Laura Melian Gutierrez - Cognitive Radio in HF Communications

Transcript

1. Cogni7ve(Radio(in(HF(Communica7ons( Laura(Melián(Gu7érrez( lmelian@ide7c.eu( IDeTIC( Universidad(de(Las(Palmas(de(Gran(Canaria,(Spain( Rennes,(6th(November(2014(
2. Outline( •  IDeTIC:(A(li/le(bit(about(us.(( •  Communica7ons(in(the(HF(band( •  Cogni7ve(Radio(in(HF(communica7ons( – Using(wideband(HF(receivers(for(spectrum(sensing( – Learning(with(Hidden(Markov(Models( – Decision(making(for(dynamic(spectrum(access( •  Conclusion( 1(
3. IDeTIC( •  Research(Ins7tute( within(ULPGC( Research(ins7tute(within( Universidad(de(Las(Palmas( de(Gran(Canaria( 2( www.ide7c.eu( Around 65 people !  Staff((PhD):(((( ( ( ( (29( !  Staff((no(PhD): (((((( ( ( (((8( !  Hired((no(permanent(staff):( (16( !  MSc(students: ( ( ( (((6( !  BSc(students:( ( ( ( (((3( … and 12 external collaborators
4. IDeTIC:(Academic(ac7vi7es( 3( Expert'course'on' aeronau.cal'and'airport' management' PhD'program'on'' Communica.ons'Systems' Short'course'on' “the'port'city”' Master'on'home' automa.on,' architecture'and' sustainable'for' tourism' Expert'course'on' IT'and'tourism' Master'on'IT'solu.ons' for'the'environment' and'welfare'
5. IDeTIC:(Research(fields( 4( RF'Systems'&'Radar' Sensor'networking,' SmartSpaces'&'IoT' Automa.c'geoloca.on' of'wildfires' Mari.me' communica.ons'&' surveillance' Wireless'Photonics' &'InKhome' Services' IT'applied'to'Social' Sciences' Space'&'AircraL' Electronics' Biometrics'&' Signal'Processing'
6. 66% 91% 57% 67% 87% 50% 77% 66% 91% 57% 67% 87% 50% 77% HF(Communica7ons( •  Research(line(between(IDeTIC((ULPGC)( and(GAPS((UPM).( •  LongWdistance(HF(communica7ons( •  HFDVL(System((HF(Data+Voice(Link):( –  Based( on( SoZware( Radio( with( mul7carrier(modula7ons.( –  Digital(interac7ve(voice(and(highW data(rate(transmissions.(( –  Operates( with( commercial( transceivers.( –  SISO( &( SIMO( (up( to( 4( Rx)( configura7ons.( •  Three(configura7ons(in(data( transmission:( –  File(transfer( –  Short(message((SMS)( –  HFMail( 5( [EXAMPLE , ]
7. HFDVL:(Real(Tests( 6( Tests done by the Spanish Department of Defense Arnomendi ship (60 days) Flight from Manas to Zaragoza (14 h) Hercules C-130 airplane Canary( Islands 6021 Km 7950'Km 3864'Km
8. HFSA_IDeTIC_F1_V01(Database( Spectrum(ac7vity(in(the(14(MHz(amateur(band.( •  Power(measurements(in(the(frequency(domain.( •  600(kHz(bandwidth((200(channels(simultaneously):( amateur(band(and(other(sta7ons.( •  Dura7on:(10(minutes.( •  Weekdays(&(Weekends.( •  Each(sample(represents(a(3kHz(channel(in(2(seconds.( 7( Yagi antenna Broad-band HF transceiver Spectrum power measurement Agilent Vector Signal Analyzer PC with: System Vue and VSA software
9. HFSA_IDeTIC_F1_V01(Database( Spectrum(ac7vity(in(the(14(MHz(amateur(band.( 8(
10. Cogni7ve(Radio(in(the(HF(band( •  Challenges(to(face(in(this(environment( – There(is(no(coordina7on(among(users.( •  Frequency(alloca7on(per(country.( •  TransWhorizon(behaviour.( – Use(of(wideband(receivers:( •  The(dynamic(range(of(the(received(power(is(wider(than( cellular(environments.( •  Strongly(affected(by(narrowband(interference.( 9(
11. Outline( 10( Learning/( Knowledge(extrac7on( Spectrum( sensing( Wireless( receiver( Decision( making( Wireless( transmi/er( Frequency' spectrum' Transmit) Observe)
12. NBI(in(wideband(HF(receivers( 11(
13. NBI(in(wideband(HF(receivers( 12(
14. NBI(in(wideband(HF(receivers( Wideband(HF(transceiver(–(Receiver(diagram(block( 13( 
     ADC BW = 1.0 MHz 112 MHz 122.7 MHz 82-109 MHz BW 1 MHz 10.7 MHz Image-rejection Filter 3-30MHz Band Pass Filters 3-30MHz RF Amp ≤ 20 dB DDS Low Pass 10.7 MHz AGC ≤ 75 dB RS232   µController CARDS12 
