FEE Conference – Brussels – October 2013 An Introduction to Digital Modulation Presenter: Barry Hack, Aeroflex UK

What do you know about Digital Comms ? 2 AM, Pulse FM PM  V= A(t) sin[ (t)] (Sine wave modulation signal)  V= A(t) sin[2 f(t) + (t)]  Is it really different from analog modulation? Name some early systems using digital modulation?

What do you know about Digital Comms ? Early systems employed: Sonar Morse code Semaphore Smoke signals 3

What do you know about Digital Comms ? ▼ Acronyms –GSM –TETRA –UMTS –BPSK –  /4 DQPSK –CODEC –LTE –RBER –BCCH –MNC 4  Other terms –Interleaving –Chipping Rate –Rake Receiver –Multi-path –Spreading Factor

Why do People want Digital Modulation ? ▼ Security –Princess Diana and the Sun newspaper –Prevent eaves dropping and ‘spoof’ or ‘rogue’ users…. ▼ Capacity –More users per piece of spectrum than analog –Less congestion –More revenue for operators ▼ Cost -> Seller makes more money –Digital radios have less analog bits ▼ Cheaper to produce, more reliable, easier to align 5

Why do People want Digital Modulation ? ▼ Voice Quality –Works better where signal is weak ▼ Roaming –Can speak over larger geographies –Emergency services can all communicate directly –GSM roaming across most countries (130+) ▼ Immunity to interference ▼ Capability to send voice and/or data 6

One ‘technical’ reason to move to ‘Digital’ 7 Comparison with analog FM FM Low background noise High background noise Quality Range Digital

Real radio systems ▼ They do not use one access method –They combine techniques and attempt to get the best of each (divide the users by space/location, time, frequency and/or code) ▼ There is no “best” solution –It depends on what you are trying to achieve ▼ voice, data –Geography –Regulatory constraints –Spectrum availability –Cost objectives –Services needed –User density –Politics 8

The analog implementation ▼ Information source modulates carrier directly 9 ModulatorTx Filter Demod- ulator Rx Filter Baseband filter Duplexer V.C.O. Baseband filter Low Noise Amp Power Amplifier Analog

Simplified digital transceiver 10 A to DSpeech Coder Channel Coder ModulatorTx Filter D to ASpeech Decoder Channel Decoder Demod- ulator Rx FilterFilters & Equalizer Baseband filters Duplexer V.C.O. Note: TX output PA and RX LNA removed for clarity AnalogDigital

Modulation: where is the information 11 AM, Pulse FM PM  V= A(t) sin[ (t)] (Sine wave modulation signal)  V= A(t) sin[2 f(t) + (t)] 

Basic Digital Modulation 12 Amplitude Frequency (FSK) Phase Both Amplitude and Phase

IQ Diagram: Phase and Amplitude 13 Phase Mag 0° I+ Q+ 90° Reference Phase Origin –Magnitude is an absolute value from the origin –Phase is relative to a reference signal (from I + axis)

Digital Modulation: Signal vector 14 Phase Mag 0 deg Amplitude Modulation Phase 0 deg Phase Modulation Frequency Modulation Amplitude and Phase 0 deg

Modulation Measurements ▼ Analog Systems –Power –Bandwidth –Frequency error –Modulation Accuracy (FM deviation / AM depth) ▼ Digital Systems –Power –Bandwidth –Frequency error –Modulation accuracy (Error Vector Magnitude) –Burst Timing (Power) –Symbol Timing (Data) 15

Modulation Accuracy ▼ EVM is a good measure –Some systems (i.e. FSK) would use phase error only ▼ Definition –EVM is the difference between the actual signal vector and an ideal signal vector. ▼ Some causes of EVM –Component variations –PCB track layout –Phase Noise –Spurious signals –Modulator errors 16 Q I  Magnitude error (IQ error magnitude) Measured signal Ideal (reference signal) Phase Error (IQ Phase error) Error vector

Causes of EVM - Example 1 ▼ Carrier Leakage –Some of the un-modulated local oscillator bleeds across to the output ▼ Poor screening ▼ Poor PCB layout 17 I Q IQ Modulator  /2 Carrier I: Q: Carrier Leakage

Causes of EVM – Example 2 ▼ IQ Skew –The I and Q modulation paths are not exactly 90 degrees ▼ Component tolerances ▼ Different lengths for I and Q signal paths ▼ Poor PCB layout 18 IQ Modulator  /2 Carrier I: Q: Cos (  skew ) I Q skew

