Diversity for Wireless Communications

Similar documents
An Efficient Selective Receiver for Multiple-Input Multiple-Output Scheme

Idea is to sample from a different distribution that picks points in important regions of the sample space. Want ( ) ( ) ( ) E f X = f x g x dx

Research on Efficient Turbo Frequency Domain Equalization in STBC-MIMO System

Signal,autocorrelation -0.6

Objectives of Multiple Regression

Basics of Information Theory: Markku Juntti. Basic concepts and tools 1 Introduction 2 Entropy, relative entropy and mutual information

Generative classification models

Model Fitting, RANSAC. Jana Kosecka

Transforms that are commonly used are separable

CS 2750 Machine Learning. Lecture 8. Linear regression. CS 2750 Machine Learning. Linear regression. is a linear combination of input components x

( ) = ( ) ( ) Chapter 13 Asymptotic Theory and Stochastic Regressors. Stochastic regressors model

Nonlinear joint transmit-receive processing for coordinated multi-cell systems: centralized and decentralized

EP2200 Queueing theory and teletraffic systems. Queueing networks. Viktoria Fodor KTH EES/LCN KTH EES/LCN

Reduced Complexity MIMO MMSE-DFE

General Method for Calculating Chemical Equilibrium Composition

ECE 559: Wireless Communication Project Report Diversity Multiplexing Tradeoff in MIMO Channels with partial CSIT. Hoa Pham

Kernel-based Methods and Support Vector Machines

Sample Allocation under a Population Model and Stratified Inclusion Probability Proportionate to Size Sampling

PERFORMANCE EVALUATION OF C-BLAST MIMO SYSTEMS USING MMSE DETECTION ALGORITHM

Comparison of Dual to Ratio-Cum-Product Estimators of Population Mean

Principal Components. Analysis. Basic Intuition. A Method of Self Organized Learning

Wireless Link Properties

LINEAR EQUALIZERS & NONLINEAR EQUALIZERS. Prepared by Deepa.T, Asst.Prof. /TCE

CHANNEL IMPAIRMENTS & EQUALIZATION. Prepared by Deepa.T, Asst.Prof. /TCE

Study of the capacity of Optical Network On Chip based on MIMO (Multiple Input Multiple Output) system

Reliability evaluation of distribution network based on improved non. sequential Monte Carlo method

TESTS BASED ON MAXIMUM LIKELIHOOD

Lecture 3 Probability review (cont d)

Supervised learning: Linear regression Logistic regression

Dimensionality reduction Feature selection

Lecture Notes 2. The ability to manipulate matrices is critical in economics.

ENGI 3423 Simple Linear Regression Page 12-01

3. Basic Concepts: Consequences and Properties

Polynomial Prediction RLS Channel Estimation for DS-CDMA Frequency-domain Equalization

13. Parametric and Non-Parametric Uncertainties, Radial Basis Functions and Neural Network Approximations

TCP in Presence of Bursty Losses

Functions of Random Variables

Asymptotic Formulas Composite Numbers II

hp calculators HP 30S Statistics Averages and Standard Deviations Average and Standard Deviation Practice Finding Averages and Standard Deviations

y y ˆ i i Difference between

{ }{ ( )} (, ) = ( ) ( ) ( ) Chapter 14 Exercises in Sampling Theory. Exercise 1 (Simple random sampling): Solution:

Chapter 10 Two Stage Sampling (Subsampling)

Model-Based Networked Control System Stability Based on Packet Drop Distributions

Numerical Analysis Formulae Booklet

Summary of the lecture in Biostatistics

9.1 Introduction to the probit and logit models

Lecture 02: Bounding tail distributions of a random variable

Binary classification: Support Vector Machines

Power Flow S + Buses with either or both Generator Load S G1 S G2 S G3 S D3 S D1 S D4 S D5. S Dk. Injection S G1

Error probability and error stream properties in channel with slow Rician fading

Chapter 2 - Free Vibration of Multi-Degree-of-Freedom Systems - II

Likewise, properties of the optimal policy for equipment replacement & maintenance problems can be used to reduce the computation.

Lecture 3. Sampling, sampling distributions, and parameter estimation

VARIANCE ESTIMATION FROM COMPLEX SURVEYS USING BALANCED REPEATED REPLICATION

Lecture 3. Least Squares Fitting. Optimization Trinity 2014 P.H.S.Torr. Classic least squares. Total least squares.

