1. July, 19982 - 1RF100 (c) 1998 Scott Baxter Wireless Systems: Modulation Schemes and Bandwidth Wireless Systems: Modulation Schemes and Bandwidth RF100 Chapter 2 fcfc fcfc Upper Sideband Lower Sideband fcfc fcfc I axis Q axis a b  c QPSK I axis Q axis c a  b p r v  /4 shifted DQPSK 1 0

2. July, 19982 - 2RF100 (c) 1998 Scott Baxter Characteristics of a Radio Signal nThe purpose of telecommunications is to send information from one place to another nOur civilization exploits the transmissible nature of radio signals, using them in a sense as our “carrier pigeons” nTo convey information, some characteristic of the radio signal must be altered (I.e., ‘modulated’) to represent the information nThe sender and receiver must have a consistent understanding of what the variations mean to each other “one if by land, two if by sea” nThree commonly-used RF signal characteristics which can be varied for information transmission: Amplitude Frequency Phase SIGNAL CHARACTERISTICS S (t) = A cos [  c t +  ] The complete, time- varying radio signal Amplitude (strength) of the signal Natural Frequency of the signal Phase of the signal Compare these Signals: Different Amplitudes Different Frequencies Different Phases

3. July, 19982 - 3RF100 (c) 1998 Scott Baxter AM: Our First “Toehold” for Transmission nThe early radio pioneers could only turn their crude transmitters on and off. They could form the dots and dashes of Morse code. The first successful radio experiments happened during the mid-1890’s by experimenters in Italy, England, Kentucky, and elsewhere. nBy 1910, vacuum tubes gave experimenters better control over RF power generation. RF power could now be linearly modulated in step with sound vibrations. Voices and music could now be transmitted!! Still, nobody anticipated FM, PM, or digital signals. nCommercial public AM broadcasting began in the early 1920’s. nDespite its disadvantages and antiquity, AM is still alive: AM broadcasting continues today in 540-1600 KHz. AM modulation remains the international civil aviation standard, used by all commercial aircraft (108-132 MHz. band). AM modulation is used for the visual portion of commercial television signals (sound portion carried by FM modulation) Citizens Band (“CB”) radios use AM modulation Special variations of AM featuring single or independent sidebands, with carrier suppressed or attenuated, are used for marine, commercial, military, and amateur communications SSB LSB USB

4. July, 19982 - 4RF100 (c) 1998 Scott Baxter Amplitude Modulation (“AM”) Details TIME-DOMAIN VIEW of AM MODULATOR x(t) = [1 + a m n (t)]cos  c t where: a = modulation index (0 < a <= 1) m n (t) = modulating waveform  c = 2  f c, the radian carrier freq.  a 1 + + x(t) cos  c m n (t) nAM is “linear modulation” -- the spectrum of the baseband signal translates directly into sidebands on both sides of the carrier frequency nDespite its simplicity, AM has definite drawbacks which complicate its use for wireless systems: Only part of an AM signal’s energy actually carries information (sidebands); the rest is the carrier The two identical sidebands waste bandwidth AM signals can be faithfully amplified only by linear amplifiers AM is highly vulnerable to external noise during transmission AM requires a very high C/I (~30 to 40 dB); otherwise, interference is objectionable FREQUENCY-DOMAIN VIEW Voltage Frequency 0 fcfc m n (t) BASEBAND x(t) UPPER SIDEBAND LOWER SIDEBAND CARRIER

5. July, 19982 - 5RF100 (c) 1998 Scott Baxter Circuits to Generate & Detect AM Signals nAM modulation can be simply accomplished in a saturated amplifier superimpose the modulating waveform on the supply voltage of the saturated amplifier nAM de-modulation (detection) can be easily performed using a simple envelope detector example: half-wave rectifier this “non-coherent” detection works well if S/N >10 dB. nAM demodulation can also be performed by coherent detectors incoming signal is mixed (multiplied) with a locally generated carrier enhances performance when S/N ratio is poor (<10 dB.) TIME-DOMAIN VIEW: AM MODULATOR x(t)   cos  c m n (t) [1 + a m n (t)] Sat. Lin. TIME-DOMAIN VIEW: AM DETECTOR (non-coherent) x(t)  m n (t) information RF carrier Modulated signal