Causes of EVM – Example 3 ▼ IQ Gain Imbalance –The I and Q modulation paths do not have the same gain ▼ Component tolerances –Note: Gain and skew can be seen as AM modulation in the analog domain ! 19 IQ Modulator  /2 Carrier I: Q: I Q

EVM ▼ Example 1 20 Noise can be seen because of the spread of constellation points Gain imbalance present because I and Q values are not symmetrical This angle is not 90 degrees showing skew

EVM ▼ Example 2 –Minor issues with carrier leak and phase noise –BUT the quality is well inside the measurement limits 21 Vector DiagramConstellation Diagram Rotated Vector

Receiver Tests ▼ Analog Systems –RSSI (Received Signal Strength Indicator) –Rx sensitivity (using SINAD measurement) ▼ Digital Systems –RSSI (Received Signal Strength Indicator) –Rx sensitivity (using BER measurement) 22

Receiver Tests ▼ Digital Systems –Bit Error Rate is a measure of the received bits in error as a ratio to the total received bits ▼ Other measurements include… –RBER – Residual Bit Error Rate –FER – Frame Error Rate –MER – Message Error Rate 23 If you use a 1kHz test tone, SINAD and BER can give very similar answers for receiver sensitivity ….. Rx Sens = -119dBm

Vector representation of AM and FM ▼ Remove carrier phase changes ▼ Indicate relative phase changes only 24 I QQQ I I unmodulated carrier, f c, arbitrary phase carrier, f c, with FMcarrier, f c, with AM

Vector representation of AM - Slow 25

Vector representation of AM – Faster ! 26

IQ Modulation Explained 27 serial/parallel conversion Q I A B C D I(t) Q(t) t Dibits vector phase states serial data stream t Bit period A B C D Symbol period

IQ Modulation Explained 28 A I(t) Q(t) t Dibits Q I -sin(I), +cos(Q) 0 1 t A 1 0 Bit period Symbol period L.O. 90 º ‘I’ signal ‘Q’ signal fcfc I.sin(f c ) Q.cos(f c )

IQ Modulation Explained Q I B I(t) Q(t) t Dibits 0 1 sin(I), -cos(Q) t B 0 1 Bit period Symbol period ‘Q’ signal L.O. 90 º ‘I’ signal fcfc I.sin(f c ) Q.cos(f c )

IQ Modulation Explained Q I 00 t C 1 Bit period C I(t) Q(t) t Dibits Symbol period 1 1 -sin(I), -cos(Q) L.O. 90 º ‘I’ signal ‘Q’ signal fcfc I.sin(f c ) Q.cos(f c )

IQ Modulation Explained Q I sin(I), cos(Q) t D 0 Bit period D I(t) Q(t) t Dibits 0 0 Symbol period L.O. 90 º ‘I’ signal ‘Q’ signal fcfc I.sin(f c ) Q.cos(f c )

IQ Modulation Explained Q I

Vector Timing and Synchronisation 33 ▼ To decide when the vector is at a symbol point you need to apply timing and synchronisation

QPSK Modulation – Noisy but perfect ! 34

Alternative Approaches to QPSK 35 Q I Q I Time offset QPSK Phase offset QPSK Avoid zero crossings and so minimise AM Real signals are filtered

Sampling and Speech Coding ▼ Standard telecom data rates –300 to 3.4kHz audio band –8k samples / sec, 13 bits / sample (103kbit / sec) –Compressed to 8 bits / sample (64kbit / sec) with A-law or u-law compander ▼ Even 64kbit / sec data is too high for wireless systems ▼ CODEC – (COder-DECoder) reduces data rate by up to 80% –Several approaches: model vocal tract; code book & lookup table 36 Audio kHz8, bit samples / sec (103kbit / sec) 13kbit / sec A to DCODEC

The Channel Coder 37 Speech Coder Channel Coder D to ASpeech Decoder Channel Decoder Filters & Equalizer A to D

Interleaving ▼ Loss of a single data frame makes the entire message meaningless. How do we fix this? 38 COLD FEET NEED HEAT

Interleaving – Demonstration 39

Equalisation 40 Speech Coder Channel Coder D to ASpeech Decoder Channel Decoder Filters & Equalizer A to D

Equalisation ▼ Mobile communications often rely on multi- path signals –How often do you actually have sight of the base- station when using a mobile phone? ▼ The EQUALISER in the receiver –Overcomes the effects of delay spreading –Using real-time adaptive filtering implemented in DSP –Requires some prior knowledge of the received signal ▼ i.e a training sequence 41

The End 42 Any questions ?