A High Resolution Rdr-Acoustic Sensor for Detection of Close-in Air Turbulence Grhm Brooker (Author), Jvier Mrtinez ACFR/AMME, University of Sydney Sydney, Austrli gbrooker@cfr.usyd.edu.u Duncn Robertson Physics nd Astronomy, St Andrews University St Andrews, Scotlnd dr@st-ndrews.c.uk Abstrct - This pper presents novel D imging sensor for the identifiction of regions ir turbulence in front of UAV. It comprises swsh-plte mirror scnned Rdr Acoustic Sounding System (scrass) to produce high resolution imges of the ir in front of the UAV. The 4 khz/ 17 GHz nrrow bem RASS produces imges bsed on Brgg enhnced Doppler rdr reflections from the coustic pulse s it trvels. The technique is suited to imging still ir, or ir moving in lminr fshion but returns will be disrupted by ir turbulence, so the 3D imge generted will identify cler regions in front of the UAV. chnges in density occur, smll frction of the signl will be reflected t ech of the trnsitions. A. Brgg Mtching Brgg mtching occurs where the electromgnetic wvelength is equl to twice the coustic one which results in smll reflected components dding in phse to form lrger return s seen in Fig. 1 for 4 khz coustic signl. Keywords rdio coustic sounding; RASS; imging; ultrsound I. INTRODUCTION Reflection of electromgnetic rdition from brupt chnges in tmospheric chrcteristics is now well known effect. From the beginnings of rdr use in WWII it ws one of the phenomen tht produced rtefcts clled ngels [1]. However, it ws not until the lte 195s tht the chnges in refrctive index in ir induced by coustic signls were first identified []. Over the next 5 yers the phenomenon hs been used to produce progressively more sophisticted rdiocoustic sounding systems (RASS) to exmine ir temperture, wind profiles nd turbulence in the lower troposphere [3-5]. By the erly 199s, the technique ws being pplied to indoor problems [6-9] nd the ppliction ws being widened to other fields, such s IFF [9] nd the detection of wke vortices [1, 11]. This combintion of RASS with conventionl swsh-plte scnner bsed on our mining rdrs [1] produces novel imging sensor tht will be cpble of generting imges of still or lminr-flow ir in three dimensions, nd thus, identifying regions of ir turbulence. II. OPERATIONAL PRINCIPLES As n coustic wve propgtes through the ir, the density of the medium in one region periodiclly increses nd decreses in mnner which mkes these peks nd troughs pper to trvel in the direction of propgtion. These chnges in density result in subtle differences in the refrctive index, n, between pek nd the trough. When n electromgnetic wve psses through the ir, in which these cousticlly induced Fig. 1. Coherent sum of the reflected signls shows the Brgg condition The mplitude of the coherent sum increses linerly with the number of coustic cycles, N. As the received echo power is proportionl to the squre of the mplitude, it will be proportionl to N. In ddition the width of the Brgg region decreses proportionlly. In the RASS cse, the Brgg reflector is not sttic, but pulse of sound trvelling out from the trnsducer t velocity v 34m/s. It is esy to show tht if the coustic nd electromgnetic sensors re collocted, Doppler shift equl to the coustic frequency, f occurs. B. Effect of the UAV Speed f v d f. (1) The chnge in coustic wvelength (m) for rdil velocity v r (m/s), will be v r () f
