Lab 6-8 Digital modulation and channel S6 ECE, Open lab Dr. Viswas S. Nair More details can be found in Modern Wireless Communication, S. Haykin, M. Moher, Pearson Digital Communications, John 6. Proakis, MsGraw-Hill International EditionDigital modulation * Modulator maps digital information into analog waveforms that match the channel * An M-ary modulation scheme can transmit M level-information symbol in amplitude/phase/frequency (or a combination of these) * The mapping from digital information to analog is from a block of k = log, M bits into one symbol * Symbol rate is the rate at which signal changes representing new information * Hence if the bit rate is R (number of bits to be sent per second) then the symbol rate is R/k * Common schemes + Amplitude: pulse amplitude modulation (PAM) or Amplitude shift keying (ASK) + Phase/pulse positon : Phase shift keying (PSK), Quadrature phase shift keying (QPSK), Pulse position modulation (PPM) + Frequency/ pulse period: Frequency shift keying (PSK), Pulse width modulation (PWM) * Amplitude+ phase : Quadrature amplitude modulation (QAM)Pulse amplitude modulation (PAM) * PAM (or amplitude shift keying (ASK)) is a linear ( superposition principle is followed) memoryless (output does not depend on previous waveform) modulation * The PAM waveform is of the form Sm(t) = Amg(t) cos(wt) , m=1,2,3 ... M *A,, represents m possible amplitudes in the M-ary modulation * g(t) is the pulse shape carrying pulse energy E, * For binary modulation M=2 and A,, can be A, or A, representing the amplitudes for digital 0 and 1, respectively. * The bit rate and bit period are related as T, = x, where R is the bit rate * The symbol rate (R/k) and symbol interval (T) are related as T = : = kT, * Frequency domain properties of PAM is similar to analog AM signalBinary (M=2) PAM/ASK waveform generation * The binary data is to be multiplied bya, carrier. ov * Here the symbol is consisting of one bit Input binary sequence alone since M=2 * This gives truncated sine waves where \ there are 1’s and no amplitude at 0’s ov av IM | | {Hh * Binary wave is implemented in this LabVIEW experiment using a random bit generator & digital to analog SK Modulated output wave Sine “ait converter — + AFOR loop generates sine wave that is one bit long (many sinusoidal periods - Sine x symbol ie jymbols, | |) in one bit) and the amplitude is E multiplied by each bit’s analog value st waveform mponents Index arrayBinary (M=2) PAM/ASK waveform generation * The resulting sequence of sine waves is w concatenated using a feedback loop. Input binary sequence Waveform constant a size -—Hil WHI Wh HN | ASK Modulated output wave Insert intg array * The orange loop is for attaching each truncated sine wave multiplied by the bit’s amplitude to the end of current loop * The blue loop provides index at which the new sine wave should attachBinary (M=2) PAM/ASK waveform generation * Finally, a switch case statement makes Ww sure that the output appears only once, | all bits are processed (i.e. the for loop index i reaches the number of iterations N=number of bits) Input binary sequence | ASK Modulated output waveBinary (M=2) PAM/ASK waveform generation The symbol rate, the number of samples per cycle of the sine wave carrier, the number of cycles in one symbol interval and the number of symbols are the inputs used in this labsheet. w ll \| Input binary sequence Thus, the carrier sampling rate =(no. ow of samples per cycle x no. of cycles per av symbol) x symbol rate |ASK Modulated output wave Also, No. of samples per symbol = no. of samples per cycle x no. of cycles per symbol Alternatively, the input could have been the symbol rate, number of symbols and carrier frequency and the carrier sampling rate. | I | M+Channel model * The modulated signal travels through a channel suffering * amplitude and phase distortion in a band limited channel, + attenuation and * noise addition * The frequency dependent amplitude and phase distortion can be modelled using an appropriate filter response * The frequency independent attenuation can be included as a constant loss term * The noise if additive can be modeled by adding the noise signalChannel model * Asimple model considered in this lab uses * a Butterworth bandpass response for the channel * A frequency independent attenuation term given in dB and + An additive white gaussian noise term obtained from a normal distribution specified by standard deviation. Analog modulated digital waveform Receiver Digital ip cata Raciucaesn Pee ee Peers Te : ‘ i Cl r Deesctead Se* Input binary sequence (baseband wave) w 000{11]o001]00/1111]o[1]o w * PAM modulated waveform (passband \_ wave i.e. at carrier frequency) - | * a Butterworth bandpass response for the * | | channel ov _ * A frequency independent attenuation = ff ht aH term given in dB * An additive white gaussian noise term onv ' obtained from a normal distribution specified by standard deviation. “ F M env u Input binary sequence* a Butterworth bandpass response for the channel eee eed tonto sth Waveform * A frequency independent attenuation term given in dB Lower cutoff Filtered signal wt p> b> ani eng te Filtered + attenuated signal noise SD * An additive white gaussian noise term obtained —_Carersempling ate from a normal distribution specified by standard — es deviation. | Number of came samoles per symbolSub-V| * Sub-VI (custom blocks that contain independent block diagrams working like functions in C or MATLAB) makes block diagram simple * To make a Sub-VI: select the necessary blocks, then go to Edit> Create Sub-VI to make them into Sub-VI. * Go to the front panel of the Sub-VI and locate the small boxes on the right side» top icon. The white boxes are free terminals where new variables can be a mapped. Click on the white box and then