6. July, 19982 - 6RF100 (c) 1998 Scott Baxter Better Quality: Frequency Modulation (“FM”) nFrequency Modulation (FM) is a type of angle modulation in FM, the instantaneous frequency of the signal is varied by the modulating waveform nAdvantages of FM the amplitude is constant –simple saturated amplifiers can be used –the signal is relatively immune to external noise –the signal is relatively robust; required C/I values are typically 17-18 dB. in wireless applications nDisadvantages of FM relatively complex detectors are required a large number of sidebands are produced, requiring even larger bandwidth than AM TIME-DOMAIN VIEW s FM (t) =A cos [  c t + m  m(x)dx+   ] t t0t0 where: A = signal amplitude (constant)  c = radian carrier frequency m   frequency deviation index m(x) = modulating signal   = initial phase FREQUENCY-DOMAIN VIEW Voltage Frequency 0 fcfc S FM (t) UPPER SIDEBANDS LOWER SIDEBANDS

7. July, 19982 - 7RF100 (c) 1998 Scott Baxter Circuits to Generate and Detect FM Signals nOne way to build an FM signal is a voltage-controlled oscillator the modulating signal varies a reactance (varactor, etc.) or otherwise changes the frequency of the oscillator the modulation may be performed at a low intermediate frequency, then heterodyned to a desired communications frequency nFM de-modulation (detection) can be performed by any of several types of detectors Phase-locked loop (PLL) Pulse shaper and integrator Ratio Detector TIME-DOMAIN VIEW: FM MODULATOR s FM (t) m(x) ~ VCO x LO HPA TIME-DOMAIN VIEW: FM DETECTOR x LO LNA PLL s FM (t) m(x) information FM modulated signal

8. July, 19982 - 8RF100 (c) 1998 Scott Baxter The Inventor of FM Major Edwin H. Armstrong was one of the most famous inventors in the early history of radio. In 1918, he invented the superheterodyne circuit -- and implemented the basic mixing principle of heterodyne frequency conversion used in virtually all modern radio receivers. Others got the credit. In 1933, he invented wide-band frequency modulation. Armstrong’s primary motivation was to improve the audio quality of broadcast transmission, which had suffered from noise and static because it used AM modulation. Promotion and commercial development of FM placed Armstrong in competition with David Sarnoff and Radio Corporation of America. Sarnoff and RCA were promoting television, and worried Armstrong’s FM would compete with TV and slow its public acceptance. Mainly due to RCA influence, the US FCC decided to change the frequencies allocated for FM broadcasting, obsoleting hundreds of FM transmitters and 500,000+ home receivers Armstrong had helped finance as an FM demonstration. In 1954, despondent over these setbacks, Armstrong took his life. But today, the technology he started is used not only in broadcasting and the sound portion of TV, but also in land mobile and first-generation analog cellular systems. Edwin Howard Armstrong 1890 - 1954

9. July, 19982 - 9RF100 (c) 1998 Scott Baxter Sister of FM: Phase Modulation (“PM”) nPhase Modulation (PM) is a type of angle modulation, closely related to FM the instantaneous phase of the signal is varied according to the modulating waveform nAdvantages of PM: very similar to FM the amplitude is constant –simple saturated amplifiers can be used –the signal is relatively immune to external noise –the signal is relatively robust; required C/I values are typically 17-18 dB. in wireless applications nDisadvantages of PM relatively complex detectors are required, just like FM a large number of sidebands are produced, just like FM, requiring even larger bandwidth than AM TIME-DOMAIN VIEW s PM (t) =A cos [  c t + m  m(x) +   ] where: A = signal amplitude (constant)  c = radian carrier frequency m   phase deviation index m(x) = modulating signal   = initial phase FREQUENCY-DOMAIN VIEW Voltage Frequency 0 fcfc S FM (t) UPPER SIDEBANDS LOWER SIDEBANDS information Phase-modulated signal

