CYCLOSTATIONARITY-BASED BLIND CLASSIFICATION OF ANALOG AND DIGITAL MODULATIONS

Similar documents
Consider a Binary antipodal system which produces data of δ (t)

Multicode DS-CDMA With Joint Transmit/Receive Frequency-domain Equalization

Final Exam. Tuesday, December hours, 30 minutes

Molecular Evolution and Phylogeny. Based on: Durbin et al Chapter 8

Low-complexity Algorithms for MIMO Multiplexing Systems

Combinatorial Approach to M/M/1 Queues. Using Hypergeometric Functions

8.1. a) For step response, M input is u ( t) Taking inverse Laplace transform. as α 0. Ideal response, K c. = Kc Mτ D + For ramp response, 8-1

Coupled Mass Transport and Reaction in LPCVD Reactors

Add the transfer payments and subtract the total taxes from (1): Using these definitions, the above becomes: The Saving and Investment Equation:

MEEN 617 Handout #11 MODAL ANALYSIS OF MDOF Systems with VISCOUS DAMPING

Reinforcement learning

Let. x y. denote a bivariate time series with zero mean.

The shortest path between two truths in the real domain passes through the complex domain. J. Hadamard

General Non-Arbitrage Model. I. Partial Differential Equation for Pricing A. Traded Underlying Security

Transmit Beamforming with Reduced Channel State Information in MIMO-OFDM Wireless Systems

QUASI-EXTREMUM SCALAR CONTROL FOR INDUCTION TRACTION MOTORS: RESULTS OF SIMULATION. Rodica Feştilă* Vasile Tulbure** Eva Dulf * Clement Feştilă*

Surface Contents Author Index

L. Yaroslavsky. Image data fusion. Processing system. Input scene. Output images


EECE 301 Signals & Systems Prof. Mark Fowler

7 Wave Equation in Higher Dimensions

Support Vector Machines

Exponential and Logarithmic Equations and Properties of Logarithms. Properties. Properties. log. Exponential. Logarithmic.

Lecture 17: Kinetics of Phase Growth in a Two-component System:

Representing Knowledge. CS 188: Artificial Intelligence Fall Properties of BNs. Independence? Reachability (the Bayes Ball) Example

Today - Lecture 13. Today s lecture continue with rotations, torque, Note that chapters 11, 12, 13 all involve rotations

Introduction to SLE Lecture Notes

Lecture-V Stochastic Processes and the Basic Term-Structure Equation 1 Stochastic Processes Any variable whose value changes over time in an uncertain

Lecture 18: Kinetics of Phase Growth in a Two-component System: general kinetics analysis based on the dilute-solution approximation

8.5 Circles and Lengths of Segments

UT Austin, ECE Department VLSI Design 5. CMOS Gate Characteristics

Part I. Labor- Leisure Decision (15 pts)

Sections 3.1 and 3.4 Exponential Functions (Growth and Decay)

Problem Set If all directed edges in a network have distinct capacities, then there is a unique maximum flow.

Laplace Transform. Inverse Laplace Transform. e st f(t)dt. (2)

Inference for A One Way Factorial Experiment. By Ed Stanek and Elaine Puleo

Linear Quadratic Regulator (LQR) - State Feedback Design

Degree of Approximation of a Class of Function by (C, 1) (E, q) Means of Fourier Series

4/3/2017. PHY 712 Electrodynamics 9-9:50 AM MWF Olin 103

MATHEMATICAL FOUNDATIONS FOR APPROXIMATING PARTICLE BEHAVIOUR AT RADIUS OF THE PLANCK LENGTH

ENGI 4430 Advanced Calculus for Engineering Faculty of Engineering and Applied Science Problem Set 9 Solutions [Theorems of Gauss and Stokes]

Sample Final Exam (finals03) Covering Chapters 1-9 of Fundamentals of Signals & Systems

Randomized Perfect Bipartite Matching

On The Estimation of Two Missing Values in Randomized Complete Block Designs

The sudden release of a large amount of energy E into a background fluid of density

Relative Positioning Method of Using Single-frequency Model for Formation Flying Missions

Graphs III - Network Flow

STUDY OF THE STRESS-STRENGTH RELIABILITY AMONG THE PARAMETERS OF GENERALIZED INVERSE WEIBULL DISTRIBUTION

Identification of Fault Locations using Transient State Estimation

Engineering Accreditation. Heat Transfer Basics. Assessment Results II. Assessment Results. Review Definitions. Outline

Fall 2014 Final Exam (250 pts)

, on the power of the transmitter P t fed to it, and on the distance R between the antenna and the observation point as. r r t

Chapter Finite Difference Method for Ordinary Differential Equations

12. Nyquist Sampling, Pulse-Amplitude Modulation, and Time- Division Multiplexing

Lecture 9. Transport Properties in Mesoscopic Systems. Over the last 1-2 decades, various techniques have been developed to synthesize

Sensors, Signals and Noise

The International Diversification Puzzle when Goods Prices are Sticky: It s Really about Exchange-Rate Hedging, not Equity Portfolios

On Control Problem Described by Infinite System of First-Order Differential Equations

