17
137 1 Introduction Local Bureaus of Telecommunications confirmed that there were approximately 36,000 illegal radio stations in 1997, approxi- mately 45,000 in 1998 and approximately 37,000 in 1999[ ]. Illegal stations cause prob- lems by leading to interference and jamming of important or general-operational radio com- munications. Adequate measures must be taken against such problems. We conducted a study to identify press-to-talk transmitters by analyzing their transient-response characteris- tics at the time of the transmitters are turned on and off[ ]. A previous study reported that a transient response lasted for several μs to sev- eral hundreds of ms[ ]. Insufficient reports were available regarding detailed descriptions of the reproducibility, stability, and effect (on the environment) of a transient-response dura- tion, thus it was difficult to develop an effec- tive procedure for identifying radio transmit- ters. In the earlier part of this report, in order to determine the general characteristics and tendencies of the transient-response duration, we applied the saturation voltage to normalize an amplitude envelope at the time of a signal rise[ ]. We also examined the effects of varia- tions in power voltage and ambient tempera- ture on performance in the identification of radio transmitters. In the latter part of this report, we compare the results of the above- mentioned indoor experiment with the experi- mental results obtained using an antenna, to determine the effectiveness of the present sys- tem with various types of transmitters. Tsutomu SUGIYAMA et al. Transmitter Identification Experimental Techniques and Results Tsutomu SUGIYAMA, Masaaki SHIBUKI, Ken IWASAKI, and Takayuki HIRANO We delineated the transient response patterns of several different radio transmitters in order to determine the patterns most useful in the development of a transmitter identifica- tion system. Using a high-speed data acquisition system, we first obtained rise and fall data for various transient response patterns that were produced by six different FM radio trans- mitters when the press-to-talk buttons were swicthed on and off. We next evaluated the effect of these transient patterns on the time domain and time-frequency spectrograms by measuring the changes in the receiver input level at Pin = -120 dBm and SNR = 7.2 dB obtained from three different bandwidths : 250, 50, and 12 kHz. Similarly, spectrogram analysis was used to obtain information about the relationship between bandwidth and noise. Comparison of the spectrogram patterns obtained in both laboratory and field exper- iments independently corrobrated our results. However, the transient patterns in the time domain could not be used because of extensive distortion related to noise and interference from neighboring radio stations. These results suggest that spectrogram analysis of tran- sient reponse patterns may be the most effective way to secure reliable identification of the FM radio transmitters. Keywords Radio transmitter identification, Spectrogram pattern, Wigner-distribution

Transmitter Identification Experimental Techniques and Results · Transmitter Identification ... A 145-MHz FM transceiver, TR51 of Manu-facturer i, was used under conditions in which

  • Upload
    others

  • View
    3

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Transmitter Identification Experimental Techniques and Results · Transmitter Identification ... A 145-MHz FM transceiver, TR51 of Manu-facturer i, was used under conditions in which

137

1 Introduction

Local Bureaus of Telecommunicationsconfirmed that there were approximately36,000 illegal radio stations in 1997, approxi-mately 45,000 in 1998 and approximately37,000 in 1999[1]. Illegal stations cause prob-lems by leading to interference and jammingof important or general-operational radio com-munications. Adequate measures must betaken against such problems. We conducted astudy to identify press-to-talk transmitters byanalyzing their transient-response characteris-tics at the time of the transmitters are turnedon and off[2]. A previous study reported that atransient response lasted for severalμs to sev-eral hundreds of ms[3]. Insufficient reportswere available regarding detailed descriptions

of the reproducibility, stability, and effect (onthe environment) of a transient-response dura-tion, thus it was difficult to develop an effec-tive procedure for identifying radio transmit-ters. In the earlier part of this report, in orderto determine the general characteristics andtendencies of the transient-response duration,we applied the saturation voltage to normalizean amplitude envelope at the time of a signalrise[4]. We also examined the effects of varia-tions in power voltage and ambient tempera-ture on performance in the identification ofradio transmitters. In the latter part of thisreport, we compare the results of the above-mentioned indoor experiment with the experi-mental results obtained using an antenna, todetermine the effectiveness of the present sys-tem with various types of transmitters.

