impulse radio ultra-wideband antenna array correlation beamformingimpulse radio ultra-wideband...

Upload: ragaconi

Post on 07-Jul-2018

256 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    1/27

    Impulse Radio Ultra-Wideband Antenna ArrayCorrelation Beamforming

    Igor Dotlić, Kamya Yekeh Yazdandoost, Huan-Bang Li and Ryu Miura

    National Institute for Information and Communications Technology, Japan

    2016 International Conference on Electronics, Information andCommunication

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 1 / 27

    http://find/http://goback/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    2/27

    Outline1 Introduction

    2 PreliminariesWideband antenna array time-domain modelCorrelation beamformer

    Received signal down-conversionSignal correlationBeamformer output

    Beamformer gainBeamformer gain denitionMaximum attainable beamformer gain

    3 Gain pattern synthesisGeneral principlesProposed method

    Overall optimization problemGain reduction constraint

    4 Numerical examplesSimulation setup

    Pulse p (t )Antenna array element

    Dotlić, Yazdan doost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 2 / 27

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    3/27

    Introduction

    1 IntroductionWideband antenna array time-domain modelCorrelation beamformerBeamformer gainGeneral principlesProposed methodSimulation setupBeamformer pattern shaping examples

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 3 / 27

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    4/27

    Introduction

    Introduction of Impulse Radio Ultra-Wideband (IR-UWB)

    UWB RegulationsIn 2002 US FCC published its sub-part F of Part 15 regulations

    Receiver architecturesEnergy Detection (ED) receivers

    Low complexityFirst to appearHigh sensitivity to multiple access interference.

    Commercially available coherent IR-UWB

    Appeared considerably laterHigh accuracy two-way indoor positioning and radar

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 4 / 27

    http://find/http://goback/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    5/27

    Introduction

    Rationale for beamforming in IR-UWB

    Low regulated EIRP spectral density of only -41.3dBm/ MHz.The most limiting factor for range and data rate in the IR-UWBsystems.

    Range and/or data rate may be increased by employing antennaarrays.Phase shifters can be used for beamforming in small UWB arrays.For larger arrays more complex signal processing techniques arenecessary

    UWB beamforming in the digital domain ⇒ high sampling rates.Analog delay lines ⇒ large space required

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 5 / 27

    http://find/http://goback/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    6/27

    Introduction

    Correlation beamforming paradigm

    Radar ⇒ single range bin is of interest (range gating).Radio receiver

    ⇒ known TOA and DOA.

    Array pattern shaping in correlation beamformers is little investigated.

    Topic of this work.Arbitrary array geometry.Arbitrary elements’ patterns.

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 6 / 27

    l

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    7/27

    Preliminaries

    2 PreliminariesWideband antenna array time-domain modelCorrelation beamformerBeamformer gainGeneral principlesProposed methodSimulation setupBeamformer pattern shaping examples

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 7 / 27

    P li i i Wid b d i d i d l

    http://goforward/http://find/http://goback/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    8/27

    Preliminaries Wideband antenna array time-domain model

    Wideband antenna array time-domain model

    Antenna array comprising M elements with indexesk ∈ {0, 1, . . . , M −1}.Position vector of the k -th element r k .

    Time-domain radiation pattern is denoted g k ( u , t ).Wideband pulse p (t ) incident on the array from direction u .

    Signal produced at the output of the k -th array element:

    s k ( u , t ) = p (t )⊗g k ( u , t ). (1)

    ⊗ denotes convolution.

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 8 / 27

    Preliminaries Correlation beamformer

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    9/27

    Preliminaries Correlation beamformer

    Received signal down-conversion

    × LPF

    π / 2

    cos(2π f 0t ) s̃ k ( u , t ) × (· ) d t z k ( u )

    × LPF ζ ∗k (t )

    s k ( u , t )

    s̃ Q k ( u , t )

    s̃ I k ( u , t )

    To other array elements.