This will hve the effect shifting the wvelength rtio, nd the coherent sum shown in Fig.1 will shift wy from the pek. To ccount for this, it will be necessry to djust the coustic or the electromgnetic frequency slightly. C. Focus Effect One feture of collocting the coustic nd rdr sensors is the focus effect shown in Fig.. In this geometry, both the wve-fronts expnd with the sme rdius of curvture nd so coherence is mintined over the full re of the expnding pulse [9]. N R Acoustic Pulse Rdr nd Acoustic Source Fig.. RASS geometry illustrting the focus effect of collocted sensors According to Clifford [13], the received power, P r (W) returned from reflection off n coustic pulse, if the RF bemwidth is wider thn the coustic bemwidth, cn be pproximted by sin k k e 17 N P P t ge r 4.61 R N g k ke P where P t trnsmitted RF power (W) P trnsmitted coustic power (W) g coustic ntenn gin N number of coustic pulses R Rnge (m) coustic wvelength (m) k = / k e = / e N This is only true if the conditions re perfect nd there re no tmospheric perturbtions. In the event tht the rdr bemwidth is nrrower thn the coustic bemwidth, then g e /g. is replced by g. Mrshll [5] provides n lmost identicl eqution but fctor of three lrger. The rdr cross section (RCS), (m ), of this expnding coustic pulse cn be determined in terms of the coustic power, P, the coustic ntenn bemwidth, (rd) nd the rnge, R (m). If the rdr bem is wider thn the coustic bem [9] then (3) 15 5 1.76 1 4 R N Pg 1 cos (4) 16 With some simplifiction (4) reduces to 1.69 1 R N P (5) 1 where (rd) is the coustic bemwidth. D. Effect of Turbulence The effects of turbulence re twofold. Firstly, locl chnges in the direction of the irflow cn distort the propgtion of the coustic pulse to reduce the effectiveness of the Brgg mtching. Secondly, more globl turbulence cn ffect the curvture of the pulse to reduce the focus effect for collocted sensor configurtion. Together these effects will reduce the effective RCS with the result tht the trcked pulse will be extinguished over short distnce. The rte t which the reduction in the echo return occurs, will be indictive of the mgnitude of the turbulence in tht direction. E. Atmospheric Attenution The eqution describing the RCS does not consider the ttenution of the coustic signl, which increses significntly with incresing frequency[14] [15]. At 4 khz the ttenution vries between 1.1dB/m nd 1.4dB/m depending on the reltive humidity. The ttenution corresponding to the rnge of opertion should be subtrcted from the RCS clculted in (4) to produce the true vlue. III. SYSTEM CONFIGURATION A. Monosttic RASS Rdr System The wvelength of n coustic system operting t 4kHz is 8.5mm. The rdr system must operte t wvelength of 17mm to stisfy the Brgg condition, which equtes to frequency of 17.65GHz. A conventionl Doppler rdr system with reflected power cnceller (RPC) hs been constructed from discrete components s shown in Fig. 3. To ADC YIG Oscilltor Premp Filter Coupler Mixer PA LNA Reflected Power Cnceller Circultor Horn Lens Antenn Fig. 3. Schemtic digrm of the Doppler rdr nd reflected power cnceller The RASS performnce hs been determined in simultion. The RCS defined in (4) nd modified by the tmospheric ttenution is plotted in Fig. 4. For n coustic power of 1 W, it cn be seen tht the RCS reches mximum t rnge of 6.5m before flling off s the tmospheric ttenution begins to dominte over the R term.
RCS (dbm ) -7-8 -9-1 -11-1 5 cycles 5 cycles 5 cycles 1 cycles -13 1-1 1 1 1 Rnge (m) Fig. 4. Rdr cross section of coustic pulse with the number of cycles, N, in pulse s prmeter The Doppler rdr model ssumes tht the received signl to noise rtio is limited by therml noise becuse the RPC cncels the phse noise lekge effects. The system performnce shown in Fig. 5 is determined for the following prmeters: Opertionl frequency 17.65GHz RF Trnsmit power 4dBm Antenn gin 5dB Receive filter bndwidth 3kHz System noise figure 5dB 1 pulses integrted Power (dbm) - -4-6 -8-1 -1 Received Power N=5 Received Power N=5 Received Power N=5 Received Power N=1 Integrted Noise Floor B. Dt Acquisition nd Processing The controller consists of lptop nd the Dt Trnsltion DT9836 USB bsed IO module. This module includes 16bit ADC smpling t up to 5 ks/s to cquire the Doppler signl nd 16-bit DACs running t 5 ks/s to generte the coustic pulses. A one chnnel of stereo hifi mplifier ws used to drive the coustic rry. Softwre ws developed using DT Mesure