Solutions to the Exam Digital Communications I given on the 11th of June = 111 and g 2. c 2

Similar documents
UNIVERSITY OF CALIFORNIA College of Engineering Department of Electrical Engineering and Computer Sciences EECS 121 FINAL EXAM

Block Diagram of a DCS in 411

EE3723 : Digital Communications

Lecture 8. Digital Communications Part III. Digital Demodulation

Multiphase Shift Keying (MPSK) Lecture 8. Constellation. Decision Regions. s i. 2 T cos 2π f c t iφ 0 t As iφ 1 t. t As. A c i.

Lecture 4. Goals: Be able to determine bandwidth of digital signals. Be able to convert a signal from baseband to passband and back IV-1

28. Narrowband Noise Representation

A First Course in Digital Communications

EE456 Digital Communications

Chapter 3: Signal Transmission and Filtering. A. Bruce Carlson Paul B. Crilly 2010 The McGraw-Hill Companies

Demodulation of Digitally Modulated Signals

( ) ( ) if t = t. It must satisfy the identity. So, bulkiness of the unit impulse (hyper)function is equal to 1. The defining characteristic is

Zürich. ETH Master Course: L Autonomous Mobile Robots Localization II

Vehicle Arrival Models : Headway

(10) (a) Derive and plot the spectrum of y. Discuss how the seasonality in the process is evident in spectrum.

GMM - Generalized Method of Moments

Transform Techniques. Moment Generating Function

Spring Ammar Abu-Hudrouss Islamic University Gaza

ENERGY SEPARATION AND DEMODULATION OF CPM SIGNALS

6.003 Homework #13 Solutions

A Bayesian Approach to Spectral Analysis

ADDITIONAL PROBLEMS (a) Find the Fourier transform of the half-cosine pulse shown in Fig. 2.40(a). Additional Problems 91

control properties under both Gaussian and burst noise conditions. In the ~isappointing in comparison with convolutional code systems designed

Answers to Exercises in Chapter 7 - Correlation Functions

CHAPTER 2 Signals And Spectra

Problem Formulation in Communication Systems

Impact of Channel Estimation Errors on Space Time Trellis Codes

Macroeconomic Theory Ph.D. Qualifying Examination Fall 2005 ANSWER EACH PART IN A SEPARATE BLUE BOOK. PART ONE: ANSWER IN BOOK 1 WEIGHT 1/3

EE 313 Linear Signals & Systems (Fall 2018) Solution Set for Homework #8 on Continuous-Time Signals & Systems

CS Homework Week 2 ( 2.25, 3.22, 4.9)

Asymptotic Equipartition Property - Seminar 3, part 1

Georey E. Hinton. University oftoronto. Technical Report CRG-TR February 22, Abstract

Comparing Means: t-tests for One Sample & Two Related Samples

Lecture 20: Riccati Equations and Least Squares Feedback Control

20. Applications of the Genetic-Drift Model

6.003 Homework #8 Solutions

Lecture 1 Overview. course mechanics. outline & topics. what is a linear dynamical system? why study linear systems? some examples

Introduction D P. r = constant discount rate, g = Gordon Model (1962): constant dividend growth rate.

Anno accademico 2006/2007. Davide Migliore

EE 301 Lab 2 Convolution

KINEMATICS IN ONE DIMENSION

m = 41 members n = 27 (nonfounders), f = 14 (founders) 8 markers from chromosome 19

I Let E(v! v 0 ) denote the event that v 0 is selected instead of v I The block error probability is the union of such events

Essential Microeconomics : OPTIMAL CONTROL 1. Consider the following class of optimization problems

AC Circuits AC Circuit with only R AC circuit with only L AC circuit with only C AC circuit with LRC phasors Resonance Transformers

Elements of Stochastic Processes Lecture II Hamid R. Rabiee

ERROR LOCATING CODES AND EXTENDED HAMMING CODE. Pankaj Kumar Das. 1. Introduction and preliminaries

( ) = Q 0. ( ) R = R dq. ( t) = I t

Problem Set 5. Graduate Macro II, Spring 2017 The University of Notre Dame Professor Sims

Suboptimal MIMO Detector based on Viterbi Algorithm

d 1 = c 1 b 2 - b 1 c 2 d 2 = c 1 b 3 - b 1 c 3

Sequential Importance Resampling (SIR) Particle Filter

INTRODUCTION TO MACHINE LEARNING 3RD EDITION

Math 106: Review for Final Exam, Part II. (x x 0 ) 2 = !

0.1 MAXIMUM LIKELIHOOD ESTIMATION EXPLAINED

Impacts of both Tx and Rx IQ Imbalances on OFDM Systems - Analytical Approach

L07. KALMAN FILTERING FOR NON-LINEAR SYSTEMS. NA568 Mobile Robotics: Methods & Algorithms

Learning a Class from Examples. Training set X. Class C 1. Class C of a family car. Output: Input representation: x 1 : price, x 2 : engine power

Some Basic Information about M-S-D Systems

The Potential Effectiveness of the Detection of Pulsed Signals in the Non-Uniform Sampling

Diebold, Chapter 7. Francis X. Diebold, Elements of Forecasting, 4th Edition (Mason, Ohio: Cengage Learning, 2006). Chapter 7. Characterizing Cycles

Kalman filtering for maximum likelihood estimation given corrupted observations.