    Figure: A single element of correlation beamformer.

    s̃ k ( u , t ) = p̃ (t ) ⊗̃g k ( u , t ), (2)

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 9 / 27

    Preliminaries Correlation beamformer

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    10/27

    Preliminaries Correlation beamformer

    Signal correlation

    The correlation template has the form:

    ζ k (t ) = p̃ (t −τ maxk ). (3)The correlator output z k ( u ) is calculated as

    z k ( u ) =+ ∞

    −∞

    s̃ k ( u , t )ζ ∗

    k (t ) dt . (4)

    τ maxk = argmaxτ

    + ∞

    −∞

    s̃ k ( u max , t )p̃ ∗ (t −τ ) dt (5)

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 10 / 27

    Preliminaries Correlation beamformer

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    11/27

    Preliminaries Correlation beamformer

    Beamformer output

    Calculated as a weighted sum of correlators’ outputs:

    v ( u ) = w H z ( u ). (6)

    z ( u ) = [ z k ( u )]M − 1

    k =0 – the column vector of the correlator outputs.Superscript “H ” denotes Hermitian transposition.

    w – column vector of complex weights.

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 11 / 27

    Preliminaries Beamformer gain

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    12/27

    Preliminaries Beamformer gain

    Beamformer gain denition

    Dened as signal-to-noise ration (SNR) gain relative to the SNR at theoutput of the reference isotropic receiver.

    G ( u ) =w H z ( u ) 2E 2p w 2

    , (7)

    where E p =+ ∞

    −∞

    |p̃ (t )

    |2 d t is the energy of the pulse p̃ (t ).

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 12 / 27

    Preliminaries Beamformer gain

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    13/27

    Preliminaries Beamformer gain

    Maximum attainable beamformer gain

    The weighting distribution that maximizes the gain for some direction u

    w opt ( u ) = C w z ( u ), (8)

    where C w is an arbitrary complex constant. Inserting (8) in (7) yields

    G max ( u ) = z ( u ) 2E 2p

    . (9)

    w opt ( u max ) and G max ( u max ) are of particular importance.

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 13 / 27

    Gain pattern synthesis

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    14/27

    p y

    Wideband antenna array time-domain model

    Correlation beamformerBeamformer gain

    3 Gain pattern synthesisGeneral principlesProposed methodSimulation setupBeamformer pattern shaping examples

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 14 / 27

    Gain pattern synthesis General principles

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    15/27

    y

    General principles

    z ( u ) is analogous to the so-called steering vector of the narrowband

    antenna array.Analogous to the classic narrowband beamforming paradigm.Giving up the requirement for maximum possible gain

    Optimize different parameters of the array pattern.

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 15 / 27

    Gain pattern synthesis Proposed method

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    16/27

    Overall optimization problem

    Second Order Cone Programming (SOCP) method.Mixed-norm minimization of the beamformer gain pattern sidelobes.Additional gain loss constraint.

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 16 / 27

    Gain pattern synthesis Proposed method

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    17/27

    Overall optimization problem (math)

    w o : Minimizew

    βα + γ (1 −α),s.t.: z ( u max )H w = 1 ,

    Z H sl w ≤β,|z ( u n )H w | ≤γ, for n = 0 , 1, . . . , N −1,

    w ≤√ ηmax

    z ( u max ).

    (10)

    u n for n = 0 , 1, . . . , N −1 is the set of directions uniformly distributedin the sidelobes area.N ≥10M 0 ≤α ≤1 is the factor of the α–norm of sidelobes

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 17 / 27

    Gain pattern synthesis Proposed method

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    18/27

    Gain reduction constraint

    Gain reduction:

    η = G max ( u max )

    G ( u max ) = w 2 z ( u max ) 2. (11)

    Including in the problem the SOCP constraint

    w ≤√ ηmax

    z ( u max ), (12)

    assurs that the loss in gain is no more than ηmax .

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 18 / 27

    Numerical examples

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    19/27

    Wideband antenna array time-domain model

    Correlation beamformerBeamformer gainGeneral principlesProposed method

    4 Numerical examplesSimulation setupBeamformer pattern shaping examples

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 19 / 27

    Numerical examples Simulation setup

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    20/27

    Pulse p (t )

    The pulse used in the numerical examples:Linear chirp pulse.

    The parameters specied in the mandatory mode of IR–UWB PHY inthe IEEE 802.15.6-2012 standard for Body Area Networks (BAN).Pulse duration: T p = 64 ns .Chirp frequency sweep: ∆f c = 520 MHz .

