EE5530: Wireless Communications
3. Digital Modulation
Bandpass digitally modulated signals Signal space, vector representations Spectral characteristics Optimal detection in AWGN Error probability in flat fading channels Linear, memoryless modulation (QAM, PSK) Nonlinear modulation with memory (CPM, MSK, OFDM)
Modulation Considerations
Want high rates, high spectral efficiency, high power efficiency, robust to channel, low complexity (cost) Linear Modulation (MPSK, MQAM, MPAM)
Information encoded in amplitude/phase More spectrally efficient than nonlinear Easier to adapt Issues: differential encoding, pulse shaping, bit mapping.
Nonlinear modulation (FSK)
Information encoded in frequency More robust to channel and amplifier nonlinearities
Linear Modulation
Info. encoded in amp./phase (PSK,QAM) Offset modulation avoids zero crossing of phase trajectories Optimal bit mapping reduces BER Pulse shaping improves spectral properties Differential modulation mitigates phase drift
Linear Modulation in AWGN
ML detection induces decision regions
Example: 8PSK
dmin
Probability of symbol error Ps depends on
# of nearest neighbors Minimum distance dmin (depends on s) Approximate expression
Ps ( s ) M Q M s
Linear Modulation in Fading
In fading, s and therefore Ps are random Performance metrics:
Outage: probability that Ps is above target (used when Tc>>Ts ) Average Ps: expected value of r.v. Ps (used when Tc~Ts ) Error probability much higher than in AWGN alone Combined outage and average Ps
Ts Ps(target)
Ps
Ps =
Ps
Ts
( s ) p ( s ) d
Outage
Combined Outage and Average Ps
Ps Ps Pstarget
Used in combined shadowing and flat-fading Outage determined by shadowing (above target) Local Ps determined by flat fading
Nonlinear Modulation (FSK)
Information encoded in signal frequency
Constant envelope signal Robust to amplitude distortion and amplifier nonliearity
Uses FM modulation (continuous phase) Bandwidth depends on both fc and Bu Pulse shaping improves spectral properties
Adaptive Modulation
Adaptively change modulation relative to fading
Adapt parameters of modulation to fade values
Parameters to adapt
Constellation size Transmit power Instantaneous BER Symbol period Coding rate/scheme
Optimization criteria
Maximize throughput, Minimize average power; Minimize average BER
Only 1-2 degrees of freedom need be exploited for good performance
Modulation in Wireless Applications
AMPS: analog FM GSM/DCS1800/PCS1900/DECT (Europe): GMSK IS-54/IS-136 (North America): /4-DQPSK IS-95 (North America): QPSK(FL)/DQPSK(RL) PDC/PHS (Japan): /4-DQPSK CT2 (Canada): GMSK IEEE802.11: BPSK, QPSK, OFDM Bluetooth: GFSK
Summary
Modulation
Digital modulation encodes bit information in amplitude, phase, or frequency of transmitted signal Modulation choices involve tradeoffs in cost, performance, BW efficiency, and power efficiency. Linear modulation more spectrally efficient but less robust. Design choices for linear modulation include differential modulation and pulse shaping
Fading
Average fade duration dictates performance (BER or outage). Fading greatly increases average Ps , thus BER In fading Ps random: characterized by outage or average value.