Recursive Least-Squares Fixed-Interval Smoother Using Covariance Information based on Innovation Approach in Linear Continuous Stochastic Systems

Excel-Based Solution Method For The Optimal Policy Of The Hadley And Whittin s Exact Model With Arma Demand

Suggested Solutions to Assignment 4 (REQUIRED) Submisson Deadline and Location: March 27 in Class

Solutions for Assignment 2

Økonomisk Kandidateksamen 2005(II) Econometrics 2. Solution

Ecological Archives E A1. Meghan A. Duffy, Spencer R. Hall, Carla E. Cáceres, and Anthony R. Ives.

SOLUTIONS TO ECE 3084

State-Space Models. Initialization, Estimation and Smoothing of the Kalman Filter

Reliability of Technical Systems

Solutions from Chapter 9.1 and 9.2

2.160 System Identification, Estimation, and Learning. Lecture Notes No. 8. March 6, 2006

Wireless Communication Channel Overview

Financial Econometrics Kalman Filter: some applications to Finance University of Evry - Master 2

DEPARTMENT OF ELECTRICAL AND ELECTRONIC ENGINEERING EXAMINATIONS 2008

Robert Kollmann. 6 September 2017

Explaining Total Factor Productivity. Ulrich Kohli University of Geneva December 2015

Learning a Class from Examples. Training set X. Class C 1. Class C of a family car. Output: Input representation: x 1 : price, x 2 : engine power

( ) ( ) ( ) () () Signals And Systems Exam#1. 1. Given x(t) and y(t) below: x(t) y(t) (A) Give the expression of x(t) in terms of step functions.

Two Popular Bayesian Estimators: Particle and Kalman Filters. McGill COMP 765 Sept 14 th, 2017

STA 114: Statistics. Notes 2. Statistical Models and the Likelihood Function

Filtering Turbulent Signals Using Gaussian and non-gaussian Filters with Model Error

S Radio transmission and network access Exercise 1-2

MATH 128A, SUMMER 2009, FINAL EXAM SOLUTION

Cooperative Ph.D. Program in School of Economic Sciences and Finance QUALIFYING EXAMINATION IN MACROECONOMICS. August 8, :45 a.m. to 1:00 p.m.

The Brock-Mirman Stochastic Growth Model

Solutions - Midterm Exam

Chapter Floating Point Representation

Outline Chapter 2: Signals and Systems

WEEK-3 Recitation PHYS 131. of the projectile s velocity remains constant throughout the motion, since the acceleration a x

Representing a Signal. Continuous-Time Fourier Methods. Linearity and Superposition. Real and Complex Sinusoids. Jean Baptiste Joseph Fourier

A complex discrete (or digital) signal x(n) is defined in a

Testing for a Single Factor Model in the Multivariate State Space Framework

EE102 Homework 5 and 6 Solutions

1.6. Slopes of Tangents and Instantaneous Rate of Change

Y 0.4Y 0.45Y Y to a proper ARMA specification.

Linear Response Theory: The connection between QFT and experiments

5. Stochastic processes (1)

Christos Papadimitriou & Luca Trevisan November 22, 2016

Transcription:

Soluions o he Exam Digial Communicaions I given on he 11h of June 2007 Quesion 1 (14p) a) (2p) If X and Y are independen Gaussian variables, hen E [ XY ]=0 always. (Answer wih RUE or FALSE) ANSWER: False. (Only if E(X)=0 or E(Y)=0.) b) (3p) Consider a rae 1/2 convoluional code wih generaor sequences g 1 = 111 and g 2 = 110. Assume ha BPSK modulaion is used o ransmi he coded bis. Draw a shif regiser for he encoder. ANSWER: c) (2p) One advanage of non-coheren demodulaion over coheren demodulaion, is ha he demodulaor does no need o implemen phase esimaion. (Answer wih RUE or FALSE) ANSWER: rue. d) (3p) A receiver ha implemens he ML decision rule is always opimal in he sense of minimum symbol error probabiliy. (Answer wih RUE or FALSE) ANSWER: False. e) (2p) he MAP decision rule is a special case of he ML rule. ANSWER: False. ML is a special case of MAP. c 2 (Answer wih RUE or FALSE) f) (2p) PSK is a special case of QAM. (Answer wih RUE or FALSE) ANSWER: rue. Quesion 2 (12p) Here we sudy a binary sysem wih signal wo alernaives: s 1 and s 2. he bis are assumed o be saisically independen and of equal probabiliy. he sysem is an AWGN channel wih noise power specral densiy given by N 0 /2. c 1

