You are on page 1of 21
DAT OOM, 13a" ece 2 211 DIGITAL COMMUNICATION TEST-1 Aug-Dec. 2016 Max. Marks=25 fe the axioms of iincertainty and show that there is only one function that satisfies ~~ fall the axioms. Is “Information” a measurable quantity? Justify your answer. (An unbiased die with six faces is thrown twice and the following is disclosed: "The a sum of the faces appeared is even.”. How much information is conveyed by this disclosure? _ _ Prove the inequality: Q(a) < $e F What is the significance (importance) of this inequality? Z a = (DV Show that given M real-valued, Gnite-cnergy functions (zm(l)}, taking non-zero values only in the interval [0,], there exist N g Namaidvia Unit n'\avan , \251)2,9, dy eck Com Men) ©2211 DIGITAL COMMUNICATION Major (Final) Exam, Aug-Dee, 2016 Max. Marks=50. y (9) Conse the thre signa fork = 0,152 ru(t) = yieon (anf fk), OST aund 0 elsewhere, ‘These ate to be used to send one of Uh channel of PSD 48, When the messages are equally probability of error is given by P= Jou ee vrti [l-@(u+ Vel} ly fA] Four frequeney-orthogonal signals (0) = V2EF()sin(umt), m= 1,2,3,4, each [4 of duration 7 seconds is sed to communicate in a non-coherent AWGN channel Draw the encoder, modulator and optimum structure of the ML demodulator for / non-coh ee ‘that will not have, Give an example ofa practical system for each. (4) Define the terms “diversity gain” and “coding gj a fading channel 7 “A good communication scheme exploits all the available degrees of freedom in the (2) channel” - Justify this statement with a couple of examples of SISO communication cystems. messages over an AWGN , show that the average rence and those [2] n” of a communication scheme for (2) 1 (cY Define the “Deep fade event” and justify the definition. For a system operating ina [3] fading channel if the PSD of the additive noise alone is doubled (all other quantities remaining same), does the deep fade event change? Jpistify your answer. (af Define the following and give an example for each:~() time-selective and frequency- (3) selective channel, (jf non-time-selec prance sh channel, Af time- selective, non-frequency selective channel, (jy non-time-seletive, non-frequency se- lective channel. J.) Siate and prove the Chemoff bound i = gige™. Show that [2 A Vir {+ wis exponentially distributed, What is the mean of X? When his complex Gaussian with zero mean and unit variance show that Py E [a /2RPSNR)| = § (1 ~ sSfa) - ) > > . 2 o 3 a 3 st Show that in a Rayleigh fading channel with channel gains known parfectly at the BL. _ receiver, the average error probability 9. of coherent detection of BPSK on a symbol by symbol basis is given by (1 — Vain) whose high-SNR. approximation is 3 1 2 2 4 2 2 2 For xX 1,2, let rw; be real Gaussian RVs with pdf f (wy ahr: 4/ (Explain the “Successive nulling and cancelling” (, ‘multiplexing (VBLAST) system. Is this detection ML? Justify your answer. (YF Show that the “best first cancelling approach at each stage in the detection pro wv Jeads to the globally optimum ordering in the maxmin cae Show that in a time-selective SISO fading channel repetition code of length LA Pp emouaes le of length L gives [4] (with ordering) detection for spatial [2] _ sense, State and prove the rank and the determinant criterion to design @ spacetime block code for a MIMO fading channel (arbitrary number of transmit and receive ntennas) . : (Mf Show that the Alamouti code | 7! —73 . r mo mit diversity gain with the ML decoding complexity same as that of the uncoded SISO system. When the variables x1, x2 take values from the finite signal set S, ___._ calculate its coding gain in terms of the parameters of S. So What is the rate of the code [ pee a Zita, appa}. | complex symbols per channe! ; Provides the maximum possible trans- use? Justify your answer. tu + jt2q | yf Shoy“that the code . 