Tsutomu SUGIYAMA et al.

Transmitter Identification ―ExperimentalTechniques and Results―Tsutomu SUGIYAMA, Masaaki SHIBUKI, Ken IWASAKI, and Takayuki HIRANO

We delineated the transient response patterns of several different radio transmitters inorder to determine the patterns most useful in the development of a transmitter identifica-tion system. Using a high-speed data acquisition system, we first obtained rise and fall datafor various transient response patterns that were produced by six different FM radio trans-mitters when the press-to-talk buttons were swicthed on and off. We next evaluated theeffect of these transient patterns on the time domain and time-frequency spectrograms bymeasuring the changes in the receiver input level at Pin = -120 dBm and SNR = 7.2 dBobtained from three different bandwidths : 250, 50, and 12 kHz. Similarly, spectrogramanalysis was used to obtain information about the relationship between bandwidth andnoise. Comparison of the spectrogram patterns obtained in both laboratory and field exper-iments independently corrobrated our results. However, the transient patterns in the timedomain could not be used because of extensive distortion related to noise and interferencefrom neighboring radio stations. These results suggest that spectrogram analysis of tran-sient reponse patterns may be the most effective way to secure reliable identification of theFM radio transmitters.

Keywords Radio transmitter identification, Spectrogram pattern, Wigner-distribution

Page 2: Transmitter Identification Experimental Techniques and Results · Transmitter Identification ... A 145-MHz FM transceiver, TR51 of Manu-facturer i, was used under conditions in which

138

In the field experiment, we qualitativelyand quantitatively evaluated 1) a change in theinput level (Pin) and 2) its effects on an ampli-tude envelope waveform and spectrogram atthe time of a rise in the time domain. Theevaluation of the spectrogram was comparedwith that made in the indoor experiment. Thiscomparison revealed that the patterns of aspectrogram with characteristic values of -120dBm and SNR = 7.2 dB agreed in the indoorand field experiments. Therefore, such pat-terns may be applied to the identification ofradio transmitters.

2 Indoor experiment

In this indoor experiment, we used thesame high-speed data-acquisition system inthe previous report that described the configu-ration of the system, connection to signals,and the installation. Twenty-four radio trans-mitters made by three manufacturers wereexamined in this experiment, divided into 6models. In this report,αindicates a 144-MHzband, whileβindicates a 430-MHz band. ‘A,’‘B,’ and ‘C’ are manufacturers. For details,refer to Appendix 1 for the specifications ofthe transmitters.

Fig.1 shows typical amplitude envelopewaveforms at the time of a rise for 6 modelsof the three manufacturers. The vertical axisindicates the amplitude in units of mV, whilethe horizontal axis indicates the duration oftime in units of ms. Manufacturer A, shown atthe top of this figure, had a pulse-like wave-form that converged in a short period before amajor rise appeared. At the major rise, thewaveform overshot and then slowly attenuat-ed, reaching a constant level. After overshoot-ing, Manufacturer B’s waveform graduallyattenuated and converged at a constant level.Unlike A, B had no pulse-like waveform,overshot for a longer time, and had a roundpattern. Manufacturer C’s waveform did notovershoot, increased slowly, and saturatedafter an extended period. Thus, amplitudeenvelope waveforms differed by transmittermodel.

Fig.2 shows falling waveforms. All trans-mitters had their signal levels cut off, greatlyreducing the time required for falling. Thisproduced very little of the information neces-sary to identify transmitters, compared withthe use of data for a rise. Therefore, we decid-ed to examine data for a rise in order to identi-fy transmitters.

2.1 Processing and evaluation ofacquired data

We now know from Fig.1 that amplitudeenvelope waveforms differed by transmittermanufacturers and model. To processacquired data, we define an inter-thresholdtime lag, as shown in Fig.3. Assuming the sat-uration level (V) of the amplitude to be 100,the amplitude voltage (Vs) is normalizedevery 10% during transition. We refer to thetransition process as a “threshold.” A time lagbetween thresholds, Δt, is defined as the timerequired for the voltage to reach a value of Vs