s 1 A s 2 A /2 -A a) Please express E b (he average energy per bi) as a funcion of A and. (2p) Soluion: he energy for s 1 is E 1 = 0 s1 2 d= A 2 and he energy for s 2 is E 2 = 0 s2 2 d= A 2. Hence, he average energy per bi is E b = A 2. b) Find a basis for he signal space and draw he signal consellaion ha is he vecors corresponding corresponding o s 1 and s 2. Please label he axis in erms of E b. (4p) Soluion: One possible basis is { 1, 2 } where i =s i / E b. Hence, he signal consellaion is: s 1 =[ E b, 0] and s 2 =[0, E b ] c) Wha is he smalles value of E b / N 0 in db required o reach a bi error probabiliy of 10 3 if we assume ha we are using he opimum receiver (which minimizes he bi error probabiliy)? (4p) Soluion: he maximum likelihood receiver will minimize he bi error probabiliy since all signal

alernaives are equally likely. he disance beween he signal alernaives is d = 2 E b and he bi error probabiliy is P b =Q d 2 /2N 0 =Q E b / N 0 =10 3. he required E b / N 0 in db is approximaely 9.80dB. d) Please draw he decision regions for he receiver in par c). (2p) Soluion: he ML receiver is a minimum-disance receiver and he decision regions Z 1 and Z 2 are depiced below. Quesion 3 (12p) In his problem, we will consider a convoluional encoder. I is given in he figure below. he informaion bi b gives rise o hree coded bis c 1, c 2 and c 3, which are ransmied over an AWGN channel using binary PAM wih roo-raised cosine pulses wih roll-off facor =0.2. he PAM pulses are ransmied a a rae of 1000 pulses/second. If he coded bi is 1, hen he ransmied ampliude is posiive. Conversely, if he coded bi is 0, hen he ransmied ampliude is negaive. b c1 c2 a) Draw he sae diagram and a rellis secion of he encoder. Please make sure o label he ransiions. (4p) c3

Soluion: he sae diagram and rellis are shown in he figure below. c) Now suppose he sampled oupu of he receiver mached filer is {1.51, 0.63, 0.04, 1.14, 0.56, 0.57, 0.07, 1.53, 0.9, 1.68, 0.9, 0.98, 1.99, 0.04, 0.76} his sequence corresponds o he ransmied sequence c 1, c 2, c 3, c 1, c 2, c 3,... which is generaed from a packe of hree informaion bis. he encoder is assumed o be in he all-zero sae a he beginning and ending of he ransmission. he encoder is forced ino he ending sae by appending wo zero bis o he informaion bi sequence. Use hard decision decoding o esimae he ransmied informaion bis. (8p) Soluion: he hard decoded channel bis are {1, 1, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1}. he rellis wih he decoded pah (he pah a minimum Hamming disance from he received channel bis) is show below.

We see ha he decoded informaion bis are {0, 0, 1}. Quesion 4 (12p) In his problem, we are considering a binary PAM sysem. he pulse shape is given by g and he daa rae is R b =1/ b. We can hus conclude ha he ime beween consecuive pulses equals b seconds. Furhermore, s = k= b k g k b, where b k { 1, 1} is he k h ransmied bi. he sysem is depiced in he block diagram below. Here, sgn{q} denoes he sign of q, ha is if q is posiive hen sgn{q}=1 and if q is negaive hen sgn{q}=-1. Furhermore, he sampling of he kh bi is performed a ime =k b as indicaed in he figure. he channel is an AWGN channel. he addiive noise is denoed n() and has noise power specral densiy N 0 /2. he decision on he k h bi is given by he sign of y k b, where y is he oupu from he mached filer (which has impulse response g R ). he arge bi error rae is P b =10 4, and he energy per bi is denoed by E b. n() =k b b k Binary PAM g s() g R y() sgn{.} b k g R =g A a) Wha is he maximum possible daa rae if he ransmission mus be ISI-free? Answer wih an expression in. (6p) Soluion: I follows from he block diagram ha y = l = b l x l b v where x =g g R and v =n g R, and he sampled oupu from he mached filer is y k b = l = b l x k b l b v k b =b k x, l k b l x k l b v k b here will be no ISI if x n b =0 for all nonzero n. From he figure, i is obvious ha his is equivalen o n b 2 b. Hence, he maximum daa rae for ISI free ransmission is R b =1 /.

x A 2 b) Wha is he required E b / N 0 in db for he ISI-free sysem o mee he bi error arge? (6p) Soluion: he kh noise sample is v k b =n g R =k b. he noise sample will be a zeromean Gaussian random variable wih variance E g N 0 /2, where E g is he energy of g R. g 2 R d= A 2. Clearly, E g is also he energy of g and since b k =±1, we E g = have ha E b =E g. Condiioned on ha b k = 1, a bi error occurs wih probabiliy Pr { y k b 0 b k = 1}=Pr { A 2 n k b 0}=Pr { E b n k b 0} Pr {n k b / E b N 0 /2 2 E b / N 0 }=Q 2 E b / N 0 Due o symmery, P b = P b b k = 1 = P b b k = 1 and E b / N 0 =...=8.39dB. 2