Problem Set #1. i z. the complex propagation constant. For the characteristic impedance:

Flow networks. Flow Networks. A flow on a network. Flow networks. The maximum-flow problem. Introduction to Algorithms, Lecture 22 December 5, 2001

#5 Demodulation and Detection Error due to Noise

Vehicle Arrival Models : Headway

Primal and Weakly Primal Sub Semi Modules

Control Systems -- Final Exam (Spring 2006)

Network Flow. Data Structures and Algorithms Andrei Bulatov

Chapter 7: Inverse-Response Systems

The distribution of the interval of the Cox process with shot noise intensity for insurance claims and its moments

Topic 3. Single factor ANOVA: Introduction [ST&D Chapter 7] 3.1. The F distribution [ST&D p. 99]

Revision of Lecture Eight

Lecture 26: Leapers and Creepers

COMPARING MORE THAN TWO POPULATION MEANS: AN ANALYSIS OF VARIANCE

Variance and Covariance Processes

Then the number of elements of S of weight n is exactly the number of compositions of n into k parts.

18.03SC Unit 3 Practice Exam and Solutions

Stochastic dynamics of ion channels in sensory cilia: regulation by feedback

Computer Propagation Analysis Tools

Mahgoub Transform Method for Solving Linear Fractional Differential Equations

Estimation and Confidence Intervals: Additional Topics

A Theoretical Model of a Voltage Controlled Oscillator

Algorithmic Discrete Mathematics 6. Exercise Sheet

The Residual Graph. 11 Augmenting Path Algorithms. Augmenting Path Algorithm. Augmenting Path Algorithm

ADAPTIVE BACKSTEPPING CONTROL FOR WIND TURBINES WITH DOUBLY-FED INDUCTION GENERATOR UNDER UNKNOWN PARAMETERS

EE Control Systems LECTURE 2

Orthotropic Materials

[ ] 0. = (2) = a q dimensional vector of observable instrumental variables that are in the information set m constituents of u

Chapter 7. Interference

International Journal of Mathematical Archive-5(6), 2014, Available online through ISSN

The Production of Well-Being: Conventional Goods, Relational Goods and Status Goods

Control Volume Derivation

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.

The Production of Polarization

The Residual Graph. 12 Augmenting Path Algorithms. Augmenting Path Algorithm. Augmenting Path Algorithm

To become more mathematically correct, Circuit equations are Algebraic Differential equations. from KVL, KCL from the constitutive relationship

Lecture 22 Electromagnetic Waves

and substitute in to the 1 st period budget constraint b. Derive the utility maximization condition (10 pts)

Lecture 5. Chapter 3. Electromagnetic Theory, Photons, and Light

TRAVELING WAVES. Chapter Simple Wave Motion. Waves in which the disturbance is parallel to the direction of propagation are called the

Introduction to Congestion Games

The Non-Truncated Bulk Arrival Queue M x /M/1 with Reneging, Balking, State-Dependent and an Additional Server for Longer Queues

Interpolation and Pulse Shaping

Transcription:

CYCLOSAIONARIY-BASED BLIND CLASSIFICAION OF ANALOG AND DIGIAL MODULAIONS Oavia A. Dobe Ali Abdi 2 Yehekel Ba-Ne 2 Wei Su 3 Fauly of Eng. Applied Siene Memoial Univeiy of Newfoundl S. John NL AB 3X5 Canada 2 CCSPR Dep. of ECE New Jeey Iniue of ehnology Newak NJ 72 USA 3 RDECOM Fo Monmouh NJ 773 USA dobe@eng.mun.a abdi@adm.nji.edu bane@yegal.nji.edu wei.u@u.amy.mil Aba - he poblem of blind analog digial modulaion laifiaion i akled in hi pape. A yloaionaiy-baed laifie i popoed whih doe no euie eimaion of aie phae feueny offe ignal noie powe iming eovey a pepoeing ak. Numeial imulaion ae pefomed o validae heoeial developmen. I. INRODUCION Blind modulaion laifiaion (MC i a majo ak of an inelligen eeive wih boh miliay ommeial appliaion []. A modulaion laifie eenially involve wo ep i.e. ignal pepoeing modulaion eogniion. Pepoeing ak may inlude eimaion of aie phae feueny offe ignal noie powe iming eovey e. Depending on he laifiaion algoihm employed in he eond ep diffeen pepoeing ak an be euied. Claifiaion algoihm whih ely le on pepoeing ae ough. Reeah on MC ha been aied ou fo a lea a deade []-[9]. Many algoihm popoed in he lieaue adde he digial modulaion laifiaion poblem []-[5]. A uvey of uh laifiaion ehniue i peened in []. Nonehele laifiaion of analog digial modulaion i of inee a ome exiing yem ill ue analog ommuniaion ehniue. Algoihm fo he eogniion of analog digial modulaion ae alo popoed in he lieaue [5]-[8]. Howeve he pefomane of mo of hee algoihm i peened unde he aumpion of pefe pepoeing i an ubanially degade wih model mimahe uh a aie phae feueny offe. In hi pape we popoe a novel algoihm whih exploi ignal yloaionaiy fo analog digial modulaion laifiaion. he algoihm doe no euie eimaion of aie phae feueny offe ignal noie powe iming eovey in he pepoeing ep i appliable o he following pool of modulaion: ampliude modulaion (AM double ideb (DSB ingle ideb (SSB M-ay phae hif keying (M-PSK M-ay uadaue ampliude modulaion (M-QAM. he e of he pape i oganized a follow. he ignal model oeponding aiial haaeizaion ae peened in Seion II he popoed yloaionaiy-baed laifiaion algoihm i inodued in Seion III imulaion eul ae given in Seion IV onluion ae dawn in Seion V. empoal yloaionaiy paamee ae defined in Appendix A eul fo he fi- / o eond-ode yloaionaiy of he ignal of inee ae deived in Appendix B. II. SIGNAL MODEL AND CORRESPONDING SAISICAL CHARACERIAION A. Signal Model Le he eeived baeb wavefom ( be he um jθ j2π f ( = e e ( + w( ( whee θ i he aie phae f i he aie feueny offe ( = I ( + jq ( i he anmied ignal wih modulaion i w ( = wi( + jwq( i zeo-mean omplex Gauian noie of powe N wih indexe I Q ing fo in-phae uadaue epeively j =. he in-phae uadaue noie omponen wi ( wq ( ae zeo-mean unoelaed Gauian poee unoelaed wih (. he anmied ignal i given by ( = A( +µ Ax( fo i=am ( = Ax( fo i=dsb ( = A( x( ± jx (( fo i=ssb ( = A k px( k ε fo i=m-psk k M-QAM whee A i he ignal ampliude µ A i he modulaion index x( i he zeo-mean eal-valued b-limied modulaing ignal ( x( i he Hilbe anfom of x( px ( i he anmi pule hape i he ymbol peiod ε epeen a iming eo wih ε< k = k I + jk Q i he ymbol anmied wihin he k h peiod dawn fom he ignal onellaion i i=m-psk M-QAM M i he numbe of poin in he ignal onellaion []. he daa ymbol { k } ae aumed o be zeo-mean independen idenially diibued ( om vaiable wih { i ki } ( { i kq } unoelaed. A he eeive-ide nomalizaion of he ignal i aied ou wih epe o he eeived ignal powe o emove any ale fao fom daa hen ampling i pefomed a a ampling ae f (no aliaing ou []. he nomalized diee-ime ignal i ( k: = ( / S + N wih S a he ignal powe. = kf

B. Cyloaionaiy of Signal of Inee Reul deived in Appendix B fo he fi- /o eond-ode yli empoal umulan funion (CCF of he ignal of inee ae peened in he euel wih emphai on hei appliaion o modulaion laifiaion. he fi-ode/ zeo-onjugae CCF of AM DSB SSB M-PSK M-QAM nomalized diee-ime eeived ignal a yle feueny (CF β ae a follow jθ /2 (AM ( β = Ae ( S + N wih β= f f (2 ( i ( β = any β i = DSB SSB M-PSK M-QAM. (3 One an hu onlude ha he eeived AM ignal exhibi fiode yloaionaiy wheea all ohe ignal of inee do no. hi popey of he AM ignal i exploied fo i eogniion. he eond-ode/ zeo-onjugae CCF of DSB SSB M-PSK M-QAM M 4 BPSK nomalized diee-ime eeived ignal a CF β delay τ ae a follow (DSB 2 2 j 2 f f j θ π τ x (; βτ = A ( S + N e e m ( τ f (4 wih 2 f f β= mx( τ f a he eond-ode/ zeoonjugae momen of xk ( = x ( = kf ( i (; β τ = any β i =SSB M-PSK M-QAM M 4(5 2 j2πγfε (BPSK (; βτ = A ( S + N f e (6 j2θ j2π fτf j2πγk e e px ( k px ( k+τ e k = wih 2 f f β=γ+ γ= f k/ k inege 2 p X ( k = p X ( kf. = hu fo i=dsb BPSK (; βτ a CF β= 2 f f 2 ff + f k/ 2 epeively wheea fo i=ssb M-PSK M-QAM M 4 (; β τ = any β. hi popey i exploied hee o diinguih beween DSB BPSK SSB M-PSK M-QAM M 4. he eond-ode/ one-onjugae CCF of DSB SSB M-PSK M-QAM nomalized diee-ime eeived ignal a CF β delay τ ae a follow 2 j 2 f f π τ x (DSB (; βτ = 2 A ( S + N e m ( τf (7 + w( β; τ wih β= mx( τ f a he eond-ode/ one-onjugae momen of x( k (; w βτ a he of eond-ode/ one-onjugae yli umulan of wk ( = w ( / S+ N a CF β delay τ = kf 2 j2π fτf (SSB (; βτ =2 A ( S + N e (8 ( mx( τf m jm ( x( τf + w( β; τ wih β = m( (. x a he Hilbe anfom of m (. x 2 j2πγfε j2π fτf (; βτ = A ( S + N f e e (9 * j2πγk px ( k px ( k+τ e + w( β; τ k = wih β=γ= f k/ k inege 2 i=m-psk M-QAM M 2 * a he omplex onjugae. hu fo i = DSB SSB (; β τ a β = only wheea fo i=m-psk M-QAM M 2 (; β τ a β= f k/ k inege 2. hi popey i exploied o diinguih beween DSB BPSK a well a beween SSB M-PSK M-QAM M 4. III. PROPOSED CYCLOSAIONARIY-BASED CLASSIFIER he binay deiion ee algoihm popoed fo blind eogniion of analog digial modulaion i depied in Fig.. A eah node of he ee ignal yloaionaiy i exploied o make a deiion on he modulaion foma of he eeived ignal. A. Binay Deiion-ee Claifiaion Algoihm A Node of he ee diiminaion beween AM DSB SSB M-PSK M-QAM M 2 i pefomed by exploiing he peene of a CF in he fi-ode/ zeo-onjugae CCF of he AM ignal. he eogniion poe i blindly pefomed oniing of he following wo ep: Sep I: he magniude of he fi-ode/ zeo-onjugae CCF i eimaed fom he finie lengh eeived daa euene a idae CF β' ove he [ //2 ange. Fo an AM ignal a peak ou in hi magniude a CF β= f f i value deeae wih a eduion in he ignal o noie aio (SNR defined a SNR:= S/ N (ee (2. hu below a eain SNR he value of hi peak beome ompaable wih he aiially inignifian pike whih appea in he eimaed CCF magniude a idae CF β' β due o he finie lengh of daa euene 3 (example ae povided fo illuaion in Seion IV. If a peak i deeed in he eimaed magniude of he fi-ode/ zeo-onjugae CCF Sep II i applied o he idae CF oeponding o hi peak. Ohewie i i deided ha he modulaion foma i no AM. Sep II: he yloaionaiy e developed in [2] i ued o e he idae CF oeponding o he peak deeed in Sep I. Definiion of he empoal yloaionaiy paamee i given in Appendix A. 2 Noe ha he inege k ake value in a denumeable e whih depend on he yloaionaiy ode ignal bwidh [2]. 3 Appaenly wih a finie lengh daa euene neihe he fi-ode eimaed CCF of AM a idae β β ' no of DSB SSB M-PSK M-QAM M 2 a any idae CF β ' ae idenially zeo. Noe ha eul of Appendix B need o be undeood a aympoial value whih ae obained by aveaging ove an infinie-ime ineval.