Journal of the Communications Research Laboratory Vol.49 No.1 2002

Amplitude envelope waveforms at thetime of a rise

Fig.1

Amplitude envelope waveforms at thetime of falling

Fig.2

Page 3: Transmitter Identification Experimental Techniques and Results · Transmitter Identification ... A 145-MHz FM transceiver, TR51 of Manu-facturer i, was used under conditions in which

139

with the saturation time defined as 0 ms.Fig.4 shows a typical relationship between thetime lag and the normalized amplitude volt-age. This example is a result obtained from10 continuous measurements using the 144-MHz band Model 1 of Manufacturer A. Thebars in the figure indicate the error rangesaround an average. The evaluation of enve-lope waveforms was conducted in comparisonwith the indoor-experiment data, and isdescribed below.

2.2 Dependence of the time-lag-nor-malized voltage relationship on thepower voltage

Fig.5 shows the measured relationshipsbetween the inter-threshold time lag and thenormalized voltage at differing power volt-ages. These results were obtained by averag-ing the measurements for supply voltages of15.2 V, 13.8 V, and 12.4 V at a temperature of+25℃. The transmitters were placed in a ther-

mostatic chamber. As can be seen from thisfigure, the time lag tended to decrease with anincrease in the supply voltage. This proved tobe common to all of the models we examined.Modelα1 of Manufacturer C, however,showed the strongest tendency in this regard.

2.3 Dependence of the time-lag-nor-malized voltage relationship on thetemperature

Fig.6 shows the measured relationshipsbetween the inter-threshold time lag and thenormalized voltage at differing temperatures.We placed the transmitters in a thermostaticchamber to obtain these results by averagingten consecutive measurements each at temper-atures of +40℃, +25℃, and –10℃. In Fig.6,the vertical bars indicate errors. It was com-mon to the six models of the three manufac-turers that, as the set temperature lowered, theinter-threshold time decreased. Most of themodels behaved similarly, except that Modelsαandβof Manufacturer C showed relativelylarge dependencies. Modelβof ManufacturerA showed a large change in voltage ofbetween 10% and 20% at –10℃, due theeffect of a pulse-like waveform prior to a rise.

2.4 Difference in the time-lag-normal-ized voltage relationship among fourtransmitters of the same model

Fig.7 shows the differences in the time-lag-normalized voltage relationships amongfour transmitters of the same model. Themeasurement conditions were a constant tem-perature of +25℃ and power voltage of 13.8V. The four transmitters in each groupbehaved similarly, except for ManufacturerA’s Modelα1. Fig.8 shows the averaged rela-tionships for all models, indicating that thereis a good possibility of identifying transmittersbased on the significant differences amongtheir six types. As all models of each manu-facturer behaved similarly, it may be possibleto identify the manufacturer of a transmitter.It may also be possible to distinguish betweenthe different models of one manufacturer.However, it will be difficult to identify trans-

Tsutomu SUGIYAMA et al.

Definition of inter-threshold time lag Fig.3

Plotting of the inter-threshold time lagagainst normalized voltage

Fig.4

Page 4: Transmitter Identification Experimental Techniques and Results · Transmitter Identification ... A 145-MHz FM transceiver, TR51 of Manu-facturer i, was used under conditions in which

140 Journal of the Communications Research Laboratory Vol.49 No.1 2002

Dependence of the time-lag-normalized voltage relationship on the power voltageFig.5

Dependence of the time-lag-normalized voltage relationship on the temperatureFig.6

Page 5: Transmitter Identification Experimental Techniques and Results · Transmitter Identification ... A 145-MHz FM transceiver, TR51 of Manu-facturer i, was used under conditions in which

141

mitters of the same model and manufacturer.The possibility of this last identification willbe described in view of time-frequency spacein another report[5].

3 Experiment using an antenna

We conducted an experiment using anantenna by changing conditions such as the

receiver input intensity level, S/N ratio, IFbandwidth, and trigger point. To eliminate asmany variables as possible and improve relia-bility, one common receiver antenna and onecommon transmitter antenna were installed.They were separated by approximately 300meters to prevent the effect of box radiation.No hindrance, such as tall building, was pres-ent between them.