     ADC BW = 1.0 MHz 112 MHz BW = 1.0 MHz 112 MHz 122.7 MHz 82-109 MHz BW 1 MHz 10.7 MHz BW 1 MHz 10.7 MHz Image-rejection Filter 3-30MHz Image-rejection Filter 3-30MHz Band Pass Filters 3-30MHz Band Pass Filters 3-30MHz RF Amp ≤ 20 dB ≤ 20 dB DDS Low Pass 10.7 MHz AGC ≤ 75 dB RS232   µController CARDS12 We(must(detect(and(mi7gate(in(the(analog(domain(
15. NBI(detec7on(in(wideband(HF(receivers( 14( Our(proposal(based(on(Compressive(Sensing(for(NBI(detec7on( ( ( ( ((A(parallel(system(to(the(wideband(receiver(with:( (W(Detec7on(phase( (W(Mi7ga7on(block( ADC(with(fs (<<(fNyq(
16. 15( NBI(detec7on(in(wideband(HF(receivers( Linear(Measurement( Process( Φ( M x N transforma7on( (M < N)( x =(Original(Signal( (N(samples)( y =(Compressive(Measurements( (M(measurements)( y = Φ x " Basics(of(Compressive(Sensing:(
17. 16( NBI(detec7on(in(wideband(HF(receivers( y = Φ x " Basics(of(Compressive(Sensing:( x = Ψ s " Sparse(Signal( Fourier(basis( Compressible(Signal(
18. 17( NBI(detec7on(in(wideband(HF(receivers( y =(Compressive(Measurements ( ( (K =(Sparsity(level) ( Φ =(Measurement(Matrix " " " "s =(Sparse(Signal( " x =(Original(Signal( ( ( ( ( ( (Ψ =(Sparsity(Basis( "
19. NBI(detec7on(in(wideband(HF(receivers( 18( Our(proposal(based(on(Compressive(Sensing(for(NBI(detec7on( ( ( ( ( Unpublished(Results(
20. Outline( 19( Learning/( Knowledge(extrac7on( Spectrum( sensing( Wireless( receiver( Decision( making( Wireless( transmi/er( Frequency' spectrum' Transmit) Observe)
21. Hidden(Markov(Model( 20( Hidden'Markov'Model' Doubly( embedded( stochas7c( process( with( an( underlying( stochas7c(process(that(is(not(observable((it(is(hidden),(but(it(can( only( be( observed( through( another( set( of( stochas7c( processes( that(produce(the(sequence(of(observa7ons.( ... Box 1 Box 2 Box N
22. Hidden(Markov(Model( •  The(three(basic(problems(of(HMM( – The(evalua7on(problem:( (( (Forward(step(of(ForwardWBackward(procedure( – The(decoding(problem:(( (Viterbi(algorithm( – The(learning(problem:(( (BaumWWelch(method( 21(
23. Data(segmenta7on( 22( 2 s. 1 minute 1 0 0 ... ... ... ... 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 3 1 2 3 ... 1 10 minutes
24. HF(Spectrum(Ac7vity(Predic7on( •  Main(model:( (Ergodic(HMM( (10(minutes(sequences( •  3(submodels:( (LeZWright(HMM( (Observa7on(symbols(for(1( (minute( 23( HMM1 HMM2 HMM3 1 2 3 4 39 40 ... 38
25. HF(Spectrum(Ac7vity(Predic7on( Submodel 1 λ 1 = (A 1 ,B 1 ,π 1 ) Submodel 2 λ 2 = (A 2 ,B 2 ,π 2 ) Submodel 3 λ 3 = (A 3 ,B 3 ,π 3 ) P(O T |λ 2 ) P(O T |λ 3 ) P(O T |λ 1 ) B’ 11 B’ 22 B’ 33 High-level model New definition of the high-level model: Previous detected states High-level model λ = (A,B’,π) max(P(OT+1 |λ)) Predicted state B’ 11 0 0 0 B’ 22 0 0 0 B’ 33 B’ = O T Observations t (min) T T-1 Submodels t (min) T-1 T T-2 ... ST 24(
26. HF(Spectrum(Ac7vity(Predic7on( 1 2 3 4 5 6 7 8 4 6 8 10 12 14 16 Acquired knowledge (min.) Average error rate (%) Global performance Normal activity High activity 25(
27. Outline( 26( Learning/( Knowledge(extrac7on( Spectrum( sensing( Wireless( receiver( Decision( making( Wireless( transmi/er( Frequency' spectrum' Transmit) Observe) Joint(work(at(Supélec(
28. Decision(making(with(UCB( Upper(Confidence(Bound((UCB)(algorithm(to(provide(HF( secondary(users(with(dynamic(access(to(the(spectrum.( 27( Joint(work(at(Supélec( Exploita7on( Explora7on(
29. Decision(making(with(UCB( 28( Joint(work(at(Supélec( Unpublished(Results(
30. Conclusion( The(applica7on(of(Cogni7ve(Radio(to(HF( communica7ons(might(be(feasible( ( •  Both(Learning(with(HMM(and(decision(making(with( UCB(can(help(secondary(HF(users(to(avoid(collisions( with(other(users.( •  A(compressive(detector(can(be(used(in(wideband(HF( receivers(to(detect(NBI.( 29(
31. Thank(you(for(your(a/en7on!(
32. Cogni7ve(Radio(in(HF(Communica7ons( Laura(Melián(Gu7érrez( lmelian@ide7c.eu( IDeTIC( Universidad(de(Las(Palmas(de(Gran(Canaria,(Spain( Rennes,(6th(November(2014(