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 20 / 27

    Numerical examples Simulation setup

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    21/27

    Antenna array element

    UWB L-Loop antenna was used.Designed for the frequency range of 3.1 GHz–5.1 GHz.The carrier frequency used for p (t ): f 0 = 4 GHz .

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 21 / 27

    Numerical examples Simulation setup

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    22/27

    Array geometry

    Linear array with M = 15 elements.Elements are spaced at d 0 = 0 .4c / f 0, where c is the speed of light.

    The k -th element position vector is r k = ( k −(M −1)/ 2) d 0 i z fork = 0 , 1, . . . , M −1. i z : unit vector along the z axis.

    x and y : axes of the substrate on which the array elements are

    printed.

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 22 / 27

    Numerical examples Beamformer pattern shaping examples

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    23/27

    α parameter effect to the shape of sidelobes

    − 90 − 60 − 30 0 30 60 90− 40

    − 20

    0

    Angle ( ◦ )

    G a

    i n ( d B i )

    w opt (0◦ )

    w o (0◦ ) , α = 0 . 05

    w o (0◦ ) , α = 0 . 5

    w o (0◦ ) , α = 0 . 95

    Figure: Effects of α parameter to level and

    shape of the sidelobes, θmax = 0◦

    and= 1 dB and BW = 14 ◦ .

    Fig. illustrates the physicalmeaning of the α parameter.Low value of α ⇒ sidelobemaximum level reduction.High value of α ⇒sidelobes’ energy reduction.

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 23 / 27

    Numerical examples Beamformer pattern shaping examples

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    24/27

    Effects of η max parameter

    − 90 − 60 − 30 0 30 60 90− 40

    − 20

    0

    Angle ( ◦ )

    G a

    i n ( d

    B i )

    w opt (0◦ )

    w o (0◦ ) , η = 0 . 1 dB

    w o (0◦ ) , η = 0 . 5 dB

    w o (0◦ ) , η = 1 dB

    w o (0◦ ) , η = 3 dB

    Figure: Effects of η parameter, θmax = 0 ◦ ,

    α = 0 .5 and BW = 14◦

    .

    Fig. illustrates the physicalmeaning of the ηmax

    parameter as the maximumgain reduction.

    η attained in theoptimization is practicallyalways equal to ηmax level.Cases with η < ηmax happenwith high levels of ηmax (notshown here).Increasing ηmax reduces thesidelobes level relative to themaximum.

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 24 / 27

    Numerical examples Beamformer pattern shaping examples

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    25/27

    Effects of predened main-lobe beam-width to thesidelobes level

    − 90 − 60 − 30 0 30 60 90− 40

    − 20

    0

    Angle ( ◦ )

    G a

    i n ( d B i )

    w opt (10◦ )

    w o (10◦ ) , BW = 16 ◦ , η = 0 . 1 dB

    wo (10

    ), BW

    = 18◦

    , η = 1 dB

    w o (10◦ ) , BW = 24 ◦ , η = 1 dB

    Figure: Effects of predened main-lobebeam-width to the sidelobes level,

    larger the main lobebeam-width, effective area inwhich the sidelobes need to

    be reduced gets smaller ⇒The optimization has moredegrees of freedom to spendin suppressing sidelobes.With increasing the mainlobe beam width sidelobeslevel is reduced.

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 25 / 27

    Conclusions

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    26/27

    Wideband antenna array time-domain model

    Correlation beamformerBeamformer gainGeneral principlesProposed methodSimulation setup

    Beamformer pattern shaping examples

    5 Conclusions

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 26 / 27

    Conclusions

    http://find/

  • 8/19/2019 Impulse Radio Ultra-Wideband Antenna Array Correlation BeamformingImpulse Radio Ultra-Wideband Antenna Array Correlation Beamforming

    27/27

    Conclusions

    The paper presented the SOCP-based method for shaping the gainpatterns of the IR-UWB correlation beamformers.

    A designer is able to tailor several parameters of the optimization.For example, if larger degree of gain reduction and main lobebeamwidth increase is allowed, then the method is able to reducemixed norm of sidelobes to lower levels.

    Dotlić, Yazdandoost, Li & Miura (NICT) IR-UWB Correlation Beamforming ICEIC 2016 27 / 27

    http://goforward/http://find/http://goback/