Fig. 9 shows the discone antenna that wasavailable on the market and installed on abuilding (Building No.3) rooftop. An attenua-tor was placed between the antenna and trans-mitter to adjust the output power to thatrequired by the receiver. A stabilized powersupply was used for the transmitters, and theywere examined under the same conditions asin the indoor experiment. Appendix 1 shows alist of the transmitters tested in the indoorexperiment. The circled transmitters in thislist were tested in the present experiment.

Tsutomu SUGIYAMA et al.

Difference in the time-lag-normalized voltage relationship among four transmitters of thesame model

Fig.7

Averaged time-lag-normalized volt-age relationships for all models

Fig.8

Page 6: Transmitter Identification Experimental Techniques and Results · Transmitter Identification ... A 145-MHz FM transceiver, TR51 of Manu-facturer i, was used under conditions in which

142

3.1 Data acquisition through an an-tenna

In the experiment using an antenna toacquire data, it was likely that data acquisitionwas affected by interference and a loweredS/N ratio caused by neighboring radio stationsin the same band. We first studied theseeffects on transmitter identification. Fig.10shows a measured amplitude waveform thatwas affected by such interference and noise.A 145-MHz FM transceiver, TR51 of Manu-facturer i, was used under conditions in whichAtt of 30 dB was added to a transmission out-put of 5 W, the receiver input-level Pin was–110 dBm, and the S/N ratio was 7.2 dB.

Fig.11 shows an amplitude waveformobtained in the indoor experiment. A compar-ison of Figs. 10 and 11 revealed that dataacquired through an antenna contained a largeamount of noise, and that the distorted signalsmade it difficult to obtain the characteristics atthe rise of a waveform. Fig.12 shows a spec-trogram obtained by using fast Fourier trans-formation (FFT) to convert I-Q data into a fre-quency region. Fig.13 shows a spectrogram

for the indoor experiment obtained by usingFFT to convert the field-experiment I-Q datainto a frequency region.

In Fig.10, the waveform seems to havebeen affected greatly by noise. However, itproved to be interference that actually affectedthe waveform. As can be seen in Fig.12, theraw data can be separated into the signal,interference-induced wave, and noise. Theeffect on the waveform depended on thedegree of interference. A comparison betweenthe waveform and interference by overlayingrevealed no deformation or distortion of thespectrogram. This will be described in greaterdetail in the figure of a spectrogram patternshown below. Unlike in the indoor experi-ment, waveform data acquired through anantenna proved to be useless for identifyingtransmitters unless noise and interference weresufficiently negligible. On the other hand,very little effect was seen on the spectrogram

Journal of the Communications Research Laboratory Vol.49 No.1 2002

Discone antenna for use in the presentexperiment

Fig. 9

Measured amplitude waveformaffected by interference on Tr51

Fig.10

Amplitude waveform obtained in theindoor experiment using Tr51

Fig.11

Page 7: Transmitter Identification Experimental Techniques and Results · Transmitter Identification ... A 145-MHz FM transceiver, TR51 of Manu-facturer i, was used under conditions in which

143

calculated from the data acquired through anantenna. The pattern of a spectrogram provedto be useful for identifying transmitters in boththe indoor experiment and field antennaexperiment. The next section will describe theusefulness of the spectrogram pattern.

3.2 Data processingThe previous section discussed the capa-

bility of a spectrogram obtained by convertingI-Q data through FFT. This application soft-ware enables determination of the frequencyand spectrum intensity at any point specifiedusing a mouse cursor. When this function isused, the displayable range of the screen is 10ms. In this study, to obtain the entire patternof the spectrogram, we moved the cursor tothe peak spectrum every 1 ms in order to readthe frequency and spectrum intensity.

Fig.14 shows the frequency (upper) andspectrum intensity (lower) obtained for trans-ceiver Tr51 of Manufacturer i. The attenua-

tion level for transmission was 0 dB, 10 dB,20 dB, and 30 dB. The frequency graduallychanged with the attenuation level, up to 20dB. With 30-dB attenuation, the pattern shift-ed forward significantly. Now we will explainthe results of the examination about the triggersetting and the shifted pattern. The indoorexperiment confirmed that this type of trans-mitter produced by Manufacturer i caused afrequency step of between 20 ms and 40 ms.For attenuation of 0 dB, Fig.14 indicates thatfrequency stepping took place 33 ms after thetrigger started. In this figure, ① through ④indicate the occurrence of frequency steppingwhen the input-level Pin was switched usingthe attenuator. For ① through ③, the step-ping was small and shifted forward by approx-imately 2 ms. For ③ and ④, the steppingshifted by 5 ms. It should be noted that thetrigger setting was unchanged between 0 dBand 20 dB, while at 30 dB it was changed by–97 dBm to –107 dBm. The entire patternsshifted due to the effects of a change in thetrigger setting and the lowered S/N ratio.These effects did not result in distortion of thepattern.