click on the indicator or control that needs to be made a terminal. * Pass the global variables from main program to Sub-VI to avoid duplication and mistakes in re-entering the same value in many Sub-VIs Please remember * The implementations shown here are just examples for learning the features of LabVIEW. There may be more/better ways. Try to explore. * There may be more diverse applications of the same ideas. Explore that tooDigital receiver * Consists of two parts * a demodulator that uses either a matched filter or correlation to optimally demodulate the signal and provide an N-dimensional vector matching the transmitted data dimension * Adetector that decides the transmitted signal (or level) in an M-ary scheme that matches closely with the received vector Analog modulated digital waveform Digital ip Attenuation ’ ’ =>) Modulator Se Output eee oreo] filter pena Coc Channel modelDigital Demodulation * Demodulator decomposes the signal+noise into N-dimensional basis vectors + N(S M) basis vectors span the signal space * Here, for PAM N=1 since 0 and 1 both can be represented as the same wave with amplitude 0 and 1, respectively (the two states are linearly dependent). * Cross correlation function finds the projection of one function on the other when the delay t=0 ‘ Ry(7) = x(t)y*(t — 7) dt * Convolution performs filtering by multiplying and adding the product of a signal function with an impulse response function y(t) = [xome —1)dr * Both operations can be used to find the projections of the received signal on the basis functionsDigital Demodulation — bringing to baseband + In our case there is just one basis function and we know it as a pulse in baseband * The passband truncated sine can be converted to baseband pulse by rectification + and low pass filtering =| (basic amplitude demodulation) = fe Demod fier upper cutoff bm. i i] Exonpas=P 1. " os * The output will be a series of pulses with noise in it.Digital Demodulation — * Acrude implementation for PAM reception can skip the demodulation and perform detection by just comparing with the mean of the wave to see if the amplitude is above or below at a particular time (prone to error and not advised) to decide if the received bit is a 0 or 1. * A better implementation will use correlation or matched filter receivers. f(t) Correlation receiver r1) = tp) + n() n(T ) Decision Threshold deviceDigital Demodulation-correlation receiver implementation * Here, correlation (with zero time delay t=0) demodulator is used. 1%} (T) = I r(Of;e (dt 0 * It involves multiplying received signal + noise (r(t)) with the basis functions (f(t) and integrating over one symbol period (0 to T), taking the sample at T. * There is only one basis function for a binary PAM and is a pulse. * Since the amplitude is constant the implementation skips the step and an appropriate level as detection threshold us used. * The one symbol time long subarray of the PAM wave is extracted and integrated in a FOR loop that runs for all symbols pase ieetien ‘array subset =a i/p PAM wave samples per symbolDetection * The detector is implemented by comparing the sample from the demodulator at the end of each symbol period with a detection forename eae threshold in the same loop anteiaiid ; * For PAM there are only two levels and a comparator checks if the level is above or below the mean (assuming binary symmetric channel) * Then a switch case assigns the 0 or 1 level based on the comparator m— output + The generated bits are concatenated to form an array and is the output coc i —fen f ae ‘etection thesold past sia and emodsator ofpWhy channel model? Channel - capacity * Channel capacity theorem by Shannon (1948) * Channel capacity of bandlimited AWGN waveform channel with a bandlimited and average power-limited input bandwidth of the channel / C=B log, (1+S/N) Channel capacity signal-to-noise sn bits /s rate * Channel coding can be used to add redundancy in data to enable error correction at receiver side. * Noisy channel coding theorem (Shannon (1948)) says that Arbitrarily low error rates can ig obtained with the help of channel codes if transmission rate (R) < channel capacity C)Channel- Models * Simplifications to suit different scenarios * Binary symmetric channels (BSC)- two levels, both have same probability of error. — i.e. chances of a 0 becoming 1 is same as a 1 becoming 0 * Discrete memoryless channel (DMC)- Discrete input Discrete output, output depends on the present state of input —-no dependency on past input * Discrete-input, continuous-output (ex. AWGN) + Waveform channel-separate modulator and demodulator from channel and visualize channel with input and output as waveformsEffect not considered here: Channel-Fading + Scattering and diffraction due to obstacles can lead to multipath for rays. * The diffraction or interference between waves coming from multiple paths lead to fading when there is a relative motion between obstacles and source/receiver. * Slow(shadowing or lognormal fading)/fast fading based on the relative speed and distance. Small distance and fast movement for fast fading. * Fast fading with no dominant path, (ex. mobile devices), is often modeled as Rayleigh fading distribution. * Rician fading is used when there is one dominant path. * Doppler effect also leads to a change in frequency * Time selective channels have attenuation or phase rotation as time varying quantities * Frequency selective channels has frequency dependenceMore on the attenuation/Loss term * Attenuation due to atmosphere Tw * Free-space propagation (Friis transmission equation) = Power atthe receiving antenna P.G,G,2? Ae dani niaeantgimexe 6,= Gain ofthe transmitting antenna z (4nky 6,= Gain of te receiving antenna A= Wavelength [R= Distance between the antennas R= hepecokomoneacatech team cso