PGE 310: Formulation and Solution in Geosystems Engineering. Dr. Balhoff. Interpolation

Chapter Two. An Introduction to Regression ( )

Homework 1: Solutions Sid Banerjee Problem 1: (Practice with Asymptotic Notation) ORIE 4520: Stochastics at Scale Fall 2015

SIMULTANEOUS wireless information and power transfer

Basics of Information Theory: Markku Juntti. Basic concepts and tools 1 Introduction 2 Entropy, relative entropy and mutual information

On the Delay-Throughput Tradeoff in Distributed Wireless Networks

IN Massive Multiple-Input Multiple-Output (MIMO), the

CODING & MODULATION Prof. Ing. Anton Čižmár, PhD.

THE ROYAL STATISTICAL SOCIETY 2016 EXAMINATIONS SOLUTIONS HIGHER CERTIFICATE MODULE 5

Lecture 16: Backpropogation Algorithm Neural Networks with smooth activation functions

WIDE AREA, FINE RESOLUTION SAR FROM MULTI-APERTURE RADAR ARRAYS

1 Review and Overview

Application of Calibration Approach for Regression Coefficient Estimation under Two-stage Sampling Design

Solving Constrained Flow-Shop Scheduling. Problems with Three Machines

Nonparametric Regression with Trapezoidal Fuzzy Data

ESS Line Fitting

Simple Linear Regression

Multiple Linear Regression Analysis

Introduction to Numerical Differentiation and Interpolation March 10, !=1 1!=1 2!=2 3!=6 4!=24 5!= 120

LECTURE 9: Principal Components Analysis

Overcoming Limitations of Sampling for Aggregation Queries

4. Standard Regression Model and Spatial Dependence Tests

PREAMBLE DESIGN FOR THE DIGITAL COMPENSATION OF TX LEAKAGE IN ZERO-IF RECEIVERS. Andreas Frotzscher and Gerhard Fettweis

ECONOMETRIC THEORY. MODULE VIII Lecture - 26 Heteroskedasticity

Lecture Notes Forecasting the process of estimating or predicting unknown situations

TOWARDS INCREASING THE CAPACITY IN MIMO WIRELESS COMMUNICATION SYSTEMS

best estimate (mean) for X uncertainty or error in the measurement (systematic, random or statistical) best

Correlation and Regression Analysis

Statistics MINITAB - Lab 5

Lecture 7. Confidence Intervals and Hypothesis Tests in the Simple CLR Model

UNIVERSITY OF OSLO DEPARTMENT OF ECONOMICS

Stochastic control viewpoint in coding and information theory for communications

Unsupervised Learning and Other Neural Networks

Linear Regression Linear Regression with Shrinkage. Some slides are due to Tommi Jaakkola, MIT AI Lab

ELEC 6041 LECTURE NOTES WEEK 1 Dr. Amir G. Aghdam Concordia University

Detection and Estimation Theory

THE ROYAL STATISTICAL SOCIETY 2009 EXAMINATIONS SOLUTIONS GRADUATE DIPLOMA MODULAR FORMAT MODULE 2 STATISTICAL INFERENCE

Energy Efficient Transmission Probability and Power Control in Random Access Networks

Linear regression (cont.) Linear methods for classification

Wu-Hausman Test: But if X and ε are independent, βˆ. ECON 324 Page 1

MA/CSSE 473 Day 27. Dynamic programming

A Note on Ratio Estimators in two Stage Sampling

Lecture 7: Linear and quadratic classifiers

Chapter 13 Student Lecture Notes 13-1

Decode. Encode. Page 6. Speech Coding E.4.14 Speech Processing. Speech Coding. Lecture 8. Speech Coding. s(n)

Transcription:

Dverst for Wreless Commucatos For AWG caels, te probablt of error decas epoetall wt SR c.f. Q. fucto P e Q SR SR e For fadg caels, te deca s muc slower. For Raleg-fadg caels cael coeffcet s comple Gaussa=>ampltude s Raleg te probablt of error decas versel wt SR P e K SR

Proof

Proof 3

Sce power s a lmted ad precous resource wreless commucatos, t s udesrable to crease trasmt power to compesate for fadg power cotrol eve f cael s ow at trasmtter wc s ot realstc. Oter problems wt creasg te power are creased terferece to oter users ad creased pea to average rato wc would crease amplfer bac off to operate te lear rego

A alteratve more desrable soluto to ts fadg problem s troug dverst trasmsso/recepto were we sed te same formato over D depedet caels wc are ulel to all fade togeter Ts dverst tecque reduces probablt of error to P e SR e D.As D we aceve epoetal deca wt SR same as Gaussa. I practce, eve D=-4 results sgfcat performace mprovemet

Dverst ca be aceved several forms cludg Tme Dverst: Trasmt same formato smbols at tmes separated b more ta te coerece tme of te cael=> same smbols wll eperece depedet fadg=> temporal dverst Frequec Dverst: est were ee we trasmt ta t same fo over frequec carrers separated b more ta te coerece badwdt of te cael