0 ta +jxrig | 12 = 12 + j72q is single complex symbol decodable (like the Alamouti code) when used with one receive antenna. Also show that the transmit diversity gain depends oy the signal set from which the variables x; and r2 take values. Ident . where 2 =21;+ jr and one signal for which the transmit diversity gain is 2 and one for which it is 1 oh Assuming, that the channel response of a frequency selective fading channel has a finite number of taps L, eypsfain how viewing this cha _combifved with Maximum Likelih (pf Show that in a frequency-selecti . SSS (Direct Sequence - Spr n of L. "Explain how freq channel using OFDM. annel as a MISO fading channel ain of L can be achieved with uncoded transmission 1 Sequence Detection (MLSD). ead Spectrum System) with RAKE receiver can extract a diversity cy diversity can be achieved in a frequency selective fading 2] (3) facing channel with L number of taps the DS- GOOD LUCK !!' E22 SITAL Ci on 11 DIGITAL COMMUNICATION Aug-Dec. 2015 Max, Marks=25 State the axioms of wncertaint jon that satisfi ncerLainty and show that there is only one fynction that satisfies (2) Sima y that there is only one fynction that sati i An unbiased coin is thrown four times and the following odd number of heads in the first three throws". How much information is conveyed Define "Vector Space”. Show that given i real-valued, fnite-energy functions {zm(¢)}, taking non-zero values only in the interval (0,7), there exist N < M unit- energy (normalized) orthogonal functions da(t) such that zm(t) = Dnw (2), M1 Emad mn = fp tm(l)ba(t)dt. Furthermore, )} is linearly independent over the set of real for m = 1,2,-++M, where for each m end N = M Mand only if the set (2m(t numbers. ‘Define the following and give an example for each: (a) Minimum probability of (2). error decoder (b) MAP decoder (c) ML Decoder and (d) Minimum-Distance decoder and (¢) Bounded-Distance Decoder. Under what conditions/situations a bounded: - distance decoder will be an ML decoder? “{by Show that when a binary code is used on a BSC (Binary Symmetric Channel) the [2] receiver which decides in favor of the codeword which is closest to the received vector in Hamming distance.is an ML receiver. ‘Draw a block diagram for an optimum ML receiver for the 6-ary communication [2] ~ "using the signals s, i= 1,2,--- ,6, given below (for the ranges of ¢ with no value specified, the signal takes the value zero) in a AWGN channel of PSD 48 : aA@=LOStsh sal rstsz sal 0sts% salt) = 1, 2ECE3 s(t) = 1, OS CSI KICESH sean cts3. Tb). Obtain the union bound for a binary coded signal set (with M codewords) with [3] biphase modulation used in AWGN channel with power spectral density (PSD) with soft-decision decoding. What can you say about the probability of error as M tends to infinity? A” GY Derive the Unioi-Bhattacharya bound. Calculate this bound for a pair of codewords [3] of a binary linear code employed in output symmetric channel. JS For a sufficiently large value of the carrier frequency f., find the minimum frequency [3] separation Ay required for the following two pairs of signals to be orthogonal, where the signals are shown for the interval 0 < t < T, outside of which they take the value zero: (i) si(0) = VFeos(2fet) and an(t) = VFeostan(fe + O,}t) and (i) (2) = [eos fet) and salt) = fFsinQan(fe + O)0- _ the help of f vs fa diagram for AWGN channel explain (compare) the fo} [2] (i) Bandwidth efficient modulatiox schemes; (i?) Power efficient modulation schemes; (ii) Conventional binary coded biphase-morlulated schemesi(iv) Trellis coded mod- ulation (TCM); (v) Turbo Coding Scheme A Show that for a linear lock code the Hamming distance distribution seen from any [I] codeword is the same and it is the Hamming weight distribution of the code. J) Show that the computational requirement for bounded distance decoding of a linear [3] code can be much smaller than the ML decoding of the code using appropriate Hamming code or any other code: 6 (a) Show that the ML soft-decision decoding of convolutional codes when employed in a AWGN channel witha constant energy signal set is the minimum accumulated ae squared Euclidean distance decoding. ) For the encoder of Fig.l, find the transfer function