10. July, 19982 - 10RF100 (c) 1998 Scott Baxter Circuits to Generate and Detect PM Signals nPM and FM signals are identical with only one exception: in FM, the analog modulating signal is inherently de- emphasized by 1/F nConsequences of this realization: the same types of circuitry can be used to generate and detect both analog PM or FM, determined by filtering the modulating signal at baseband FM has poorer signal-to-noise ratio than PM at high modulating frequencies. Therefore, pre- emphasis and de-emphasis are often used in FM systems TIME-DOMAIN VIEW: FM DETECTOR FOR PM x LO LNA PLL s FM (t) m(x) The phase of an FM signal is proportional to the integral of the amplitude of the modulating signal. The phase of a PM signal is proportional to the amplitude of the modulating signal. TIME-DOMAIN VIEW: PHASE MODULATOR s FM (t) m(x) ~ Phase Shifter x LO HPA information Phase-modulated signal

11. July, 19982 - 11RF100 (c) 1998 Scott Baxter How Much Bandwidth do Signals Occupy? nThe bandwidth occupied by a signal depends on: input information bandwidth modulation method nInformation to be transmitted, called “input” or “baseband” bandwidth usually is small, much lower than frequency of carrier nUnmodulated carrier the carrier itself has Zero bandwidth!! nAM-modulated carrier Notice the upper & lower sidebands total bandwidth = 2 x baseband nFM-modulated carrier Many sidebands! bandwidth is a complex Bessel function Carson’s Rule approximation 2(F+D) nPM-modulated carrier Many sidebands! bandwidth is a complex Bessel function Voltage Time Time-Domain (as viewed on an Oscilloscope) Frequency-Domain (as viewed on a Spectrum Analyzer) Voltage Frequency 0 fcfc fcfc Upper Sideband Lower Sideband fcfc fcfc

12. July, 19982 - 12RF100 (c) 1998 Scott Baxter Digital Sampling and Vocoding

13. July, 19982 - 13RF100 (c) 1998 Scott Baxter Introduction to Digital Modulation nThe modulating signals shown in previous slides were all analog. It is also possible to quantize modulating signals, restricting them to discrete values, and use such signals to perform digital modulation. Digital modulation has several advantages over analog modulation: nDigital signals can be more easily regenerated than analog in analog systems, the effects of noise and distortion are cumulative: each demodulation and remodulation introduces new noise and distortion, added to the noise and distortion from previous demodulations/remodulations. in digital systems, each demodulation and remodulation produces a clean output signal free of past noise and distortion nDigital bit streams are ideally suited to multiplexing - carrying multiple streams of information intermixed using time-sharing transmission demodulation-remodulation transmission demodulation-remodulation transmission demodulation-remodulation

14. July, 19982 - 14RF100 (c) 1998 Scott Baxter Theory of Digital Modulation: Sampling nVoice and other analog signals first must be converted to digital form (“sampled”) before they can be transmitted digitally nThe sampling theorem gives the requirements for successful sampling The signal must be sampled at least twice during each cycle of f M, its highest frequency. 2 x f M is called the Nyquist Rate. to prevent “aliasing”, the analog signal is low-pass filtered so it contains no frequencies above f M nRequired Bandwidth for Samples, p(t) If each sample p(t) is expressed as an n-bit binary number, the bandwidth required to convey p(t) as a digital signal is at least N*2* f M this follows Shannon’s Theorem: at least one Hertz of bandwidth is required to convey one bit per second of data Notice: lots of bandwidth required! The Sampling Theorem: Two Parts If the signal contains no frequency higher than f M Hz., it is completely described by specifying its samples taken at instants of time spaced 1/2 f M s. The signal can be completely recovered from its samples taken at the rate of 2 f M samples per second or higher. m(t) Sampling Recovery m(t) p(t)