Bot tme ad frequec dverstes are aceved at te epese of rate loss 3Spatal Dverst : trasmt or receve formato to/from depedet ateas separated b more ta te coerece dstace of te cael. Does t volve rate loss but addtoal ateas & RF cas. Icludes polarzato dverstt To esure depedece, ateas are separated b more ta alf wavelegt at termal ad b more ta wavelegts at base stato

Remar: Multple trasmt ateas ca be used to Icrease spatal dverst of sstems b sedg te same formato stream over depedet spatal caels Space-Tme Codg Or crease rate of sstem b spatall multpleg depedet formato streams BLAST 3Aceve a sutable tradeoff of rate ad dverstt

Sgle Iput Multple Output g p p p Receve DverstSIMO M r r r r M M M r M to mamze SR at recever => use matced r r r M M M flter also ow as mamal rato comber

Sgle Iput Multple Output z r... M dverst order M r ~ Cael assumed perfectl ow at recever REMARK: Receve dverst s more sutable for upl moble to base stato were we ca mplemet multple ateas more easl ta at user termal

Spato-Temporal Equalzato To mtgate sgal fadg effects o wreless caels, multple receve ateas are used for dverst ad ter outputs are combed Select sgal wt gest SR selecto dverst Sum wt equal wegtg equal-ga combg Sum wt wegtg proportoal to SR mamal- rato combg O frequec-selectve caels, atea outputs are fltered b tapped-dela-les TDL ad ter outputs are combed b tapped-dela-les TDL ad ter outputs are combed. TDL FIR wegts optmzed to mmze MSE

Spato-Temporal Equalzato H H I I w w - Dec ˆ H M r M r M r M r w b I M r j j j z I : j,,..., r Sgle-Iput Multple-Output SIMO Cael M were s o. of receve ateas

bloc Ts structure ca be geeralzed to perform mult-user detecto usg full cross-coupled mult-put mult-output feedforward & feedbac flters 3

Trasmt Dverst MISO I cellular sstems, dowl from base stato to termal s te bottleec Iteret traffc asmmetr To mprove dowl performace, trasmt same formato usg multple ateas at base stato Trasmt Dverst Trasmt dverst eeps user termals cost effectve small sze ad low power Trasmt ateas are separated wde eoug to esure depedet fadg over multple trasmsso caels Sce o cael formato s assumed at trasmtter, total trasmt power dvded equall amog ateas Smplest form of spatal trasmt dverst s dela dverst dverst ga ol. Advaced forms ow as space-tme codg aceve codg & dverst gas 4

Dela Dverst over Flat-Fadg Caels Here, we assume cael s OT ow at trasmtter Dela Dverst Sceme D D D D D D D D D Y Dela dverst creates a artfcal ISI cael tat ca be equalzed e.g. usg D Vterb algortm to get performace mprovemet secod-orderorder dverst

Dela Dverst over ISI Caels D D D D Preflter D L D D MLSE Equalzer Aceves dverst b creatg cotrolled ISI Bacward compatble wt estg SISO stadards ad user termals Equalzer sees a equvalet SISO cael wt more taps spatal dverst trasformed to temporal dverst D D D D D D L D D D D D How to coose L? D D D eq 6

Te Alamout Sceme Code over two cosecutve b l d l - smbols ad assume cael s fed over tese smbols * * H Aceves dverst order H le dela dverst but at lower decodg complet

Alamout Sceme Cot d REMARK: Te equvalet cael matr s ortogoal, ece matced flter s ML optmal H H HH I Performace of Trasmt/Receve atea dverst s degraded te presece of atea a correlato 3 Te Alamout sceme falls uder te class of space-tme bloc codg STBC scemes

Alamout Sceme Decodg: Matced flter s optmal ere! z H H H H ~ ~ d order dverst ~ s stll Gaussa, zero mea ad as varace SR mproved b factor

Alamout Sceme Advatages of Alamout Sceme Mamumdverst d order Rate formato smbols tme slots=> full- rateuder restrcto of o costellato epaso Ope loop o eed for cael owledge at T Low decodg complet lear Drawbac: more o ts later Caot be eteded to more ta trasmt ateas for comple sgal costellatos wtout rate loss.e. rate < or sacrfcg smple lear decodg complet or costellato epaso

Te Alamout Sceme for ISI Caels For ISI Caels, te Alamout sceme sould be mplemeted at a bloc ot smbol level as flat-fadgfadg case order to realze multpat dverst addto to te d order spatal dverst. Tere are at least 3 was of dog ts te tme doma called tme-reversal spacetme bloc codg TR-STBC or te frequec- doma usg sgle-carrer FDE or usg multcarrer OFDM trasmsso. I te sequel, sgle carrer FDE-STBC wll be descrbed

Te Alamout SC-FDE Sceme for ISI Caels CP CP Remove CP FFT ˆ ˆ SBS IFFT FDE

ECODIG RULE Deote te t smbol of te t trasmtted bloc of legt from atea "" b were deotes,.te te modulo dl - for operato. Tag te DFT of te put blocs, we get tme de,,.. -,,4,... wc s m m m m for frequec b m te Alamout sceme at te bloc level.,,.. -,,4,...