Fig.14 (lower) shows the dependencies ofthe peak of a spectrum signal on the elapsedtime. The signal level increased slowly fol-lowing a rise for a duration of over ten ms,and then became constant. Switching of theattenuator every 10 dB caused an expectedchange in the signal level. This confirmedthat the equipment and related software oper-ated normally.

Fig.15 shows the time at which steppingtook place for the four transmitters at a triggerlevel of 0 dBm based on the results of theindoor experiment. For Tr51, all data pointsare plotted.

Next, we examined transmitter Tr39 ofManufacturer y, as it showed a large frequencychange at the time of a rise, and the changewas not reproducible in a number of measure-ments. Fig.16 shows the averages and errorsfor the peak frequency with an interval of 1ms, in ten measurements that were taken dur-ing the first 40 ms after a rise. This figure

Tsutomu SUGIYAMA et al.

Spectrogram affected by interfer-ence on Tr51

Fig.12

Spectrogram obtained in the indoorexperiment using Tr51

Fig.13

Page 8: Transmitter Identification Experimental Techniques and Results · Transmitter Identification ... A 145-MHz FM transceiver, TR51 of Manu-facturer i, was used under conditions in which

144

indicates that approximately 16 ms were

required for the peak frequency to stabilizefollowing convergence. As the attenuationincreased from 0 dB to 30 dB, the peak-fre-quency average peaked earlier. As with Tr51,this was an effect of the trigger-level setting.As can also be seen in this figure, the averagedid not lie in the center of the error bar.

Fig.17 shows the results for four transmit-ters of the same model. When data overlapsas shown in this figure, it is difficult to identi-fy the transmitters. This is a problem thatremains to be solved in the future.

Fig.18 shows the results obtained in a fieldexperiment in which Att = 10 dB (○) and inan indoor experiment (●). The indoor-experi-ment result is the average of measurements ofthe peak frequency. The arrows indicate fluc-tuations. The average peak frequency wassimilar in the two experiments, while the timeof occurrence of the peak frequency in the

Journal of the Communications Research Laboratory Vol.49 No.1 2002

Frequency (upper) and spectrum intensity (lower) with an interval of 10 ms for transceiverTr51 of Manufacturer i

Fig.14

Fig.15 Time at which stepping occurred forthe four transmitters at a trigger levelof 0 dBm. For Tr51 of Manufacturer i,all data points are plotted

Page 9: Transmitter Identification Experimental Techniques and Results · Transmitter Identification ... A 145-MHz FM transceiver, TR51 of Manu-facturer i, was used under conditions in which

145Tsutomu SUGIYAMA et al.

Averages and errors for the peak frequency with an interval of 10 ms for transmitter Tr39 ofManufacturer y

Fig.16

Page 10: Transmitter Identification Experimental Techniques and Results · Transmitter Identification ... A 145-MHz FM transceiver, TR51 of Manu-facturer i, was used under conditions in which

146

field experiment showed fluctuations twicethat seen in the indoor experiment. This isdue to the fact that the receiver functionsusing the antenna produced noise that seemedto have an adverse effect on data.

4 Spectrogram pattern

Acquired data was converted into a spec-trogram. A spectrogram pattern was used toidentify the adequate features of the transmit-ter. A pattern such as that obtained using a 10-ms interval in the previous section was notsufficient to identify the transmitter. We need-ed a tool that would enable the quick andaccurate acquisition of a spectrogram pattern.For this purpose, we applied software devel-oped in the course of analyzing data.