Deail on hi e ae given in Seion III. B. If he eed idae CF β ' i deided o be indeed a CF β he ignal i idenified a AM; ohewie a belonging o he DSB SSB M-PSK M-QAM ignal la. A Node 2 diiminaion beween DSB BPSK SSB M-PSK M-QAM M 4 i pefomed by exploiing he peene of a CF in he eond-ode/ zeo-onjugae CCF of he DSB BPSK ignal fo zeo delay ( τ=. he laifiaion poe i imila o ha fo Node. A Node 3 Node 4 diiminaion beween DSB BPSK SSB M-PSK M-QAM M 4 i epeively pefomed by exploiing he peene of non-zeo CF in he eond-ode/ one-onjugae CCF of M-PSK M-QAM ignal M 2 fo zeo-delay ( τ=. he laifiaion poe i imila o ha fo Node wih Sep I applied o he ixh-ode/ hee-onjugae yli empoal momen funion (CMF Sep II o he eond-ode/ one-onjugae CCF. he eondode/ one-onjugae CCF magniude i no ued in Sep I a i value a non-zeo CF β=± f / i low even a high SNR peak oeponding o hee CF anno be diinguihed fom he aiially inignifian pike whih appea due o eimaion baed on a finie lengh daa euene. Inead he wo ymmei peak ae idenified fom he ixh-ode/ hee-onjugae CMF magniude 4. he idae CF whih oepond o he highe peak ou of he wo i hen eed in Sep II. I i o be noed ha he laifiaion poe doe no euie knowledge of he aie phae feueny offe ignal noie powe iming eovey. In addiion i i o be noed ha laifiaion beyond Node 4 an be pefomed by applying ohe algoihm popoed in lieaue uh a []-[3] fo linea digial modulaion [7] fo SSB ignal. B. Cyloaionaiy e A menioned in Seion III. A he yloaionaiy e developed in [2] i ued o hek he peene of a CF a eah node l l =...4 of he popoed laifiaion algoihm. he e i fomulaed a a wo hypohei-eing poblem i.e. H : he eed idae CF β ' i no a CF H : he eed idae CF β ' i indeed a CF. he e oni of hee ep i applied hee fo zeo delay 5 a follow. Sep : he nh-ode/ -onjugae CCF 5 i eimaed fom K ample a he idae CF β ' eleed in Sep I fo zeo delay i.e. ˆ ( β'; 5. hen he veo 4 By uing he yli umulan-o-momen fomula along wih (2-(22 (25-(26 one an eaily pove ha he magniude of he ixh-ode/ hee-onjugae CMF a β=± f / i geae han ha of he eondode/ one-onjugae CCF a he ame CF. 5 Fo Node n = = fo Node 2 n = 2 = fo Node 3 4 n = 2 =. ˆ : [Re{ ˆ ( '; } Im{ ˆ ( '; }] ( ( = K β K n n β 5 ( i fomed wih Re{.} Im{.} a he eal imaginay pa epeively. Sep 2: he aii ˆ = K Σ n n n ˆ ˆ 5 ( i ompued fo he eed idae CF β '. Hee - denoe maix anpoe invee epeively ˆ Σ i n an eimae of he maix Re {( Q + Q / 2} Im {( Q Q / 2} Σ = 5 (2 n Im {( Q + Q /2} Re {( Q Q /2} whee ( K ( K Q : = lim Cum[ ˆ ( β'; ˆ ( β'; ] n K ( K ( K* Q = ˆ β ˆ β K : lim Cum[ ( '; ( '; ] wih Cum[.] a he umulan opeao 6. he ovaiane Q Q ae given epeively by 7 8 [2] K = k n K = +ξ ξ= j2π2 β' k j2 πβ' ξ Q lim K Cum[ L ( k; L ( k ; ] e e K * = =- n K k= ξ +ξ j2 π( β' ξ Q lim K Cum[ L ( k; L ( k ; ] e whee L ((* ( ( k n i u i ; : = ( k i he n h-ode/ u = -onjugae lag podu of ( k fo zeo delay-veo wih ( u u =... n a a poible onjugaion o ha he oal numbe of onjugaion i. Sep 3: he aii 5 eimaed a eed idae CF β ' fo zeo delay i ompaed again a hehold Γ ( l ( l l =...4 fo deiion-making. If 5 one deide ha he eed idae CF i indeed a CF fo zeo delay; ohewie i i no delaed a CF. he hehold i e fo a given pobabiliy of ( l ( K ( l fale alam PF = P{ n H} 5 l =...4 by aking ino aoun ha 5 ha an aympoi χ 2 diibuion wih wo degee of feedom unde H [2]. 5 IV. SIMULAION RESULS A. Simulaion Seup Fo he geneaion of analog ignal he modulaing ignal x( i obained by low-pa fileing a euene of zeo-mean Gauian om numbe wih uni vaiane. he analog ignal 6 Fo he umulan definiion ee e.g. [3] Ch.2. 7 Noe ha hee eualiie ae alway valid if n = wheea only fo a zeo-mean poe if n = 2. 8 Fo he ovaiane eimao ee e.g. [2] e. (48. 5

