within a coherence bandwidth B c. Frequencies f 1 and f 2 fade independently if f 1 f 2 >> B c. 38

Coherence Bandwidth and Delay Spread delay spread τ channel gain coherence bandwidth B c delay spread τ channel gain frequency coherence bandwidth B c frequency Coherence bandwidth is inversely proportional to delay spread: 39 B c 1/τ.

Narrowband Signal channel gain signal power (narrowband) coherence bandwidth B c Frequencies far outside the coherence bandwidth are affected differently by multipath. f 1 f 2 frequency The signal power is confined within a coherence band. Flat fading: all signal frequencies are affected the same way. 40

Wideband Signals channel gain signal power (wideband) coherence bandwidth B c Frequencies far outside the coherence bandwidth are affected differently by multipath. f 1 f 2 frequency A wideband signal spans many coherence bands. Frequency-selective fading: different parts of the signal (in frequency) are affected differently by the channel. 41

Frequency Diversity channel gain signal power (wideband) coherence bandwidth B c Frequencies far outside the coherence bandwidth are affected differently by multipath. f 1 f 2 frequency Wideband signals exploit frequency diversity. Spreading power across many coherence bands reduces the chances of severe fading. Wideband signals are distorted by the channel fading (distortion causes intersymbol interference). 42

Coherence Bandwidth for Cellular channel gain signal power (wideband) coherence bandwidth B c Frequencies far outside the coherence bandwidth are affected differently by multipath. f 1 f 2 frequency For the cellular band, B c is around 100 to 300 khz. How does this compare with the bandwidth of cellular systems? 43

Fading Experienced by Wireless Systems Standard Bandwidth Fade rate AMPS 30 khz (NB) Fast IS-136 30 khz Fast GSM 200 khz Slow IS-95 (CDMA) 1.25 MHz (WB) Fast 3G 1.25-5 MHz Slow to Fast (depends on rate) LTE up to 20 MHz Slow 802.11 > 20 MHz Slow Bluetooth > 5 MHz (?) Slow 44

Pulse Width vs. Bandwidth signal pulse Narrowband Power bandwidth = 1/T T time frequency signal pulse Wideband Power bandwidth = 1/T T time frequency 45

Radar Pulse Bandwidth reflection s(t) delay τ = 2 x distance/c s(t) delay τ r(t) time t r(t) Narrow bandwidth pulse time t High bandwidth pulse 46

Bandwidth and Resolution delay τ = 2 x distance/c reflection s(t) r(t) The resolution of the delay measurement is roughly the width of the pulse. Low bandwidth è wide pulse è low resolution High bandwidth è narrow pulse è high resolution time t If the delay measurement changes by 1 microsec, the distance error is c x 10-6 /2 = 150 meters! 47

Propagation and Handoff Received Signal Strength (RSS) from right BST from left BST unacceptable (call is dropped) time 48

Propagation and Handoff Received Signal Strength (RSS) handoff threshold from right BST with handoff from left BST unacceptable (call is dropped) time 49

Propagation and Handoff Received Signal Strength (RSS) handoff threshold RSS margin time needed for handoff from right BST with handoff from left BST unacceptable (call is dropped) time 50

Propagation and Handoff Received Signal Strength (RSS) handoff threshold RSS margin time needed for handoff from right BST from left BST unacceptable (call is dropped) time 51

Handoff Threshold Received Signal Strength (RSS) handoff threshold RSS margin time needed for handoff from right BST from left BST unacceptable (call is dropped) time Handoff threshold too high è too many handoffs (ping pong) Handoff threshold too low è dropped calls are likely Threshold should depend on slope on vehicle speed (Doppler). 52

Handoff Measurements (3G) Mobile maintains a list of neighbor cells to monitor. Mobile periodically measures signal strength from BST pilot signals. Mobile sends measurements to network to request handoff. Handoff decision is made by network. Depends on available resources (e.g., channels/time slots/codes). Handoffs take priority over new requests (why?). Hysteresis needed to avoid handoffs due to rapid variations in signal strength. Received Signal Strength (RSS) handoff threshold unacceptable (call is dropped) time 57