Fig.19 shows an example obtained usingthe above software. The upper graph showsthe absolute values of amplitude. To obtainthe lower graph, fast Fourier transformation(FFT) was first applied to the raw data with adata length (Nfft) of 1024 bits in order toobtain a spectrogram. The lower graph inFig.19 shows the contours of this spectrogram,with an interval of 0.05% between 10% and95% of the peak level (Pmax). The contourscorrespond to decibel steps down to –22 dB,with step intervals being between 1 dB and 2dB. The data length of 1024 bits was selectedin consideration of the frequency resolution,time resolution, and spectrum density distribu-tion. The vertical axis in the lower graph indi-cates the deviation from the center frequencywithin a range of ±15 kHz. The horizontalaxis indicates the elapsed time from the startof triggering to 102 ms. This tool enables theselection of a linear scale (as shown in thisgraph) or decibel scale. This graph clearlyconfirms the frequency step characteristics oftransmitter Tr51 in the spectrogram pattern.

Fig.20 shows an example, similar to thatshown in Fig.19, in which attenuation of 30dB was applied to create strong interference.While Fig.19 indicates only small effects ofinterference and noise on the raw amplitudedata, Fig.20 indicates that the effects of inter-ference and noise are too large to confirm theoriginal waveform. As shown in Fig.14, thechange in the trigger settings and the S/N ratio(from 24.3 dB to 7.2 dB) caused the entirespectrogram pattern to shift backward byapproximately 4μs. The spectrogram patterns

Journal of the Communications Research Laboratory Vol.49 No.1 2002

Time at which a frequency peakoccurred for four transmitters of Tr39of Manufacturer y

Fig.17

Comparison of the time at which afrequency peak occurred in a fieldexperiment and indoor experiment

Fig.18

Page 11: Transmitter Identification Experimental Techniques and Results · Transmitter Identification ... A 145-MHz FM transceiver, TR51 of Manu-facturer i, was used under conditions in which

147

shown in Figs. 14 and 20 were examined andfound to be in good agreement without signifi-cant distortion.

Fig. 21 and 22 show the calculated spec-trogram patterns for transmitter Tr39 of Manu-facturer y. This transmitter experienced asteep change in frequency in the initial period(0 to 30 ms) after a rise. Thereafter, the fre-quency converged quickly to the set value.The graphs in these figures contain calcula-tions of the largest and smallest changes infrequency in ten measurements. As shown inFig.17, four transmitters of the same type asTr39 produced very similar characteristics forfrequency stepping. As this data overlapsclosely, it was impossible to identify the trans-mitters. This is a problem that remains to besolved in the future.

Fig.23 shows the calculated spectrogrampatterns for transmitter Tr32 of Manufacturerk. This transmitter experienced a steepchange in frequency in the initial period aftera rise, followed by an oscillatory attenuation.Other transmitters produced complex patternsand were identifiable.

Tsutomu SUGIYAMA et al.

Amplitude waveform and calculatedspectrogram pattern for transmitterTr51

Fig.19

Amplitude waveform and calculatedspectrogram pattern affected byattenuation of 30 dB to create stronginterference

Fig.20

Calculated spectrogram patterns (1)for transmitter Tr39

Fig.21

Page 12: Transmitter Identification Experimental Techniques and Results · Transmitter Identification ... A 145-MHz FM transceiver, TR51 of Manu-facturer i, was used under conditions in which

148

5 Spectrogram pattern and IFbandwidth

All measurements described in the preced-ing sections were made with the IF bandwidthfixed at 250 kHz. A low-pass filter for 40 kHzwas applied to produce absolute values ofamplified waveform and spectrogram patterns.In this section, we describe the IF bandwidthand the low-pass filter.

Fig.24 shows the spectrogram patternobtained by applying a low-pass filter with acutoff frequency of 20 kHz to the I-Q datameasured with an IF bandwidth of 50 kHz bytransmitter Tr39 of Manufacturer y. In Figs.21 and 22, which were obtained using thesame transmitter, Tr39, the amplitude wave-form initially overshot and then maintained aconstant level. In Fig.24, the initial portionshowed distortion due to the effect of the low-pass filter for eliminating frequencies abovethe cutoff value. However, the spectrogrampattern was not affected.

