In the Dictionary of the American National StandardsInstitute (ANSI) [6], the bandwidth of a receiver, an amplifier,or a network was defined as the extent of a continuous rangeof frequencies over which the amplitude (gain) does notdiffer from its maximum value by more than a specifiedamount. This is equivalent to signal-processing definitions,where a signal’s spectrum replaces a device’s transferfunction. The ANSI definition is consistent and, therefore,acceptable when considering a narrowband portion of adevice’s spectrum. However, in a general characterizationof device transfer functions, Giri and Tesche have arguedthat the gain of a device’s transfer function is linearlyincreasing for low frequencies, is linearly decreasing forhigh frequencies, and has resonance behavior at intermediatefrequencies [7]. For the impulse-radiating antennas (IRAs)of the HPEM community, radiation takes place at lowfrequencies, where an impulse-radiating antenna’s transferfunction has linearly decreasing phase and linearly increasinggain: that is, the radiated signal is the derivative of theimpulse-radiating antenna’s input signal. Consequently,the frequency range near the maximum gain (the resonantmode of the antenna) is not usable for HPEM applications.In contrast, the gains of antenna transfer functions associatedwith UWB radars have narrower bandwidths that are centeredabout the resonance region, because the goal of a radarsystem is to maximize the radiated energy in a desireddirection to achieve enough signal-to-noise ratio to detectobjects over a two-way propagation path. Consequently,the behavior of the gain of a device’s transfer function canvary significantly according to the device’s intended use.As the ANSI definition does not account for the specificbehavior of a device, the bandwidth definition can identifythe wrong frequency range and, as a result, give a misleadingbandwidth value – which supports the argument for havingdistinct bandwidth definitions for signals and devices.Since the ANSI definition includes neither the phasebehavior, the dispersion, nor the specific application of adevice, it clearly focuses on the transmission/reception ofmore narrowband signals and, consequently, should not beapplied to UWB devices without appropriate modifications.4.1 DeviceAnother bandwidth definition for devices is found inthe IEEE Standard Definitions of Terms for Antennas(IEEE Std 145) [8], where the bandwidth of an antenna wasdefined to be the range of frequencies within which theperformance of the antenna, with respect to somecharacteristics, conforms to a specific standard. Replacingantenna by device leads to a general definition of bandwidth:The bandwidth of a device, component, or system is thecontinuous range of frequencies over which theperformance, with respect to some characteristics, doesnot differ from some chosen behavior by more than aspecified amount.Since this definition does not provide exactmathematical limits, further interpretation and specificationare needed. To this end, a device can be characterized by theamplitude and phase of its transfer function. Conventionally,the limits of bandwidth are determined by deviations of3 dB in amplitude and 1° in phase (if a phase specificationexists). Note that the preceding bandwidth definition isstrongly related to the specific application or mode ofoperation. For example, an antenna is used in the resonantmode for standard communications and radar applications,or in the differentiating sub-resonant frequency range forHPEM applications. Generally, the bandwidths of bothmodes differ significantly. Although a mathematicallyprecise definition of a device’s bandwidth might be desired,it is difficult to achieve, and is seldom useful for practicalapplications.4.2 SignalThe extreme diversity of signals makes it hard toconstruct