Handoff Decision Depends on RSS, time to execute handoff, hysteresis, and dwell (duration of RSS) Proprietary methods Handoff may also be initiated for balancing traffic. 1G (AMPS): Network Controlled Handoff (NCHO) Handoff is based on measurements at BS, supervised by MSC. 2G, GPRS, 3G: Mobile Assisted Handoff (MAHO) Handoff relies on measurements at mobile Enables faster handoff Mobile data, WLANs (802.11): Mobile Controlled Handoff (MCHO) Handoff controlled by mobile 58

Example Diagnostic Measurements: 1XEV-DO drive test measurements drive path 59

Why Digital Communications? 1G (analog) à 2G (digital) à 3G (digital) Digitized voice requires about 64 kbps, therefore the required bandwidth is >> the bandwidth of the voice signal (3 4 khz)! 60

Why Digital Communications? 1G (analog) à 2G (digital) à 3G (digital) Digitized voice requires about 64 kbps, therefore the required bandwidth is >> the bandwidth of the voice signal (3 4 khz)! Can combine with sophisticated signal processing (voice compression) and error protection. Greater immunity to noise/channel impairments. Can multiplex different traffic (voice, data, video). Security through digital encryption. Flexible design possible (software radio). VLSI + special purpose digital signal processing à digital is more cost-effective than analog! 61

Binary Frequency-Shift Keying (FSK) Bits: 1 0 1 1 0 Amplitude time 62

Quadrature Phase Shift Keying (QPSK) Bits: 00 01 10 11 Amplitude time 63

Binary Phase Shift Keying (BPSK) Bits: 1 0 1 1 0 Baseband signal Amplitude time 64

Amplitude Shift Keying (4-Level ASK) Bits: 00 01 10 11 Baseband signal Amplitude symbol duration time 65

Baseband à RF Conversion Baseband signal sin 2πf c t Passband (RF) signal T time X Modulate to the carrier frequency f c Power signal bandwidth is roughly 2/T Power 0 frequency ß 0 f c frequency 66

Why Modulate? 67

Why Modulate? The baseband spectrum is centered around f=0. Without modulation all signals would occupy low frequencies and interfere with each other. It is difficult to build effective antennas at low frequencies since the dimension should be on the order of a wavelength. Low frequencies propagate further, causing more interference. 68

Selection Criteria How do we decide on which modulation technique to use? 69

Selection Criteria How do we decide on which modulation technique to use? Performance: probability of error P e. Probability that a 0 (1) is transmitted and the receiver decodes as a 1 (0). Complexity: how difficult is it for the receiver to recover the bits (demodulate)? FSK was used in early voiceband modems because it is simple to implement. Bandwidth or spectral efficiency: bandwidth (B) needed to accommodate data rate R bps, i.e., R/B measured in bits per second per Hz. Power efficiency: energy needed per bit to achieve a satisfactory P e. Performance in the presence of fading, multipath, and interference. 70

Example: Binary vs. 4-Level ASK 3A A A -A -A -3A Rate = 1/T symbols/sec Bandwidth is roughly 1/T Hz Bandwidth efficiency = 1 bps/hz Rate = 2/T symbols/sec Bandwidth is roughly 1/T Hz Bandwidth efficiency = 2 bps/hz What about power efficiency? 71

Noisy Baseband Signals 3A A A -A -A -3A Rate = 1/T symbols/sec Bandwidth is roughly 1/T Hz Bandwidth efficiency = 1 bps/hz Power =A 2 (amplitude squared). Rate = 2/T symbols/sec Bandwidth is roughly 1/T Bandwidth efficiency = 2 bps/hz Power = (A 2 + 9A 2 )/2 = 5A 2 What about probability of error vs transmitted power? 72

