Adaptive Demodulation Techniques for Next Generation Software - PowerPoint PPT Presentation

Adaptive Demodulation Techniques for Next Generation Software Defined Radios U.S. Army RDECOM Communication-Electronics RD&E Center Fort Monmouth, NJ 07703, USA

Contents � Introduction � Modulation classification overview � Research on commercial applications � Challenges

Modulation Classifier From: http://www.ottawa.drdc-rddc.gc.ca

What is Automatic Modulation Classification ? 1 2 3 4 5 6 7 8 9 0 Demodulated signal signal Demodulation IF Demodulation Demodulation Filter A/D Automatic Classification LO Modulation Channel Filter Equalizer Recognition A non-cooperative communication technique which uses ………… BW ………… … statistical methods to ………… Center Estimation … Channel SNR … Frequency estimate the signal Estimation Symbol Rate Estimation Phase Estimation modulation types Estimation Estimation Statistical Estimation

Modulation Classification Preprocessed IF A non-cooperative SNR Estimation communication technique Templates Building Coarse Modulation Estimation which uses statistical Analog Digital methods to estimate the PSK/QAM Analog FSK/MSK modulation type of a Preprocessing Preprocessing Preprocessing unknown signal PSK/QAM Analog FSK/MSK Feature Feature Feature Extraction Extraction Extraction Analog PSK/QAM FSK/MSK Classification Modulation Modulation Modulation Decision Estimation Estimation Estimation Modulation Scheme Estimation Confidence Unknown Classification Confidence Type Modulation Parameters Rating Failure

SDR Applications (1) � Overcome channel fading � Monitor communication spectrum � Remove co-channel interferences 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 1 2 3 4 5 6 7 8 9 1 2 3 0 4 5 6 7 8 9 7 8 9 0 0

SDR Applications (2) Deep Space Communication Reduce the scheduling and configuration 1 2 3 4 5 6 7 8 9 0 burdens of communications 1 2 3 4 5 6 7 8 9 0 2 3 1 5 6 4 8 9 7 0

Modulation Classification Overview

Feature Extraction: Amplitude, Differential Phase, and Frequency •Input: IF •Feather: Amplitude, phase, diff phase, frequency •Statistics: histogram, STD •Classifier: max correlation, decision tree •Reference: Liedtke 1984 baud rate BPF timing circuit templates detector Freq . Δ φ STD IF I Δ 2 t recognition tree CW Δ φ Delta tan -1 / correlation histogram PSK2 Δ phase T PSK4 Q PSK8 ( ) 2 ASK2 FSK2 ( ) 2 sqrt + Amplitude

Higher-order Transform of Constellations c40 c80 c20 c21 c 41 c 42 c60 c61 c62 c 63 c81 c82 c83 c84 QPSK PSK8 PSK16 QAM16 V29-8 and V29-16 constellations QAM64 QAM32 QAM4-12 QAM16-16 QAM44-20 The 4th order constellations

M r k Higher-order Statistical 2 nd order Features Power-law: 4 k ( ) r 1 st order Moment: K ∑ 4 th order = 4 ( ) m r k 40 4 th order transformation of QPSK = ⎧ ⎫ k 1 dominant M K 1 ∏ ∑ i = ( i ) ⎨ ⎬ ( | ) ( ) G H r p r i K j k points ⎩ ⎭ M 2 nd order Cumulant: = = 1 1 k j i = − 2 3 m C m 40 40 20 Q I 4 th order transformation of QAM16

Cumulants vs. SNRs

Cyclic Spectral Analysis Time varying autocorrelation { } + τ = + τ * ( , ) ( ) ( ) R xx t t E x t x t Baseband * Cyclic autocorrelation 1 / 2 ∫ T τ = + τ − π 2 a j at ( ) lim ( , ) R R t t e dt * * xx − xx → ∞ T / 2 T T Spectrum correlation density ∞ ∫ π d − τ = τ τ 2 a a j f ( ) ( ) S f R e * * xx − ∞ xx •Input: IF Cycle Freq •Features: cycle frequencies •Reference: Menguc, 2004 Decision Templates Theoretical spectrum correlation magnitude Gardner and Spooner 1992

Feature Classification: Maximum Likelihood (ALRT) Templates K ∏ ( | ( )) l H r k PDF BPSK (.) BPSK = k 1 Modulation Baseband K ( | ( )) ∏ l H r k PDF QPSK MAX (.) QPSK Scheme = k 1 K ∏ ( | ( )) l H r k PDF 8PSK (.) 8 PSK = k 1 BPSK • Input: baseband • Feature: Complex envelop • Classifier: maximum likelihood QPSK • References: Polydoros and Kim 1995 unknown 8PSK