a general definition of signal bandwidth thatworks for all signals. Actually, it is far easier to taper asignal and its spectrum in such a way that a particulardefinition makes sense. Most tapered (worst-case) signalsare not physically realizable. As applications of interest(radar, communications, interference) typically transmitreal-valued, causal signals with finite energy, the discussionis restricted to this class of signals.A signal, s, with finite energy can be representedeither as a function of time, t, or as a function, Ŝ , offrequency, f. Both representations are related by the Fouriertransform pair:∞j2() ( )ˆ π fts t = ∫ S f e df,(3)−∞∞ˆ − j2πft( ) ( )−∞S f = ∫ s t e dt . (4)The time-domain function s is called the signal or waveform,and [ 0,2 ] is the corresponding frequency-domain representation(spectrum) [6]. Although s()t and Ŝ( f ) can be complexquantities, specified through an amplitude and a phase orthrough real and imaginary parts, the condition that s bereal-valued implies that∗() (),s t = s t(5)where the asterisk denotes complex conjugation. Moreover,applying Equation (5) to Equation (4) yields the relation( ) ˆ∗( )S ˆ − f = S f(6)for the spectrum at negative and positive frequencies.Hence, for the special case of real-valued signals, only half16The<strong>Radio</strong> <strong>Science</strong> <strong>Bulletin</strong> No <strong>313</strong> (<strong>June</strong>, <strong>2005</strong>)
of the spectrum is required to completely specify the signalwaveform. Consequently, the inverse Fourier transformation(Equation (3)) can be rewritten as [9]⎧⎪⎫2() 2 ˆ js t = R⎨ S( f) e df ⎬⎪⎩⎭∞π ft ⎪∫ , (7)0⎪where R {} x denotes the real part of the argument x. Sincethe signals under consideration have finite energy, themagnitude of Ŝ must have a maximum value that isattained for at least one frequency, f M , that is,for every f.( ) ˆ( )Sˆf ≤ S f M(8)4.2.1 General AxiomsBefore discussing various definitions of signalbandwidth B, three axioms associated with the notion ofspectral width are stated because every appropriate measureof the nominal width of the spectrum Ŝ must comply withthis minimal set of axioms.4.2.1.1 Bandwidth is non-negative:{ ( )}B S ˆ f ≥ 0 . (9)4.2.1.2 Bandwidth is independent of height:{ˆ( )} ˆ( ){ }B kS f = B S f . (10)4.2.1.3 Stretching the argument of Ŝ by a constantfactor c stretches the bandwidth by c:{ˆ( )} ˆ( ){ }B S cf = cB S f . (11)Axiom 4.2.1.2 permits normalization of Ŝ by a constant k,so that kSˆ( f ) has unit area without affecting the bandwidth.Axiom 4.2.1.3 is useful in determining the bandwidth of amodulated signal.4.2.2 Root Mean SquareBandwidthA bandwidth definition that is often used in radarsignal theory is the rms bandwidth, B RMS . The Institute ofElectrical and Electronics Engineers (IEEE) has a standard(IEEE Std 686-1997) [10] that used moment theory todefine B RMS as the second moment of the square magnitudeof Ŝ about a designated frequency, typically 0 or thespectral center, where the integration is taken over allfrequencies ( −∞ < f < +∞ ). Unfortunately, this definitionis ambiguous, and has difficulties when used with UWBsignals. In particular, since the first moment about 0 Hz (themean frequency) is 0 Hz for every real-valued waveform,the notion of spectral center is not very meaningful, even for2narrowband radar signals, since Ŝ (the power spectraldensity or energy density) has a maximum value at nonzerofrequencies. To address this shortcoming, Rihaczek [9]defined the moments over the positive spectrum. Althoughthe IEEE standard is not ambiguous for signals with spectraldensities that have a unique absolute maximum at 0 Hz, theauthors followed Rihaczek’s definitions:BRMS=∞∫0( − ) ⎤ ˆmp ( )∫0( )2 2⎡2πf f S f df⎣⎦∞2 , (12)Sˆf dfwhere the denominator of the radicand is half of the signalenergy, the mean