Iput-Output Relatosp p p p z H H j j j j j j j j cclc pref due to te use of crculat matrces are ad H H were cclc pref due to te use of :dagoal matr : FFT Q Q Q H Q Q H j j j j g Q Q

Recever Operatos p FFT : j j j j j j j Z Q Y Z Y blocs : Processg pars of Z - Z Y - Y Y cosecutve blocs matrces are assumed fed over were te cael

Recever Operatos Z ~ ~ Appl space tme matced flter 3 Y f f l d Z ~Z ~ ~ Y Y Y Y blocs are decoupled te two formato ow,. to ortogoalt of due : Z ~ ~ ~, : Z ~ ~ Y Y wc ca be equalzed as SISO FDE -STBC, : Z Y : Z ~ ~ Y ~ b s frequec m m m m m

3 MIMO Case B sedg depedet formato streams from multple trasmt ateas ad detectg tem at recever usg multple receve ateas acevable trougput o,we crease Assumptos: FIR Cael from t trasmt atea to j t receve atea wt memor j deoted b j

MIMO - Assumptos to use flterg based equalzato at recever to detect fo streams. Groupg te Y were H ad m cael H m m m ma, j, j m m outputs, m m we ave

MIMO - Assumptos p Over a bloc of f output smbols Y H H H H H f f f f f f f Y Y f f f f Y Bl T lt C l M d l Bloc Toepltz Cael Model

MIMO - DFE Cosder te case W, + + SBS ˆ W,, W, W + + SBS ˆ Feedforward Matr Flter +, e b, b, b Feedbac Matr Flter +, e b

MIMO Equalzato To crease te data rate over wreless ls, depedet formato streams are trasmtted ad receved smultaeousl usg multple trasmt & receve ateas. Ts spatal multpleg creates a MIMO cael MIMO caels also arse we multple l users eac equpped wt atea trasmt same cell ad same tme slot to a base stato equpped w/ multple l ateas but trasmt ateas are o colocated ts case! For MIMO case, MLSE Vterb equalzer requres M b states epoetal o. put ateas also! 3 All SISO equalzato scemes studed ts course ca be eteded to te MIMO case

3

33

MIMO - DFE Remars: Te MIMO DFE ca also be used for sgle atea mult-user commucatos Ts model ca also be used to develop a MIMO Cael sorteg sceme, e, for detals see,. Al-Dar, FIR Cael Sorteg Equalzers for MIMO ISI Caels, IEEE Tras. Comm. Feb

Matr Feed Forward Flter f W W W W Feedbac Flter Matr b b b b B B B - O I - B* O I b b B B B I,, were W w w,, were W w w

B, b,, b, b b Furtermore, defe B ~ O B were f -s te decso dela

Performace Aalss : Error Vector ~ * Y W B E f f : : Y W B E f f f f correlato matr : Error Auto - ee E E E R ~ ~ ~ ~ YY YY Y Y YY Y R R B W R R R B W B R R R R B YY R Y R B W

Usg te Ortogoalt Prcple ~ W * B * R R opt R ee p ~ B ~ B opt ~ * R R R B * R H * R H * R ~ B R ~ B ~ B 38

Performace Aalss Assumptos: Ol prevous decsos o all formato streams are avalable,.e. all 4 feedbac flters for te eample are strctl causal B I We ave te cove optmzato problem m tracer m traceb R B ee ~ B were I b ~ B ad C ~ * ~ * subject to B ~ * C I

It ca be sow tat te soluto s gve b ~ R R C Remar: B opt, C R C R ee m We stll eed to optmze decso dela. I fact, we ca allow for dfferet for eac formato stream but t would crease computatoal complet

Specal Case: FIR MIMO MMSE-LE Set B I ad B for,,... defe g I R ee,mmse-le,mimo R s : cose to g Equalzer Coeffcets : R g, mmze : W MMSE-LE f trace b of R R R ee g

Oter Specal Cases: SISO DFE/LE = = SIMO DFE/LE = 3MISO DFE/LE o = 4Zero-Forcg MIMO/MISO/SIMO SISO DFE/LE σ = Referece:.Al-Dar ad A.H.Saed Te Fte Legt Multple Iput Mult-Output MMSE-DFE, IEEE Tras o Sgal Processg, OCT