T(D,1,L) of the code. How ‘many binary errors this code code correct ? WA Find an upper bound to the bit error probability of the code with the encoder of when employed in BSC. Make any reasonable approximation. Define a catastrophic convolutional ex-oder and give an example. From the state transition diagram of a convolutional code how can one find whether the encoder is catastrophic or not? Explain the following statesient: The catastrophe property is that of the encoder of a:convolutional code and not that of the code. 7 (2) For an uncoded 4-PSK system operating in AWGN channel show that power gein can be achieved without bandwidth expansion by designing’a 2-state TCM scheme. Calculate the asymptotic coding gain of your TCM scheme. 7 Why one generally doubles the size of a signal set In a TOM scheme? Why not quadruple or more and get more benefit? (©) Show that fora TOM scheme designed for a AWGN channel with Ungerboeck rules, when the complex symbol duration is the same as that of the uncoded system, it is Aot possible to obtain coding gain without signal set expansion. (2} (3. 13] 8] cn i aI GOOD LUCK #1! i ee ee E2211 DIGITAL COMMUNICATION TEsT-2 Aug-Dee. 2018 Max. Marks=25 ie Date the terns "Intrinsic Information” and "Extrinsic Information” (2) xplain turbo coding and decoding with suitable block diagrams and mathematical [4] _ express SM Why the interleaver used in the turbo encoder needs to be a large one? fa (a) Bplin why recursive systematic convolutional encoders are used in turbo coding [2] scheme? 2 (Q Explain the BCIR algorithm to estimate thea posteriori probabilities (APP) of the (4) ~ Wane and transitions of a Markov source observed through a noisy DMC. Ean how the algorithm of 2(a) can be used for APP decoding of convolutional [I] , codes. er For the block code with the parity check matrix H consisting of three row vectors [5] {110020}, (011001) and [101100], draw a trellis. Explain how MAP decoding of linear block codes can be done using this trellis and the algorithm of 2(a). Obtain an ML receiver structure for a binary non-coherent communication system (4) operating in an AWGN channel, For this receiver calculate the average probability of error. How does this differ from the average probability of error for the corresponding Coherent communication system? Write down the probability density functions for Rayleigh, Rician and lognormal (2] distribution. Explain how a wireless diffused multipath channel gets modeled as a Rayleigh or Rician fading channel. Derive an ML receiver structure for a binary communication system operating in (4) a Rayleigh fading channel and calculate the average probability of error for this receiver. How does this differ from the probability of error of a non-coherent ML receiver of 3(a)? Is fading a ‘blessing’ or ‘curse’ or ‘ blessing in disguise’? Justify your answer. _ Jé) Draw a block diagram of a OFDM system implemented using DFT dearly explaining [2] (briefly) the function of each block. State and prove the Singular Value Decomposition (SVD) Theorem for a rectaiigular [4] complex matrix. Explain what is meant by “channel. partitioning”. for a frequency selective channel. [1] (of Explain why OFDM system is bandwidth efficient compared to the traditional fre-_(1] quency division multiplexing. _ Explain the advantages and the disadvantages of OFDM modulation compared to [2] other equalization techniques. 9 GF For a square cyclic matrix show that the eigen vectors are the columns of the DPT [ atrix of the same size, What are the eigen values? (>)Show that “cyclic prefixing” converts an ISI channel into a set of parallel AWGN [1] channels. Is “cyclic prefixing” the only way of converting an ISI channel into a set of parallel |] AWGN channel? Justify your answer, —> 2envo~ pede Show that the "water-filling (or water-pouring)" is the optimum power allocation [2] trategy for the OFDM system. OV ein how capacity of a frequency selective channel can be approached that em- [I] ploys OFDM. . yo o ae os Pe e GOOD LUCK "! iii teen eee eee eee