ae aled o ha he ignal powe i. Fo he AM ignal he modulaion index µ A i omly hoen beween. Fo he linea digial modulaion la we imulae BPSK QPSK 8-PSK 6-QAM 64-QAM wih uni vaiane onellaion. he ignal powe i alo e o he pule hape px ( i oo-aied oine wih.25 oll-off fao []. he ignal bwidh i B = 3 KHz. A he eeive-ide a low-pa file i ued o eliminae he ou-of-b noie he eeived ignal i ampled a a ae f = 48 KHz. he obevaion ineval available 4 a he eeive i eond whih yield K = 4.8 ample. he SNR i defined a he ignal powe o he noie powe a he oupu of eeive file. All eeived ignal ae affeed by a phae θ unifomly diibued ove [ π π a aie feueny offe f f =.. In addiion linea digial ignal ae affeed by a iming eo ε wih ε =.8. B. Eimaed CCF Magniude Saii Ued fo Deiion- Making hehold Seing he magniude of he fi-ode/ zeo-onjugae CCF of AM BPSK ignal ae given fo illuaion in Fig. 2 a b a -db -2dB SNR epeively. Fo AM one an noie he peak in ˆ (AM ( β ' a β=β= ' f. f = a well a he eduion of i value wih a deeae in SNR. On he ohe h non-zeo pike due o eimaion baed on a finie lengh daa ( euene ae o be noed in ˆ K ( i ( β ' i=am DSB. Baed on ˆ ( β'; 5 he aii 5 i alulaed a eah node l l =...4 aoding o (. A Kaie window of lengh 6 paamee wa ued o ompue he ovaiane eimae in (2 8. A eah node l he aii i ompaed again a hehold Γ ( l ( (2 l =...4. Hee Γ =Γ =8.42 (3 (4 Γ =Γ =3.86. hee value oepond o a 4 pobabiliy of fale alam of 3 epeively [4] ae e baed on he eimaed value of he aii ued fo deiion-making. C. Claifiaion Pefomane o evaluae he pefomane of he popoed laifiaion algoihm we define he aveage pobabiliy of oe ( l 2 ( li li laifiaion a eah node l l =... 4 a P = 2 P li = ( i i wih P l l a he aveage pobabiliy o hooe he banh l i l i = 2 when indeed he banh l i i he oe hoie. he ( i i pobabiliy l l ( N li mod ( l ( i i li P i fuhe ompued a P / i N = mod wih ( l N i mod a he numbe of poible modulaion fo he l i h banh ( i P l i a he pobabiliy o hooe he l i h banh when he ( li i h ignal i anmied i =... Nmod. Fo example wih ( ( ( l = (Node l = (AM banh N mod = P = P. ( i he P l i i eimaed baed on 3 Mone Calo ial. he aveage pobabiliy of oe laifiaion a eah node ( l of he popoed binay deiion ee algoihm P l =...4 i ploed veu SNR in Fig. 3. A Node ( l = a pobabiliy of oe laifiaion of one i ahieved a SNR a low a -23dB o diiminae AM fom DSB SSB BPSK QPSK 8-PSK 6-QAM 64-QAM. A Node 2 ( l = 2 uh a pefomane i aained a SNR a low a -8dB o diiminae DSB BPSK fom SSB QPSK 8-PSK 6-QAM 64-QAM. A Node 3 4 ( l = 3 4 5dB SNR i euied o diiminae wih pobabiliy one beween DSB BPSK SSB QPSK 8-PSK 6-QAM 64-QAM epeively. Noe ha he aii whih i uenly ued a Node 3 4 an be applied o diinguih beween analog digial modulaion a Node i.e. o diiminae AM DSB SSB fom M-PSK M-QAM. Howeve highe SNR would be euied o ahieve he ame laifiaion pefomane a Node hu he pefomane of he whole laifie would be limied. I i alo o be noed ha eul peviouly peened ae ahieved fo eond obevaion ineval. Appaenly if a lage obevaion ineval i available a he eeive-ide he eimae will be moe auae whih in un will eul in a bee laifiaion pefomane. V. CONCLUSION A yloaionaiy-baed binay deiion ee algoihm ha been popoed fo analog digial modulaion laifiaion. he algoihm povide a good laifiaion pefomane in addiive Gauian noie doe no euie eimaion of aie phae feueny offe ignal noie powe iming eovey in he pepoeing ep. REFERENCES [] O. A. Dobe A. Abdi Y. Ba-Ne W. Su A uvey of auomai modulaion laifiaion ehniue: Claial appoahe new end o be publihed in IEE Po. Commun. 26. [2] O. A. Dobe Y. Ba-Ne W. Su Highe-ode yli umulan fo high ode modulaion laifiaion in Po. IEEE MILCOM 23 pp. 2-7. [3] C. M. Spoone Claifiaion of ohannel ommuniaion ignal uing yli umulan in Po. ASILOMAR 995 pp. 53-536. [4] O. A. Dobe A. Abdi Y. Ba-Ne W. Su Seleion ombining fo modulaion eogniion in fading hannel in Po. IEEE MILCOM 25 pp.-7. [5] E. E. Azzouz A. K. Ni Auomai Modulaion Reogniion of Communiaion Signal. Kluwe 996. [6] Y.. Chan L. G. Gadboi P. Yanouni Idenifiaion of he modulaion ype of a ignal Signal Poe. vol. 6 pp. 49-54 989. [7] O. A. Dobe A. Abdi Y. Ba-Ne W. Su he laifiaion of join analog digial modulaion in Po. IEEE MILCOM 25 pp.-6. [8] Y. O. Al-Jalili Idenifiaion algoihm of uppe ideb lowe ideb SSB ignal Signal Poe. vol. 42 pp. 27-23 995.