frequency, f , over positive frequenciesmpis given byfmp=∞∫0∞∫0( )( )22f Sˆf dfSˆf df, (13)and the rms band is ⎡max{ 0, fmp− 0.5BRMS}⎣ ,fmp+ 0.5BRMS⎤⎦ , because fmp− 0.5BRMScould benegative. The dimensions of B RMS are radians per second.For well-behaved spectra, f mp is usually very near to thefrequency associated with the maximum value of the energydensity, which is usually the carrier frequency for radiatednarrowband signals.4.2.3 X dB Power-Level Bandwidth[5]The X dB power-level bandwidth, B XdB , is given byBXdB = fh − fl,(14)which is implicitly defined by the lowest ( f l ) and highestf ) frequencies that solve( h( ) ˆ( )Sˆf = S f − X (15)20log10 20log10M( f)2ˆXS−⇔ = 10 102 ,Sˆ( f )MThe<strong>Radio</strong> <strong>Science</strong> <strong>Bulletin</strong> No <strong>313</strong> (<strong>June</strong>, <strong>2005</strong>) 17
- Page 1 and 2: Radio Science BulletinISSN 1024-453
- Page 3 and 4: EditorialJune in December?Yes, this
- Page 5 and 6: URSI Accounts 2004In 2004, a year p
- Page 7 and 8: EURO EUROA2) Routine Meetings 7,315
- Page 9 and 10: URSI AWARDS 2005The URSI Board of O
- Page 11 and 12: Guest Editors’ RemarksOn February
- Page 13 and 14: UWB and the corresponding signal pa
- Page 15: Radar / CommunicationsBand TypeElec
- Page 19 and 20: Bandwidths f l / GHz f h / GHz B F
- Page 21 and 22: Figure 3a. A time-domain representa
- Page 23 and 24: Bandwidths f l / f c f h / f c B F
- Page 25 and 26: Bandwidth f l / f c f h / f c B F C
- Page 27 and 28: Quantitative Comparison BetweenMatr
- Page 29 and 30: Next, construct the ‘filtered’
- Page 31 and 32: whereH[ P ] [ U][ ][ V] ,N= Σ (27)
- Page 33 and 34: CPU Time [sec]250200150100MPMSS1SS2
- Page 35 and 36: -10-20MPMSS1SS2-30-40RMSE [dB]-50-6
- Page 37 and 38: RMSE [dB]-45-50-55-60-65-70MPMSS1SS
- Page 39 and 40: An Ultra-Compact Impulse-Radiating
- Page 41 and 42: Figure 3. UCIRA-1 splitter/balun.th
- Page 43 and 44: Figure 9. UCIRA-2 in stowed configu
- Page 45 and 46: activated first in the deployment s
- Page 47 and 48: Figure 20. Theoretical gain ofUCIRA
- Page 49 and 50: Triennial Reports CommissionsCOMMIS
- Page 51: y the Special Section Editors, and
- Page 55 and 56: 3.2 Activities of URSI-Commission C
- Page 57 and 58: It already has been decided that th
- Page 59 and 60: SCOR (Scientific Committee on Ocean
- Page 61 and 62: COMMISSION GThis triennium report w
- Page 63 and 64: GNSS-LEO occultation is a very impo
- Page 65 and 66: CPEA Contacts: Shoichiro Fukao, Pro
- Page 67 and 68:
The group wishes to continue as an
- Page 69 and 70:
total of 109 oral papers (24 thereo
- Page 71 and 72:
surface, to compensate for the rema
- Page 73 and 74:
XXVIIIth General AssemblyNEWLY ELEC
- Page 75 and 76:
Décide1. d’accepter l’invitati
- Page 77 and 78:
satellite observation, bottomside s
- Page 79 and 80:
ETTC ‘05EUROPEAN TEST AND TELEMET
- Page 81 and 82:
IRI 2005 WORKSHOPNEW SATELLITE AND
- Page 83 and 84:
financial and logistics issues conn
- Page 85 and 86:
CONFERENCE ANNOUNCEMENTS36 TH COSPA
- Page 87 and 88:
December 2006APMC 2006 - 2006 Asia-
- Page 89 and 90:
The Journal of Atmospheric and Sola
- Page 91 and 92:
SCIENTIFIC COMMISSIONSCommission A
- Page 93 and 94:
Commission E : Electromagnetic Nois
- Page 95 and 96:
Commission J : Radio AstronomyChair
- Page 97 and 98:
URSI MEMBER COMMITTEESAUSTRALIA Pre
- Page 99 and 100:
ALPHABETICAL INDEX AND CO-ORDINATES
- Page 101 and 102:
BRUSSAARD, Prof. dr.ir. G., Radicom
- Page 103 and 104:
FEICK, Dr. R., Depto. de Electronic
- Page 105 and 106:
SAUDI ARABIA, Tel. +966 1-4883555/4
- Page 107 and 108:
E-mail loulee@nspo.gov.tw (94)LEE,
- Page 109 and 110:
O’DROMA, Dr. M., Dept. of Electri
- Page 111 and 112:
+30 2310 998161, Fax +30 2310 99806
- Page 113 and 114:
TURSKI, Prof. A., ul. Krochmalna 3
- Page 115 and 116:
Information for authorsContentThe R