cee Be 211 DIGITAL COMMUNICATION t am Auug-Dec. 2015, Max. Marks=50 YC Sie a pe Cea ad ——y) IH Por i= 1,2, ian RVs with pdf fw.) = ese” m=". Show that (2) X = wi + w} is exponentially distributed. What is the mean of X? When his complex Gaussian with zero mean and unit variance show that [2] = [Q( VaTHPSNA)] = § (1~ SRF Define the “Deep fade.event” and justify the definition. What is the probability [2] for deep fade event?. For a system operating in a fading channel if the PSD of the additive noise alone is doubled (all other quantities remaining same), docs the deep fade event change? Justi Define the term "degrees of freedom” and calculate the degrees of freedom for (i Ixn, SIMO fading channel, (ii) the nex! MISO fading channel and (iii) the nm x ne MIMO fading cliainel, where n, and n, are the number of transmit and the receive antennas respectively. we (¢) Show that there is a 4-dB energy loss when using PAM on only the F-channcl rather _ [3] than using QPSK on both the I and Q channels (In-phase and Quadrature-phase channels) although both modulations convey two bits of information: Compute the corresponding loss when one wants to transmit k bits of information using 2*—PAM _Zather than 2*—QAM. Assume k to be even, How does this loss depend on k? \Ja¥ What are the schemes that exploit all the available degrees of freedom in a 2x2 [I} MIMO fading channel? Justify your answer. “@ [&) For a TCM system operating in Rician fading channel with Rician parameter IC, {31 Vv snow that the pairwise error probabilily is upper bounded as below yur answer py P(sn i) $ T] —— _assuming usual notations (as discussed in the class)(Ask now if needed)!!. DY’ ‘From the bound in'3(a) above, obtain the design criteria for both the Rayleigh and [2] ician fading channels assuming high SNR. : (Q7 Recover the design criterion for TCM in AWGN channel (no fading) from3(a) and Tf 3(b) above. . I} Give an example to show that a TCM scheme that is best for an AWGN channel [I] may not be good for a fading channel. Justify your answer. K (2) State the rank and the determinant criterion to design a space-time code for a MIMO [| fading channel. mt oY’ ‘Show that the Alamouti code | 7)“ versity g th the Mi «decoding complexity same as that of the uncoded SISO stem, Wien the variables 21, :¢ take values from the finite signal set S. calculate its coding gain in terms of the paramict~rs of S. |: provides the maximum possible di- [3] a +2 Go What is the rate of the code | tes use? Justify your answer. typ + 3229 0 Uh Compare the code [ . ure a+2} tq = tar + JEx | ‘and decoding complexity. in complex symbols per channel [I] where a =z +Jrg and (3) 19 with the Alamonti code in terms of diversity gain, coding gain T) ‘Assuming that the channel response of a frequency selective fading channel has & finite number of taps L, explain how viewing this channel as a MISO fading channel itcan be shown that a diversity gain of L can be achieved with uncoded transmission combined with Maximum Likelihood Sequence Detection (MLSD). Explain how Viterbi algorithm can be used to implement 5(2) above. JA Explain how frequency diversity can be achieved in a frequency selective fading [3 channel using OFDM. _ ee DY” Compare the "Union bound” and the "Union-Bhattacharys bound” in terms of [2 importance/significance/usefulness/(any other aspect). Why these bounds arc used very often (almost always) when studying various aspects (like modulation, signal _Aets, coding, ....) of communication systems? ‘Show that for a TCM scheme desigried for a AWGN channel with Ungerboeck rules, [2] when the complex symbol duration is the same as thet of the uncoded systern, it is not possible to obtain coding gain without signal set expansion, - (3 Show that for communication systems operating in AWGN channels the choice of [3] the orthonormal basis does not have any impact on the average probability of error. Justify with suitable examples that this is-not true for systems operating in non- ILSD for the setting in. [2] frequency selective fading