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(
   

   View Slide

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(
   

   View Slide

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
   

   View Slide

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'
   

   View Slide

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'
   

   View Slide

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 , ]
   

   View Slide

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
   

   View Slide

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
   

   View Slide

9. HFSA_IDeTIC_F1_V01(Database(
   Spectrum(ac7vity(in(the(14(MHz(amateur(band.(
   8(
   

   View Slide

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(
    

    View Slide

11. Outline(
    10(
    Learning/(
    Knowledge(extrac7on(
    Spectrum(
    sensing(
    Wireless(
    receiver(
    Decision(
    making(
    Wireless(
    transmi/er(
    Frequency'
    spectrum'
    Transmit) Observe)
    

    View Slide

12. NBI(in(wideband(HF(receivers(
    11(
    

    View Slide

13. NBI(in(wideband(HF(receivers(
    12(
    

    View Slide

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(
    

    View Slide

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
    (<

    View Slide

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:(
    

    View Slide

17. 16(
    NBI(detec7on(in(wideband(HF(receivers(
    y = Φ x "
    Basics(of(Compressive(Sensing:(
    x = Ψ s "
    Sparse(Signal(
    Fourier(basis(
    Compressible(Signal(
    

    View Slide

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(
    "
    

    View Slide

19. NBI(detec7on(in(wideband(HF(receivers(
    18(
    Our(proposal(based(on(Compressive(Sensing(for(NBI(detec7on(
    (
    (
    (
    (
    Unpublished(Results(
    

    View Slide

20. Outline(
    19(
    Learning/(
    Knowledge(extrac7on(
    Spectrum(
    sensing(
    Wireless(
    receiver(
    Decision(
    making(
    Wireless(
    transmi/er(
    Frequency'
    spectrum'
    Transmit) Observe)
    

    View Slide

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
    

    View Slide

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(
    

    View Slide

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
    

    View Slide

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
    

    View Slide

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(
    

    View Slide

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(
    

    View Slide

27. Outline(
    26(
    Learning/(
    Knowledge(extrac7on(
    Spectrum(
    sensing(
    Wireless(
    receiver(
    Decision(
    making(
    Wireless(
    transmi/er(
    Frequency'
    spectrum'
    Transmit) Observe)
    Joint(work(at(Supélec(
    

    View Slide

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(
    

    View Slide

29. Decision(making(with(UCB(
    28(
    Joint(work(at(Supélec(
    Unpublished(Results(
    

    View Slide

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(
    

    View Slide

31. Thank(you(for(your(a/en7on!(
    

    View Slide

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(
    

    View Slide