[9] W. A. Gadne Cyloaionaiy in Communiaion Signal Poeing. New Yok: IEEE Pe 994. [] A. B. Calon P. B. Cilly J. C. Ruledge Communiaion Syem 4h ed. MGaw Hill 22. [] A. Napoliano Cyli highe-ode aii: inpu/oupu elaion fo diee- oninuou-ime MIMO linea almo-peiodially imevaian yem Signal Poe. vol. 42 pp. 47-66 995. [2] A. V. Dawade G. B. Giannaki Saiial e fo peene of yloaionaiy IEEE an. Signal Poe. vol. 42 pp. 2355-2369 994. [3] C. L. Nikia A. P. Peopulu Highe-Ode Spea Analyi: A Nonlinea Signal Poeing Famewok. Penie Hall 993 [4] M. Abamowiz I. A. Segun Hbook of Mahemaial Funion. New Yok: Dove Publiaion 972. [5] R. C. Cabo A noe on he appliaion of he Hilbe anfom o ime delay eimaion IEEE an. Aou. Speeh Signal Poeing vol. 29 pp. 67-69 98. APPENDIX A: EMPORAL CYCLOSAIONARIY PARAMEERS In he faion-of-ime (FO pobabiliy famewok ignal ae modeled a ime-eie ahe han ealizaion of ohai poee [9] []. hu wih he FO pobabiliy appoah aiial paamee ae defined hough infinie-ime aveage ahe han enemble aveage. hi appoah i ubeuenly ued o peen empoal yloaionaiy paamee. Le ( be a oninuou-ime omplex-valued ime-eie L % (; τ % be he n h-ode/ -onjugae lag podu of ( defined a L n % ((* ( (; i u i : = ( + u= u (3 whee τ % = [% τ = % τ2... % τ n] i he delay-veo. he ime-eie ( i aid o be n h-ode yloaionay fo a given onjugaion onfiguaion ( -onjugae if m% % I L% % e d I /2 j2πα% ( α % ; τ : = lim ( ; τ I I /2 j2πα : =< L% % ( ; e (4 exi i non-zeo fo ome delay veo a denumeable e of eal α% wih α % [9]. E. (4 define he CMF wih α% a CF. I i o be noed ha CMF aie fom a onideaion of he finie-engh addiive ine-wave omponen in he lag podu. he um of all uh omponen i given by he empoal momen funion (MF [9] { α} j 2πα m% ( ; τ % :=E % [ L% ( ; τ % ] = m% ( α% ; τ % e % (5 ( i ( i m ( i n n α κ % % m whee κ % = { α% : m% ( α% ; } { α% E } [.] i he expeaion opeao in he FO pobabiliy famewok whih eplae he uual opeao E[.] in he ohai famewok. hi aually pefom a ine-wave exaion opeaion i.e. i exa all addiive ine-wave omponen exien in i agumen. he n h-ode/ -onjugae empoal umulan funion (CF of ( an be expeed in em of he n h- lowe-ode MF by uing he momen-o-umulan fomula [9] % ( ; : = Cum[ (... ( +τ % ( + ] ( i ( i(* ((* i n ((* i n n n = m ( i {... } % z= z nz z ( (! ( ; (6 whee {... } i a paiion of = { 2... n} wih z z =... a non-empy dijoin ube of o ha hei eunion i i he numbe of he ube in a paiion ( n z i a delay veo whoe omponen ae elemen of { } n u u= wih indie peified by z n z i he numbe of elemen in he ube z fom whih z oepond o onjugae em. Noe ha z= nz = n z= z =. Fuhemoe % ( i (; un ou o be an almo-peiodi funion of ime ha an be wien a [9] % % % % % e πβ (7 j 2 % ( i (; τ = ( i ( ; β κ % β τ % whee % % % I % % e d (8 I /2 j2πβ% ( i (; β τ = lim ( i (; I τ I /2 i he n h-ode/ -onjugae CCF a CF β % κ % = { β% : % ( β% ; }. By uing (5 (6 (7 he n h-ode/ -onjugae CCF of ( a a CF β % an be expeed a [9] % % % ( i (; β τ = ( (! {... } m ( i =β % α% z= z α% % z nz z ( ; (9 whee α % = [ α%... α% ] i a e of CF i a -dimenional one veo. E. (9 i efeed o a he yli momen-oumulan fomula. A imila expeion an be wien o expe he CMF a a funion of CCF. hi i efeed o a he yli umulan-o-momen fomula i given by [9] m% ( ( α i % ; = ( i ( β ; {... } % % % =α z= z % z nz (2 β z whee β % = [ β%... β% ] i a e of CF. Fo he diee-ime ignal ( ( k i = ( obained by = kf ampling he oninuou-ime ignal ( a a ampling ae f he n h-ode/ -onjugae CCF he e of CF ae epeively given a (unde he aumpion of no aliaing [] % f f (2 (; β τ = ( β ; τ κ = { β [ /2;/2 : β = β% / f ( i ( β; τ } (22 whee τ = f wih omponen τ u = uf u =... n. Eimao of he yloaionaiy paamee defined in he FO famewok ae obained by onideing finie-ime aveage onvege aympoially o he ue value whih ae he infinie-ime aveage of he ame uaniie [9].