channels. Is this true for systems operating in frequency selective fading channels? Justify your answer. . pf Tn the problem of designing a communication system which one(s) of the following (2) _ is more important (if not equally important) than other(s)? (i) rate of transmission (bits/sec/Hz), (ii) bit error probability, (iii) bandwidth consumed and (iv) decoding ___ complexity. Justity your answer. Da Which one(s) of the following is more impdrtant (ifnot equally important) compared (2) to others: (i) Bit error probability, (ii) Symbol error probability and (iii) Codeword error probability ? Justify your answer with suitable examples. 7 ~~ (9 How is the role of interleaver used in Turbo coding is different from the role of (2) interleavers used in systems operating in fading channels? GOOD LUCK 1!!! P2217 DIGITAL CaMatt Auge s01y HOA TLION ‘ Sle Maries 26 © * Br/ (8) For aw uncoded 4-PSK system operating in AWGN chan vel show that power yain 5) can be achieved without bandwidth esp nsion by desig ng, 4 -atate TOM scherne Calculate the asymptotic couling gai sour TCM scherne {D1 Why one generally doubles the size of signal set ina TCM scheme? Wry not {I} quadinple of more and get more benelit © _ Show that for a TCM scheme designed for a AWGN channel with Ungerbord rites [s) when the comples symbol dura that of the uncoded system. 1 \ fet expansion _ extrinsic Info: mation” fH (b) Explain turbo coding and decoding with suitable block diagrams and mathematical {4] expression fe) Why the interleaver used in the turho e sroder seeds to be a large one? it (1) Explain why recursive syst scheme? Explain the BOJR algorithm ten s natic conve tional encoders are used in turbo coding |) The @ posterior probabilities (APP) of the tes and transitions of a Markor sourse ubserved through a noisy DMC plain how the algoritiy codes. Oy) of Sta) can Lae used for APP decoding of convolutional 1] © For the block code with the party ches matrix H consisting of two row vectors [1] [11010] aud (01101). draw a tretts, Explain how MAP decoding of linear block codes can be done using this trellis anil the algurithin of 3(a) Ty Obi “coherent comin ating in an AWGN chantul, For this -ecei ver ealuulate the average probability of ecorresporaling in the ML receiver structure for a bouary feation system Jo} ror. How does this difer from the averaye probability of error for th conerent_ communication sy.stew (b) Write down the probabili ion. Explei or Rician fadir feusity: hu ons for Rayleigh, Rician and lognormal (2) Ww a wi channel, ci gets moxleled as 9 sdiflused multipath el modulat 1 state mformation is available at che receiver, ng che minifiium product «lis (2) fectPve tinsn ior code 1g the criteria of fa) above tance for codes with small elfectiv len, large effect to Be iathis & amore ia we lengths, 7 such a POM scheme of + convolutional TCM code can athieve is hinited by the mumber of input bits m anc the number of encoder memorizs v and is given by Lmax = (£J +) @ Give an example to show that a TCM scheme that is best for an AWGN channel fa ot be good for a fading channel ton that the maxima effe length Laney Gist 72) may ~~ —~~@ ~~ a) =@ ~~ -t GOOD LUCK 2 2 2 =? CL SSeS : Be : — ~~ =) A E2211 DIGITAL COMMUNICATION TEST. Aug-Dec, 2013 Max. Marks=25 T (a) Define “vector space” with suitable examples’ Show that given M vealvalued, [al Finite-energy functions {tm(t)}, taking non-zero values only in the interval {0,7}, there exist WV ) Write down the probability density: functions for Rayleigh, Rican and lognormal distri 4} bution. Explain how a wireless diflused multipath channel gets modeled as a Ravleish or Rician favling channel GFT Show tira in a Rayleigh fading chann the average error probability p: ul cok ¥ non-coherent comnunication s) tem oper i) receiver, [A wwith channel gains known perfectly at U nt detection of BPSIC o1 a symbol by symbol basis is given by } (1 ~ / SEAR) whose high-SNK approximation is zghrn* Define a “déep facle event” for a fading channel and show that at high-SNR, the