Foundry which llows the synchronous genertion of coustic pulses nd the cquisition of the rdr outputs s well s some rel-time processing. Once sufficient mesurements hve been smpled nd coherently integrted, the Doppler frequency cn be extrcted from the composite using moving-fft bsed spectrogrm [1]. The mplitude of received Doppler signl is determined s function of time, nd it is compred to the still ir profile. An unexpected knee where the profile diverges from theory indictes tht the coustic signl is being scttered. The shrpness of the knee is function of the length of the coustic pulse, the spectrl estimtion method nd the mount of overlp selected between successive estimtes. IV. SYSTEM MEASUREMENTS A. Reflected Power Cnceller Performnce Wveguide to SMA dpters were ttched to the two wveguide ports of the Doppler rdr, shown in Fig. 6, nd these coupled to the two terminls of clibrted 851 Network Anlyzer. The S11 nd S1 mesurements were conducted for number of configurtions. Receive out Adjustble Short Directionl Coupler 1dB Level Set Attenutor Circultor Circultor Antenn -14-16 Trnsmit in 1dB Directionl Coupler -18-1 -1 1 1 1 Rnge (m) Fig. 5. Received signl nd noise levels with the number of cycles, N, in pulse s prmeter Becuse the cquisition of the Doppler signl cn be synchronized with the genertion of n coustic pulse, it is possible to integrte lrge number of mesurements to improve the SNR. For exmple, for 1 m mximum rnge, ech mesurement tkes 3ms, so the coherent integrtion of 5 returns, would only require 75 ms to perform. The number of cycles in pulse should be selected depending on the sptil resolution required nd the vilble SNR. For n coustic wvelength of 8.5mm, nd N = 6, the pulse spns rnge of 51 mm which defines the sptil resolution for the received Doppler mesurement. As the rnge is decresed, the vilble SNR increses nd fewer cycles cn be used with resulting improvement in the sptil resolution. Fig. 6. Block digrm of the wveguide components of the Doppler rdr portion of the RASS B. Shorted Antenn Port This test ws conducted to determine how well the components were mtched nd to scertin the mgnitude of the wveguide nd component losses compred to specifictions. The level set ttenutor component of the RPC is set to mximum ttenution to minimise its ffect in this test. TABLE I. SPECIFIED LOSSES THROUGH THE WAVEGUIDE SECTIONS Component Loss (db) Circultor JQL JCWR6-4.3 Coupler Penn Eng 538-111A-1.46+.3 W/G bends 38-11A.3 W/G to Cox Penn Eng 1438 1AM5.55 Cox bends (mesured). Totl 1.6dB
Note tht the.46db loss through ech coupler corresponds to the proportion of the power tpped off by 1dB coupler, 1log 1(1-1 -1/1 ) =.458dB The mesured return nd trnsmission losses over 1 GHz bndwidth from 17 to 18 GHz re shown in Fig. 7. -5 S11 nd S1 for Shorted Antenn Port S11 S1 S1 for Flnn Horn Antenn into the ceiling orthogonl nd t n ngle - 1.8mm Bckshort - S1 orthogonl -5 S1 ngle -3-35 -4-45 -5 Reltive Power (db) -1-55 Reltive Power (db) -15 - -5-6 17 17.1 17. 17.3 17.4 17.5 17.6 17.7 17.8 17.9 18 Frequency (GHz) Fig. 9. Photogrph of the RASS with the stndrd gin horn ntenn nd return loss tuned to 17.6 GHz with the bem perpendiculr to the ceiling nd t n ngle -3-35 17 17.1 17. 17.3 17.4 17.5 17.6 17.7 17.8 17.9 18 Frequency (GHz) Fig. 7. Mesured return (S11) nd trnsmission (S1) losses for the wveguide sections of the RASS with miniml contribution from the RPC It cn be seen tht the overll component mtching is resonble with typicl return loss of -15dB. However, the trnsmission losses of.5db cross the bnd re nerly 1dB higher thn the 1.6 db clculted in Tble I. No ttempt ws mde to identify the culprits. C. Antenn Port with Mtched Lod This test ws conducted to determine how well the RPC could perform under idel conditions. As cn be seen from Fig. 8, return loss remined below -15 db, while the trnsmission loss could be djusted to produce extremely