Feature Classification: Histogram Correlation M 1 ∑ i ( i ) p ( r ) ⎧ ⎫ M K 1 ∏ ∑ j k i M = ⎨ ( i ) ⎬ = G ( H | r ) p ( r ) j 1 i i K j k ⎩ M ⎭ = = 1 1 k j i ⎧ ⎫ 2 − ( i ) ( ) ⎪ r b j ⎪ 1 = − k ( i ) ⎨ ⎬ p ( r ) exp ALRT πσ σ j k 2 2 2 2 ⎪ ⎪ ⎩ ⎭ M 1 ∑ i ( i ) p ( r ) j k M = j 1 i Quantize ( ) i ( ) p r j k ( i ) p q K Q Q ( i ) ∑ ∑ ∑ ∑ = = ← p ( ) ( ) ( ) ( ) i i i i ( | ) log ( ) log ( ) log L H r p r p r p D q i K k k q q HIST = = ∈ Ω = 1 1 1 k q k k q • Input: baseband/IF ( i ) D • Feature: frequency/diff phase q • Classifier: max correlation • References: Liedtke 1984

Research on Commercial Applications

Research on Adaptive Modulation Based on SDR - Cooperative � Maintain a constent BER by varying modulation schemes � Modulation schemes: QPSK, 16QAM, and 64QAM � Data frame based modulation recognition � A pilot symbol is used in forward channel � Reference: Jain, P.; Buehrer, R.M, “Implementation of adaptive modulation on the Sunrise software radio,” The proceedings of the 45th Midwest Symposium on Circuits and Systems, Volume: 3 , 4-7 Aug 2002. Pages:III-405 - III-408 Slow Flat Pilot Pilot Fading Receiver Transmitter DATA DATA Channels Pilot Data Feed Back Channel

Why Applying Non-cooperative Demodulation � Environment limitation and restriction � Elimination of the signal overhead information � Attractive for packet data services IF Choose Data Air-interface Delayer Demod RF Preprocessing Demodulator output Modulation Modulation Recognition Recognition Preprocessing

Deference Between Military and Commercial Applications . . SIGINT SDR . Real time classification demodulation SNR low high Candidates unlimited limited QoS friend / foe packet loss Pulse shape unknown known Bandwidth unknown known Baud rate unknown known Blindness more less

Research of Nolan et al. Reduce Form Constellation Q Assume: � Equally likely I � Symmetrical QAM/PSK Purpose: � Reduce the processing time Constellation for QPSK Issues: Q � Low utilization of available information I � May need longer data length for randomness Reduced form constellation

Adaptive Receiver – Ishii et al. � Automatically recognize BPSK, QPSK, 8PSK, pi/4QPSK, 16QAM, FSK, MSK, GMSK, AM, FM, CW, and SSB using decision tree for spectrum, variance, and baud detection analysis. Thresholds Transmitter Decision tree •Spectrum •Envelope IF Modulation RF BPF •Baud Estimation •Amplitude •Phase Data Demodulation OSC output

Blind Modulation Estimation Umebayshi et al. � Automatically recognize BPSK, QPSK, 8PSK, and 16QAM using amplitude and differential phase variances. Channel gain estimation is discussed (2000). � Automatically recognize BPSK, QPSK, and 8PSK using differential phase and maximum likelihood test. Transmitter IF Data BPF Demodulation RF out Noise Variance Maximum Likelihood OSC Modulation Estimation Estimation

Research of Menguc and Jondral (1) Air Interface Identification for SDR Identify (Verify?) � TDMA-GMSK � OFDM-PSK/QAM � CDMA-QPSK Transmitter Data Demodulation output •Determine Number of Interfaces Recognize Preprocessing RF Air Interfaces •Estimate Base band Carrier and Calculate BW Likelihood Threshold Cyclic Test Autocorrelation

Research of Menguc and Jondral (2) Magnitude Plot of the Cyclic Autocorrelation Estimations Issues � Processing speed � Need universal front end GMSK OFDM CDMA * O. Menguc, “Air interface identification for software radio systems,” Ph.D. Dissertation, University of Fridericiana Karlsruhe, Nov. 30, 2004.

Research of Simon and Divsalar (1) Data Format classification for SDR •Discriminate NRZ and Use two curves approximate ln cosh(x) in order to simplify Manchester code the ML computation •Reduce complexity small •Extend to non-coherent case ⎛ ⎞ ⎛ ⎞ − − K 1 K 1 2 2 2 2 P P ∑ ∑ b b ⎜ ⎟ ⎜ ⎟ < ln cosh ln cosh r r ⎜ ⎟ ⎜ ⎟ , , k NRZ k Manchester ⎝ N ⎠ ⎝ N ⎠ = = 0 0 n n 0 0 2 / 2 ; small x x ≅ ln cosh( ) { x − | | ln 2 ; large x x Ln cosh(x) vs x

Reduced Complexity ML Implementation Research of Simon and Divsalar (2) Estimation SNR Baseband Data