15. July, 19982 - 15RF100 (c) 1998 Scott Baxter The Mother of All Telephone Signals: DS-0 nTelephony has adopted a world-wide PCM standard digital signal, using a 64 kb/s stream derived from sampled voice data nVoice waveforms are band-limited (see curve) upper cutoff beyond 3500-4000 Hz. to avoid aliasing rolloff below 300 Hz. For less sensitivity to “hum” picked up from AC power mains nVoice waveforms sampled 8000 times/second A>D conversion has 1 byte (8 bit) resolution; thus 256 voltage levels possible 8000 samples x 1 byte = 64,000 bits/second Levels are defined logarithmically rather than linearly, to handle a wider range of audio levels with minimum distortion –  -law companding is used in North America & Japan – A-law companding is used in most other countries -10dB -20dB -30dB -40dB 0 dB 1003001000300010000 Frequency, Hz C-Message Weighting t 0 1 2 3 4 5 6 8 7 9 10 11 12 13 14 15 16 4 1 3 15 8 3 4 8 A-LAW y  sgn(x) A|x| ln(1  A) for0  x  1 A y  sgn(x) ln(1  A|x)| (1  A) for 1 A  x  1 µ-Law y  sgn(x) ln(1  |x|) (1  ) (where  255) Companding Band-Limiting x = analog audio voltage y = quantized level (digital)

16. July, 19982 - 16RF100 (c) 1998 Scott Baxter Was Digital Supposed to Give More Capacity!? nA DS-0 telephone signal, carrying one person talking, is a 64,000 bits/second data stream. nShannon’s theorem tells us we’ll need at least 64,000 Hz. of bandwidth to carry this signal, even with the most advanced modulation techniques (QPSK, etc.) nBut regular analog cellular signals are only 30,000 Hz. wide! So does a digital signal require more bandwidth than analog?!! nYES -- unless we do something fancy, like compression. nWe DO use compression, to reduce the number of bits being transmitted, thereby keeping the bandwidth as small as we can nThe compressing device is called a Vocoder (voice coder). It both compresses the signal being sent, and expands the signal being received nEvery digital mobile phone technology uses some type of Vocoder There are many types, with many different characteristics

17. July, 19982 - 17RF100 (c) 1998 Scott Baxter Vocoders: Compression vs. Distortion nObjective: to significantly reduce the number of bits which must be transmitted, but without creating objectionable levels of distortion nWe are concerned mainly with telephone applications, with voice signal already band-limited to 4 kHz. max. and sampled at 8 kHz. nThe objective is toll-quality voice reproduction nGeneral Categories of Speech Coders Waveform Coders –attempt to re-create the input waveform –good speech quality but at relatively high bit rates Vocoders –attempt to re-create the sound as perceived by humans –quantize and mimic speech-parameter-defined properties –lower bit rates but at some penalty in speech quality Hybrid Coders –mixed approach, using elements of Waveform Coders & Vocoders –use vector quantization against a codebook reference –low bit rates and good quality speech

18. July, 19982 - 18RF100 (c) 1998 Scott Baxter Meet some Families of Speech Coders Waveform Coders PCM (pulse-code modulation), APCM (adaptive PCM) DPCM (differential PCM), ADPCM (adaptive DPCM) DM (delta modulation), ADM (adaptive DM) CVSD (continuously variable-slope DM) APC (adaptive predictive coding) RELP (residual-excited linear prediction) SBC (subband coding) ATC (adaptive transform coding) Hybrid Coders MPLP (multipulse-excited linear prediction) RPE (regular pulse-excited linear prediction) VSELP (vector-sum excited linear prediction) CELP (code-excited linear prediction) Vocoders Channel, Formant, Phase, Cepstral, or Homomorphic LPC (linear predictive coding) STC (sinusoidal transform coding) MBE (multiband excitation), IMBE (improved MBE) nObjective: to significantly reduce the number of bits which must be transmitted, but without creating objectionable levels of distortion nWe are concerned mainly with telephone applications, with voice signal already band-limited to 4 kHz. max. and sampled at 8 kHz. nThe objective is toll-quality voice reproduction nA few different strategies and algorithms used in voice compression:

19. July, 19982 - 19RF100 (c) 1998 Scott Baxter Speech Coders Used Mobile Technologies: nVocoders are usually described by their output rate (8 kilobits/sec, etc.) and the type of algorithm they use. Here’s a list of the vocoders used in currently popular wireless technologies: bits/sec 64k 32k 16k 13/7/4/2 v 13k 9.6k 8k 6.7k 6.4k 8/4/2/1 v 4.8k 2.4k Algorithm log PCM ADPCM LD-CELP APC QCELP RPE-LTP MPLP VSELP IMBE QCELP CELP LPC-10 Standard (Year) CCITT G.711 (1972) CCITT G.721 (1984) CCITT G.728 (1992) Inmarsat-B (1985) CTIA, IS-54/J-Std008 (1995) CDMA Pan-European DMR, GSM (1991) BTI Skyphone (1990) CTIA IS-54 (1993) TDMA Japanese DMR (1993) Inmarsat-M (1993) Enhanced Vocoder, 1997 CDMA CTIA, IS-95 (1993) CDMA US, FS-1016 (1991) US, FS-1015 (1977) MOS 4.3 4.1 4.0 n/avail 3.5 3.4 3.5 3.4 n/avail 3.4 3.2 2.3 8kEFRCIS-136 (1997) TDMA enhancedn/avail

20. July, 19982 - 20RF100 (c) 1998 Scott Baxter Digital Modulation

21. July, 19982 - 21RF100 (c) 1998 Scott Baxter Modulation by Digital Inputs nFor example, modulate a signal with this digital waveform. No more continuous analog variations, now we’re “shifting” between discrete levels. We call this “shift keying”. The user gets to decide what levels mean “0” and “1” -- there are no inherent values nSteady Carrier without modulation nAmplitude Shift Keying ASK applications: digital microwave nFrequency Shift Keying FSK applications: control messages in AMPS cellular; TDMA cellular nPhase Shift Keying PSK applications: TDMA cellular, GSM & PCS-1900 Our previous modulation examples used continuously-variable analog inputs. If we quantize the inputs, restricting them to digital values, we will produce digital modulation. Voltage Time 1 0

22. July, 19982 - 22RF100 (c) 1998 Scott Baxter Digital Modulation Schemes nThere are many different schemes for digital modulation, each a compromise between complexity, immunity to errors in transmission, required channel bandwidth, and possible requirement for linear amplifiers nLinear Modulation Techniques BPSK Binary Phase Shift Keying DPSK Differential Phase Shift Keying QPSK Quadrature Phase Shift Keying IS-95 CDMA forward link –Offset QPSK IS-95 CDMA reverse link –Pi/4 DQPSK IS-54, IS-136 control and traffic channels nConstant Envelope Modulation Schemes BFSK Binary Frequency Shift Keying AMPS control channels MSK Minimum Shift Keying GMSK Gaussian Minimum Shift Keying GSM systems, CDPD nHybrid Combinations of Linear and Constant Envelope Modulation MPSK M-ary Phase Shift Keying QAM M-ary Quadrature Amplitude Modulation MFSK M-ary Frequency Shift Keying FLEX paging protocol nSpread Spectrum Multiple Access Techniques DSSS Direct-Sequence Spread Spectrum IS-95 CDMA FHSS Frequency-Hopping Spread Spectrum

23. July, 19982 - 23RF100 (c) 1998 Scott Baxter Modulation used in CDMA Systems nCDMA mobiles use offset QPSK modulation the Q-sequence is delayed half a chip, so that I and Q never change simultaneously and the mobile TX never passes through (0,0) nCDMA base stations use QPSK modulation every signal (voice, pilot, sync, paging) has its own amplitude, so the transmitter is unavoidably going through (0,0) sometimes; no reason to include 1/2 chip delay Base Stations: QPSK Q Axis I Axis Short PN Q  cos  t sin  t User’s chips Short PN I Mobiles: OQPSK Q Axis I Axis Short PN Q  cos  t sin  t User’s chips 1/2 chip Short PN I

24. July, 19982 - 24RF100 (c) 1998 Scott Baxter CDMA Base Station Modulation Views nThe view at top right shows the actual measured QPSK phase constellation of a CDMA base station in normal service nThe view at bottom right shows the measured power in the code domain for each walsh code on a CDMA BTS in actual service Notice that not all walsh codes are active Pilot, Sync, Paging, and certain traffic channels are in use

25. July, 19982 - 25RF100 (c) 1998 Scott Baxter End of Section