Fig.25 shows a spectrogram patternobtained by applying a low-pass filter with acutoff frequency of 10 kHz to the I-Q datameasured with an IF bandwidth of 12 kHz bytransmitter Tr39 of Manufacturer y. The ini-tial portion of the amplitude waveformshowed much larger distortion than in Fig.24.However, the spectrogram pattern was notaffected, except for lack of data above the cut-off value due to the filter. The transmitter wasstill identifiable.

As the bandwidth decreases, noise isreduced, thereby improving the S/N ratio.This may result in a loss of information con-tained in signals, and in significant distortionof amplitude waveforms. On the other hand,spectrogram patterns are not distorted, butlack only the above cutoff value. Otherwise,the patterns are the same as the original pat-terns. Therefore, even under conditions inwhich noise has a large effect, the processingof signals using a low-pass filter to a narrowerbandwidth should be effective in the identifi-cation of transmitters.

Journal of the Communications Research Laboratory Vol.49 No.1 2002

Calculated spectrogram patterns (2)for transmitter Tr39

Fig.22

Calculated spectrogram patterns fortransmitter Tr32

Fig.23

Page 13: Transmitter Identification Experimental Techniques and Results · Transmitter Identification ... A 145-MHz FM transceiver, TR51 of Manu-facturer i, was used under conditions in which

149

6 Wigner distribution

We first applied Wigner distributions toidentify transmitters, as the use of fast Fouriertransformation (FFT) provides good resolu-tions of frequency and time. In this report, weapplied spectrogram patterns using FFT.Although a Wigner distribution features a res-olution of frequency twice that of a spectro-gram pattern, no significant difference wasreported in previous studies between calcula-tions of different sources[6]. Although the dis-tributions produce good instant frequencyexpression over time, they are directly affect-ed by noise and interference. Thus, we decid-ed to apply spectrogram patterns using FFT toidentify transmitters. We found that the selec-tion of an adequate data length enabled suchidentification without lowering the resolutionsof frequency and time.

Fig.26 shows a spectrogram calculatedwith a data length of Nfft of 128 bits. The res-olutions of frequency and time in this spectro-gram are approximately one-eighth and eighttimes, respectively, of those in a spectrogramobtained with a data length of Nfft of 1024bits. These large differences resulted in a lossof smoothness in the entire region containinglarge frequency fluctuations. By changing thedata length, we found a length of Nfft of 1024bits to be most adequate in the analysis ofamplitude waveforms. We therefore decidedto use a data length of Nfft of 1024 bits in thisstudy.

Fig.27 shows a spectrogram pattern for alogarithmic scale. Fig.28 shows the Wignerdistribution of the same data for a logarithmicscale. These figures indicate that effects ofboth noise and the spectrum width differedbetween the two types of calculations. TheWigner distribution was better in terms of thefrequency resolution, was influenced more bynoises, and had a narrower spectrum band-width. The basic patterns are the same in thetwo modes, making them suitable for the iden-tification of transmitters. For details on theprocedure and evaluation of these analyticaltechniques, refer to a technical manual.

Tsutomu SUGIYAMA et al.

Spectrogram pattern obtained byapplying a low-pass filter with an IFbandwidth of 50 kHz using transmitterTr39

Fig.24

Spectrogram pattern obtained byapplying a low-pass filter with an IFbandwidth of 12 kHz using transmitterTr39

Fig.25

Page 14: Transmitter Identification Experimental Techniques and Results · Transmitter Identification ... A 145-MHz FM transceiver, TR51 of Manu-facturer i, was used under conditions in which

150

7 Conclusions

Using a high-speed data-acquisition sys-tem with a receiving function, we examinedthe effects of the receiver input-level conver-sion (Pin) on envelope waveforms and spec-trogram patterns at the time of a time-domainrise. As the Pin lowered, the S/N ratio alsolowered due to the external noise. This lower-ing was compared with the results obtained inan indoor experiment in which there was nointerference with neighboring radio stations.It was found that the waveforms in the timedomain were too distorted for the identifica-tion of transmitters. On the other hand, it wasfound that original signals, noise, and interfer-ence waves were separated in spectrograms.Furthermore, the control of a bandwidth usinga filter had no effects on the portions of thespectrogram pattern outside the cutoff timerange. Thus, spectrogram patterns proved tobe effective in identifying transmitters. Itshould be noted that they were capable of