deep nrrow bnd nulls of -6 db nd more. This confirms tht the RPC functions s specified. S1 for Mtched Antenn Port - RPC Set to best cncelltion t 17.6GHz nd 17.65GHz -5 S1 t 17.65GHz -3 S1 t 17.6GHz Reltive Power (db) -35-4 -45-5 E. Rdr Receiver Requirements Fig. 5 predicts tht the received Doppler echo power from n coustic pulse will be something between -135 nd -155dBm t rnge of 4m. This signl needs to be mplified by t lest 1 db to rech mv levels suitble for the ADC bord. This gin is chieved by pir of cscded RF mplifiers providing 53 db of gin nd n udio mplifier nd filter with further 57 db of gin t 4 khz s shown in Fig. 1. To ADC Amp 57dB @ 4kHz -7dB 39dB 14dB Filter Premp Mixer LNA LNA -1.5dB Circultor Fig. 1. Block digrm showing the components of the receiver chin The udio mplifier gin chrcteristics were configured with sufficient bndwidth to ccommodte vritions in the Doppler frequency of the received signl due to either chnges in temperture or the speed of the UAV. V. SYSTEM HARDWARE A RASS ws built using polrised wire reflector in spce frme structure documented by Wei [8]. This configurtion is shown in Fig. 11. To minimize microphonics, which hd plgued erlier configurtions of the system, the coustic rry ws hung from rubber bnds (not visible). -55-6 -65 17 17.1 17. 17.3 17.4 17.5 17.6 17.7 17.8 17.9 18 Frequency (GHz) Fig. 8. Mesured trnsmission losses for the wveguide sections of the RASS with the RPC djusted for mximum cncelltion t 17.6 nd 17.65 GHz D. Antenn Port with Stndrd Gin Horn The mesured results shown in Fig. 9 confirm tht reflected power cncelltion is lso dependent on close in clutter returns, but tht good cnceltion is possible under norml, non-speculr conditions under which the RASS will operte. Fig. 11. Photogrph of the scnning RASS sensor with the vrious components lbelled
A. System Clibrtion To mesure the system performnce, it is convenient to use Doppler reflector with known RCS. Conventionl moving trgets re not suitble s the receiver is tuned to provide mximum sensitivity t round 4 khz, which corresponds to velocity of 34 m/s. In ddition, the expected RCS is incredibly smll, s cn be seen in Fig. 5. To chieve this, smll Doppler trget ws developed using 4 khz piezo trnsducer nd smll bll bering. Sinusoidl excittion voltges of between 1V nd 1V produced vritions in the mesured RCS from -14.5 dbsqm to -1.5 dbsqm [16]. VI. RANGING The system ws first operted in verticl-pointing mode (without the scnner component) to determine its performnce under different conditions. To confirm tht the coustic rry ws well isolted from the Doppler rdr, system ws first operted with the trnsmit perture blocked. It cn be seen from Fig. 1 tht prt from the coustic bng pulse every ms, the signl level is constnt. This is verified in Fig. 13 which shows the spectrogrm is empty t 4 khz prt from the bng pulse period. The second test conducted in still ir outdoors shows n exponentilly decying signl mplitude s function of time with some modultion (Fig. 14). The spectrogrm shown in Fig.15, confirms tht echoes is received right up to the strt of the next pulse ms lter. This corresponds to rnge of 6.8 m t nominl speed of sound of 34 m/s. Signl (mv) 15 1 5-5 -1-15 5 1 15 5 Fig. 14. Time domin signl received by the Doppler rdr for opertion outdoors 8 6 1 4 8 Signl (mv) Frequency(kHz) 6 4 - -4 4 45 5 55 6 65 7 4 6 8 1 1 14 16 18 Fig. 1. Time domin signl received by the Doppler rdr with the trnsmit perture blocked Frequency(kHz) 1 8 6 4 Fig. 15. Spectrogrm of the Doppler signl for opertion outdoors The finl test, with results shown in Fig. 16 nd Fig. 17, shows the results for opertion indoors with roof height of 5m. In this cse, the signl ttenutes following the norml exponentil decy profile, but ends fter 15 ms. Note tht the differences in the pek mplitudes of the time domin in Fig. 14 nd Fig. 15 re s result of the chnged VSWR, nd hence sensitivity, of the system when operted outside where there is very little reflected power, nd indoors pointing towrds flt roof. 4 6 8 1 1 14 16 18 Fig. 13. Spectrogrm of the Doppler signl with the trnsmit perture blocked