Probability of Error Log of Probability of Error BPSK 4-ASK 7 db (factor of 5) Signal-to-Noise Ratio (db) 73

How to Increase Bandwidth Efficiency? 74

How to Increase Bandwidth Efficiency? Increase number of signal levels. Use more bandwidth efficient modulation scheme (e.g., PSK). Apply coding techniques: protect against errors by adding redundant bits. Note that reducing T increases the symbol rate, but also increases the signal bandwidth. There is a fundamental tradeoff between power efficiency and bandwidth efficiency. 75

The Fundamental Question Given: B Hz of bandwidth S Watts of transmitted signal power N Watts per Hz of background noise (or interference) power What is the maximum achievable data rate? (Note: depends on P e.) 76

Claude Shannon (1916-2001) Father of Information Theory Shannon s 1948 paper A Mathematical Theory of Communications laid the foundations for modern communications and networking: The fundamental problem of communication is that of reproducing at one point either exactly or approximately a message selected at another point transmitter receiver 77

Claude Shannon (1916-2001) Father of Information Theory Shannon s 1948 paper A Mathematical Theory of Communications laid the foundations for modern communications and networking: The significant aspect is that the actual message is one selected from a set of possible messages. The system must be designed to operate for each possible selection, not just the one which will actually be chosen since this is unknown at the time of design. transmitter receiver 78

Claude Shannon (1916-2001) Father of Information Theory Shannon s 1948 paper A Mathematical Theory of Communications laid the foundations for modern communications and networking: The choice of a logarithm base corresponds to the choice of a unit for measuring information. If the base 2 is used the resulting units may be called binary digits, or more briefly bits, a word suggested by J. W. Tukey. log 2 M bits Transmitter (M possible messages) receiver 79

Claude Shannon (1916-2001) Father of Information Theory Shannon s 1948 paper A Mathematical Theory of Communications laid the foundations for modern communications and networking. Other contributions and interests: digital circuits, genetics, cryptography, investing, chess-playing computer, roulette prediction, maze-solving, unicycle designs, juggling Videos: Father of the Information Age Juggling video 80

Shannon s Channel Coding Theorem (1948) noise Information Source bits Encoder input x(t) Channel output y(t) Decoder Estimated bits Information rate: Channel capacity: R bits/second C bits/second R < C è There exists an encoder/decoder combination that achieves arbitrarily low error probability. R > C è The error probability cannot be made small. 81

Shannon Capacity noise Information Source bits Encoder input x(t) Channel output y(t) Decoder Estimated bits Channel capacity: C = B log(1+s/n) bits/second B= Bandwidth, S= Signal Power, N= Noise Power No fading 82

Caveats There exists does not address complexity issues. As the rate approaches Shannon capacity, to achieve small error rates, the transmitter and (especially) the receiver are required to do more and more computations. The theorem does not say anything about delay. To achieve Shannon capacity the length of the transmitted code words must tend to infinity! The previous formula does not apply with fading, multipath, frequency-selective attenuation. It has taken communications engineers more than 50 years to find practical coding and decoding techniques, which can achieve information rates close to the Shannon capacity. 83

Example: GSM/EDGE Bandwidth = 200 khz, S/I = 9 db = 7.943 è C = 200,000 x log(8.943) 632 kbps This is what would be achievable in the absence of fading, multipath, etc. Currently, EDGE provides throughputs of about 230 kbps. Up to 470 kbps possible using additional tricks, such as adapting the modulation and coding format to match the channel Ø Preceding Shannon formula is not directly applicable. 84

Data Rates for Deep Space Applications Mariner: 1969 (Mars) Pioneer 10/11: 1972/3 (Jupiter/Saturn fly-by) Voyager: 1977 (Jupiter and Saturn) Planetary Standard: 1980 s (military satellite) BVD: Big Viterbi Decoder Galileo: 1992 (Jupiter) (uses BVD) Turbo Code: 1993 Signal to Noise Ratio 86