APPENDIX B: DERIVAIONS OF (2-(9 Subeuenly fi- / o eond-ode empoal yloaionaiy paamee of he ignal of inee ae deived. Wih a zeo-mean puely aionay modulaing ignal x( 9 by uing (3 (4 wih he ignal given in ( aking ino aoun he aiial popeie of he noie one an eaily how ha he fi-ode/ zeo-onjugae CMF of he eeived ignal ( i = AM DSB SSB ae a follow (AM j2πα% m% (AM ( α% :=< ( e jθ j j2 f 2 j Ae f e θ π e A( Ax( e πα % α= % (23 =< + µ = ohewie m% ( i ( α % = any α% i=dsb SSB. (24 A he fi-ode CMF CCF ae idenial (23 (24 hold alo fo he fi-ode/ zeo-onjugae CCF i.e. % ( β % = m% ( α% β % =α% i=am DSB SSB. he n h-ode/ -onjugae CCF of linea digial modulaion i [2]-[3] n n j2πγε % jθ( n2 j2 π f ( u uτu % = (; β % = A e e e n (25 n (* u j2πγ% p ( +τ % e d+ % ( β% ; u= X u w n whee β=γ+ % % ( n2 f (26 wih γ=k % / k inege i he n h-ode/ -onjugae n umulan fo he ignal onellaion i i=m-psk M-QAM ( u i he opional minu ign aoiaed wih he opional onjugaion (* u u =... n. Noe ha % w(; β% n in (25 i non-zeo only fo n 2 =... n β=. Fo value of he n h-ode / -onjugae umulan fo diffeen linea digial modulaion ( n =...8 =... n ee e.g. [2]-[3]. n Odd-ode umulan (n odd eual zeo fo ymmei ignal onellaion [2]. hu wih = he fi-ode/ zeo-onjugae CCF of M-PSK M-QAM ignal beome zeo i.e. % ( β % = any β % i=m-psk M-QAM. Seond-ode yloaionaiy i ubeuenly inveigaed fo DSB SSB M-PSK M-QAM ignal. One an eaily how ha he eond-ode/ zeo-onjugae CMF of he eeived ignal ( i=dsb SSB ae a follow (DSB (DSB j2πα% m% (DSB ( ατ % ;% :=< ( ( +τ % e = 2 j 2θ j2π f Ae e m% x( α % = 2 f (27 = ohewie m% ( ατ % ;% (SSB (SSB j2 :=< ( ( +τ e πα % % = any α%. (28 (SSB 9 Baed on hi aumpion one an eaily how ha < ( x ( = j2πα% j2πα% < xx ( ( + τ % e =< ( xx ((( + τ % e = any α %. he ideniie m% x = m% x( ( m% xx =m% xx( ( ( [] ae ued o deive (28 wih m% x : =< x(( x +τ % m% x : =< x(( x + τ % ( ( ( m% xx : =< x(( x +τ % ( ( m% xx : =< x(( x + τ % ( (. A he fi-ode CMF eual zeo fo i = DSB SSB a well a fo he noie w ( he yli momen-o-umulan fomula yield he eualiy of eond-ode/ zeo-onjugae CCF CFM i.e. % (; β% τ % = m% ( α% ; β % =α% i = DSB SSB. By applying (25 wih n = 2 = = [ ] = fo i=bpsk = fo i=m-psk M-QAM M 4 [2] he eond-ode/ zeo-onjugae CCF of M-PSK M-QAM ignal ae found a 2 j2πγε % j2θ j2π f A e e e j2πγ% % ( βτ %; % = px ( px ( e d i BPSK +τ % = (29 i = M-PSK M-QAM M 4 whee β % =γ+ % 2 f wih γ=k % / k inege. Similaly one an how ha he eond-ode/ one-onjugae CMF of he DSB SSB ignal ae given epeively a (DSB ( DSB * j2πα% m% ( ατ % ;% := ( ( ( + e ( DSB 2 j2π f 2 Ae mx( τ + mw( α% ; τ α % = = % % % % (3 ohewie m% ( ατ % ;% := ( ( ( + e (SSB (SSB * j2πα% (SSB 2 j2π f 2 Ae ( m% x( jm% x( + m% w( α% ; α % = = m ( (3 ohewie * whee m% x : =< x( x ( + τ % m( % ( x i he Hilbe anfom of m% ( x. he ideniy m% xx = mx( τ ( (% % [5] wih * m% xx : =< x( x ( + τ % ( ( i ued in deiving (3. Noe ha m% x = m% x( m% xx = m% xx( ( ( a x( i a ealvalued ignal. By uing he yli momen-o-umulan fomula one an eaily how ha he eond-ode/ one-onjugae CCF CMF of ( i= DSB SSB ae epeively eual. In ohe wod % (; β% τ % = m% ( α% ; β % =α% i= DSB SSB. Fo i=m-psk M-QAM by applying (25 wih n = 2 = = [2] i beome aighfowad ha 2 j2πγε % j2π f % (; βτ % % = A e e (32 * j2πγ% px ( px ( +τ % e d+ % w( β% ; wih β=γ=k % % / k inege. Wih he eul peviouly deived fo oninuou-ime ignal by applying (2 (22 aking ino aoun ignal nomalizaion one an eaily obain (2-(9 fo he nomalized diee-ime ignal.