Signl (mv) 8 6 4 - -4-6 4 45 5 55 6 65 7 Fig. 16. Time domin signl received by the Doppler rdr for opertion indoors towrds the roof t rnge of 5m Frequency(kHz) 1 8 6 4 4 6 8 1 1 14 16 18 Fig. 17. Spectrogrm of the Doppler signl for opertion indoors towrds the roof t rnge of 5m VII. CONCLUSIONS This pper hs discussed the development of novel RASS comprising n rry of one hundred 4 khz ultrsonic elements collocted with sensitive Doppler rdr. The lrge coustic perture nd high frequency result in nrrow bem which llows for the remote mesurement of tmospheric turbulence within smll volume in spce (.5m.5m.5m) t rnge of 7m. Mesurements were mde to confirm the opertion of the sensor in its verticl rnging mode, nd it ws ble to detect coustic pulses out to beyond 6.8m. A swsh plte scnner, developed for our mining pplictions, ws integrted to direct the bem but could not be tested in time for inclusion in this pper. However, once this is chieved, sector of spce in front of the system will be scnned to generte 3D mp of regions of ir turbulence. Once this proof of concept hs been estblished, the next phse will be to hrden nd integrte such system onto fixed-wing UAV for irborne tests. [] A. Tonning, "Scttering of Electromgnetic Wves by n Acoustic Disturbnce in the Atmosphere," Applied Science Res., vol. B6, pp. 41-41, 1957. [3] N. Bhtngr nd A. Peterson, "Interction of Electromgnetic nd Acoustic Wves in Stochstic Atmosphere," IEEE Trns. on Antenns nd Propgtion, vol. AP-7, pp. 385-393, My 1979. [4] M. Frnkel, N. Chng, nd M. Snders Jr, "A High-Frequency Rdio Acoustic Sounder for Remote Mesurement of Atmospheric Winds nd Temperture," Bulletin Americn Meterologicl Society, vol. 58, pp. 98-933, September 1977. [5] J. Mrshll, A. Peterson, nd A. Brnes Jr, "Combined Rdr- Acoustic Sounding System," Applied Optics, vol. II, pp. 18-11, Jnury 197. [6] M. Ds nd R. Knochel, "Microwve-coustic Mesurement System for Remote Temperture Profiling in Closed Environments," in EUMC, Helsinki, Finlnd, 199, pp. 15-13. [7] M. WeiB nd R. Knochel, "A Monosttic Rdio-Acoustic Sounding System," IEEE MTT-S Digest, vol. THF4-9, pp. 1871-1874, 1999. [8] M. WeiB nd R. Knochel, "A Monosttic Rdio-Acoustic Sounding System Used s n Indoor Remote Temperture Profiler," IEEE Trns. on Instrumenttion nd Mesurement., vol. 5, pp. 143-147, October 1 1. [9] J. Sffold, F. Willimson, K. Ahuj, L. Stein, nd M. Muller, "Rdr-coustic Interction for IFF Applictions," in IEEE Rdr Conference, Wlthm, MA, 1999, pp. 198-. [1] J. Hnson nd F. Mrcotte, "Aircrft Wke Vortex Genertion Using Continuous-Wve Rdr," Johns Hopkins Apl. Technicl Digest, vol. 18, pp. 348-357, 1997. [11] R. Mrshll, "Wingtip Generted Wke Vortices s Rdr Trgets," IEEE AES Systems Mgzine, pp. 7-3, December 1996. [1] G. Brooker, M. Bishop, nd R. Hennessey, "Evolution of Suite of Millimetre Wve Rdr Systems for Situtionl Awreness nd Automtion in Mines," presented t the 8 Austrlin Mining technology Conference, Sunshine Cost, QLD, Austrli, 8. [13] S. Clifford nd T. Wng, "The Rnge Limittion on Rdr- Acoustic Sounding Systems (RASS) due to Atmospheric Refrctive Turbulence," IEEE Trns. on Antenns nd Propgtion, vol. AP-3, pp. 319-36, My 1977. [14] L. Kinster nd A. Frey, Fundmentls of Acoustics nd Ed. New York: John Wiley & Sons, 196. [15] N. Burnside, "A Function tht Returns the Atmospheric Attenution of Sound," ed: MATLAB Centrl - http://www.mthworks.com/mtlbcentrl/fileexchnge/lodfile.d o?objectid=6&objecttype=file, 4. [16] G. M. Brooker, "An Adjustble Rdr Cross Section Doppler Clibrtion Trget," Sensors Journl, IEEE, vol. 15, pp. 476-48, 15. REFERENCES [1] M. Skolnik, Introduction to Rdr Systems, nd ed.: McGrw-Hill Kogkush, 198.