probability [11 of deep fade is approximately gq in (a) above. Define the term “egrees of freedoms” and calculate the dexrees of freedom for (i) the Mel |i. “5 MISU fading, channel, (ii) the 1x2 SIMO fading channel and (iii) the 2x2 MIMO fading J channel Show that. there-is a 4-dB energy loss when using 4-PAM on only the I-channel rather than }4j/ using QPSK on both the I and Q channels (In:phase anc the Quadrature-phase channels) althotigh Loth modulations convey two hits of information. Compute the corresponding loss when one wants" to transmit & bits of information using 2'—-PAM rather than 2*-QAM. . Yoir can ‘ass: «does the depend on eo 7 Oban tne code clesign criteria for trellis coded modulation (TC fels when the perfect channel st yw Using Ui rin of ZQay ADT i r fr ish codes with small effective lengths is more effective than lor co a large eftective leingts Q (c)_ For such a*TCM scheme of 2(a) show that the maxinmm elfective length Lor that a [af convolutional TCM code can achieve ix limited hy the number ot input bits m and che humber of encoder memories v and is given by Lanne = UZ) +1 @ Give an example to show that a ‘TEM scheme that is L be good tor a fading channel, 7 (e) (Show that for a TCM scheme designed for a AWGN chanuel w ih the complex symbol duration is the same-as that of the uncodei § to obtain coring gain without signal set expansion’ Qf) Derive the rank and the determinant c&terion fo a space-time code for a MIMO Tuling channel (b) Show that che Alamouti code provides rhe maximum possible diversity gain with the ML laf @ dlecorling complesity same as that of she wien! SISO system! Caleulate its céding gain! © What are the schemes that exploit all the available degrees of freedom iva 2x2 MIMO (2 fading channel? Ji fs © YP |Show that tor ns 47 the wnmber of autem Tading char a tify vou ANSWe emi with receive diversity eel Explain hy ina AWGN channel the choice of basis iamerions femtM}. a = 12 does not aficet the probability of error ble examples. Show that this iy not true for fading channels GOOD LUCK vooEELS 211 DIGITAL COMMUNICATION TEST:1 Aug-Dec, 2012 Max. Marks=25 1 (a Define “vector space” with suitable examples and show that given Af realsvalued finite-energy funetions (rm (t)} defined on 0,7}, there exist N ] ML Decoder and (cl) Miniimuin-Distance decocler anid'(e) Bounded-Distgnce Decoder and give an example for cach. Under what conditions a bounded-distanée decoder will be an ML decoder? © Show th ry cade Fe used on x BSC (Binary Symmetric Channel) the [2] reeciver'which decides in favor of the codeword which is closest to the received vector in Hanning distance is an optimum ML receivers State the axioms of uncertainty and show that there is 01 all the axioms. - ly one function that satishes [3] Consider. the following two signal sets for ary, communication over an AWGN channel with ML detection - . 7 Seti: se /F. osesitacess; 3 ss) VE Ostsreresess silQ= (E-acisz, mers Set 2: a(t) = VE, 3StS4; l= VE, 11 <2 s(t) = VE, 2<¢53; s(t)=VE. 0<0 <1 . When the SNR is very high which signal set is better 7 Justify your answer, {a} (b)_ Por a sufficiently Targe value of the carrier frequency Je, find the mininun frequengy 2) nals to be orthogonal, where the ile of which’ they take the value separation Ay required for the following pair of si signals arc shown for the interval 0 < ¢-< T. out © ev: (= yf feos(2a fel) and sft = \Reoaaus, Apo oo } . wt g £2211 pig ITAL MAJON EXant Communication Avg-Dee, 2010 Max. Marks=50 1 QY Prove: Given Af finite. Uunit-energy orthogonal ‘energy functions (x»(t)} defined on [0,7], there exist pendent over the field of real numbers (b)_ Prove the inequality: Q(a)¢ je. S 2) _ $ Fix + (©) Draw-a block diagram for an optimum ML receiver for the 8-ary communicatio iy) using the signals s;, i= 1,2,--- ,8, given below (for the ranges of ¢ with no value specific, the sigia! takes the value zero) in & Facn hanvel ot psp alt) = OStery solth=h eter sy(th=1, OS tS? sa) =1, 20S soltal, OSES IAPS CSH soll) srolt.= 1, A

You might also like