Bandpass Signals and Systems
Baseband representations of bandpass systems
Baseband representations by Hilbert Transform s(t) = Re[u(t) exp(j2fct)] = uI(t)cos2fct - uQ(t)sin2fct We can ignore frequency translation encountered in the modulation process For convenience, we shall deal with the transmission of baseband equivalent signals and the baseband equivalent channels u(t) = A n b(t-nT, xn), xn = (xn, xn-1, , xn-K)
Digital communication systems with Quadrature demodulators
Signal Space
Digital modulation
The process of mapping digital information into analog waveforms The mapping is performed by taking block of K =log2M bits and selecting one of the 2K waveforms
Vector representation of waveforms
Signal space: any M waveforms can be represented in a N-dim signal space spanned by a complex basic function set {1(t),, (t)}, N <= M The orthonormal basic function set is not unique, can be constructed from the Gram-Schmidt procedure Signal operations: energy, correlation, mutually orthogonal
Optimal detection in AWGN using the signal space diagram
Signal constellation: vector representation/polar coordinates Density packed constellation implies bandwidth efficiency Bit error probability is related to distances between closest points
PSD of Digitally Modulated Signals
PSD of digitally modulated signals
Sss(f)=1/2[Suu(f-fc) + Suu(-f-fc)]
PSD of a complex envelope Suu(f) Pulse shaping
Suu(f) = 1/T . Sa(f) . |P(f)|2, where u(t) = n an p(t-nT)
Rectangular pulses Raised cosine and root raised pulses Ideal Nyquist pulse (sinc)
Spectral characteristics
Bandwidth efficiency = Rb/BWs = (log2M/T)/BWs Time and frequency envelope decay rates
Estimating PSD from sampled data
Phase Shift Keying (PSK)
M-ary PSK signals
Signal constellation Probability of error Bandwidth utilization
BPSK
uses alternative sine wave phase to encode bits Simple to implement, inefficient use of bandwidth very robust, extensively used in satellite communications
QPSK
Multilevel modulation: 2 bits per symbol More spectral efficient (2x BPSK), more complex receiver
Variations on QPSK
Offset QPSK (OQPSK)
The phase difference between baud intervals can only be 90o The same PSD and BER as QPSK Phase transitions avoids center of diagram, remove some design constraints on amplifier
/4-shifted Differential QPSK (/4-DQPSK)
Restricted carrier phase transition to +/- /4 and +/- 3/4 Signaling elements selected in turn from two QPSK constellations, each shifted by /4 Maximum phase change is 135o vs. 180o for QPSK, thus maintaining constant envelope Phase transitions avoids center of diagram Nice phase transition makes symbol synchronization easier Robust to phase drift, better BER; Same PSD as QPSK, twice as efficient as GMSK
Quadrature Amplitude Modulation (QAM)
Suitable for channels with excessive power and good linearity, i.e., telephone channels, P2P microwave QAM signal constellations
Amplitude modulation on both quadrature carriers 2K discrete levels, K=2 same as QPSK
PSD and bandwidth efficiency
Amplitude pulse shaping
Error probability
For QAM, gray coding is possible so that nearest neighbor errors result in single bit error, therefore, Pb = P() For large M, QAM is generally more efficient than M-PSK
Continuous Phase Modulation (CPM)
CPM
Carrier phase varies in a continuous manner Constant envelope
An infinite variety of CPM signals can be generated by adjusting
Frequency pulse shaping functions Modulation index h Modulation alphabet size M Pulse duration L
Phase transition diagram Spectra of CPM signals
Compact PSD: narrow mainlobe and fast sidelobe roll-off
Variations on CPM
Minimum Shift Keying (MSK)
Special form of CPFSK Minimum spacing that allows two frequencies states to be orthogonal spectral efficient, easily generated
Gaussian Minimum Shift Keying (GMSK)
MSK + premodulation Gaussian lowpass filter Increase spectral efficiency with sharper cutoff, excellent power efficiency due to constant envelope Used extensively in 2nd-generation digital and cordless telephone applications
Orthogonal Frequency Division Modulation (OFDM)
OFDM signals
Block modulation with parallel symbol transmission over orthogonal sub-carriers Mitigating channel impairment (ISI) due to small subband bandwidth
FFT-based OFDM system
Digital Demodulation and Detection
Demodulation Detection
Process of removing the carrier signals Process of symbol detection Coherent (synchronous) detection
Receiver uses a local carrier of same frequency and phase for detection Cross correlate with replica signals at receiver Match within threshold to make decision Examples: PSK, FSK, ASK, CPM
Non-coherent (envelope) detection Does not exploit phase reference information Less complex receiver, but worse performance
Examples: FSK, ASK, DPSK, CPM
Differential detection