AM DSB SSB BPSK QPSK 8-PSK 6-QAM 64-QAM DSB SSB BPSK QPSK 8-PSK 6-QAM 64-QAM No ( ( Ye DSB No (2 Ye AM (2 DSB BPSK SSB QPSK 8-PSK 6-QAM 64-QAM No (3 (4 Ye No ( (3 (4 Ye K β' β' BPSK Fig.. Popoed yloaionaiy-baed binay deiion ee laifiaion algoihm. SSB QPSK 8-PSK 6-QAM 64-QAM.35.3 ^(K (AM (β'.3.25.2.5 ^(K (BPSK (β'.25.2.5...5.5 -.5 -.4 -.3 -.2 -...2.3.4.5 Cidae yle feueny β'.2 a -.5 -.4 -.3 -.2 -...2.3.4.5 Cidae yle feueny β'.4 ^(K (AM (β'..8.6.4 ^(K (BPSK (β'.35.3.25.2.5..2 -.5 -.4 -.3 -.2 -...2.3.4.5 Cidae yle feueny β' Fig. 2. he eimaed fi-ode/ zeo-onjugae CCF magniude of AM BPSK ignal veu idae CF wih 4 K = 4.8 ample a SNR=-dB b SNR=-2dB. b.5 -.5 -.4 -.3 -.2 -...2.3.4.5 Cidae yle feueny β'.95 l = l =2 l =3 l =4.9 P ( l.85.8.75.7-3 -25-2 -5 - -5 5 5 2 SNR (db Fig. 3. he aveage pobabiliy of oe laifiaion veu SNR a eah node l l =... 4.