Journal of the Communications Research Laboratory Vol.49 No.1 2002

Amplified waveform and spectro-gram pattern calculated with a datalength of Nfft of 128 bits

Fig.26

Amplified waveform and spectro-gram pattern calculated with a datalength of Nfft of 1024 bits

Fig.27

Amplified waveform and spectro-gram pattern obtained using a Wign-er distribution

Fig.28

Page 15: Transmitter Identification Experimental Techniques and Results · Transmitter Identification ... A 145-MHz FM transceiver, TR51 of Manu-facturer i, was used under conditions in which

151

identifying types of transmitters, but that indi-vidual transmitters of the same type were notalways identifiable because some of themhave almost the same spectrogram patterns.This latter fact is a problem that remains to besolved in the future.

Acknowledgements

The present study was conducted with thefinancial support of the Ministry of Posts andTelecommunications (currently the Ministryof Public Management, Home Affairs, andPosts and Telecommunications). We would

like to express our gratitude for the assistanceprovided by those at the Ministry and theCommunications Research Laboratory. Wewould also like to express our thanks to Mr.Yasuo Suzuki and Mr. Yoshitada Kato of Agi-lent Technologies for their help in the devel-opment of the data-acquisition system. Inaddition, we wish to thank Mr. Chihiro Miki,Senior Researcher, for his help with the appli-cation of a radio station. Finally, thanks toMr. Shunkichi Isobe, Leader of the ScientificTechnology Information Group, for his valu-able advice prior to publication.

Tsutomu SUGIYAMA et al.

References1 Radio Use Website, http://www.tele.soumu.go.jp/e/

2 Y.Ichino, A.Suzuki, T.Sugiyama and M.Kamata, "Application of Wigner-Ville Distribution of Radio

Equipment Identification", IEICE Trans. B- , Vol.J77-B- , No.10, pp.584-586, Oct. 1994 (in

Japanese).

3 W.Shiobara, N.Ojima, R.Chino and T.Takahashi, "Transient variation in the transmitter frequency of a

mobile FM transmitter when its press-to-talk switch is operated", Review of the Radio Research

Laboratory, Vol.17, No.90, pp.281-286, May 1971 (in Japanese).

4 T.Sugiyama, M.Shibuki, T.Hirano, K.Iwasaki, "Data Acquisition System and Radio Transmitter Rising

Envelope on Radio Transmitter Identification", Proc. IEICE General Conf., B-4-23, Mar. 2000 (in

Japanese).

5 T.Hirano, T.Sugiyama, M.Shibuki and K.Iwasaki, "Study on Time-Frequency Spectrum Pattern for

Radio Transmitter Identification", Proc. IEICE General Conf., B-4-24, Mar. 2000 (in Japanese).

6 K.Iwasaki, T.Hirano, M.Shibuki and T.Sugiyama, "A Data Analysis Method for Radio Transmitter

Identification Based on Transient Response -Root-MUSIC Method-", Research discourse meeting,

May 1999 (in Japanese).

Page 16: Transmitter Identification Experimental Techniques and Results · Transmitter Identification ... A 145-MHz FM transceiver, TR51 of Manu-facturer i, was used under conditions in which

152 Journal of the Communications Research Laboratory Vol.49 No.1 2002

The symbols given in the Remarks section indicate the names used in this report.A circle indicates a transmitter tested in the field experiment.

Page 17: Transmitter Identification Experimental Techniques and Results · Transmitter Identification ... A 145-MHz FM transceiver, TR51 of Manu-facturer i, was used under conditions in which

153Tsutomu SUGIYAMA et al.

Tsutomu SUGIYAMA

Researcher, Radio and MeasurementTechnology Group, Applied Researchand Standards Division

Development of type approval test

Masaaki SHIBUKI

Senior Researcher, Radio and Meas-urement Technology Group, AppliedResearch and Standards Division

Standard Time and Frequency

Ken IWASAKI

Senior Researcher, Radio and Meas-urement Technology Group, AppliedResearch and Standards Division

Mobile Communications

Takayuki HIRANO

STA fellowship