An Adaptive Narrowband Interference Excision Filter with Low Signal Loss for GPS Receivers

Transcription

1 ICCAS5 An Adative Narrowband Filter with Low Signal Loss for GPS s Mi-Young Shin*, Chansik Park +, Ho-Keun Lee #, Dae-Yearl Lee #, and Sang-Jeong Lee ** * Deartment of Electronics Engineering, Chungnam National University, Daejeon, Korea (Tel : ; snyh@cslab.cnu.ac.kr) + Deartment of Electrical and Comuter Engineering, Chungbuk National University, Daejeon, Korea (Tel : ; cbucc@chungbuk.ac.kr) # Agency for Defense Develoment, Daejeon, Korea (Tel : ; ldy31@yahoo.co.kr) ** Deartment of Electrical and Comuter Engineering, Chungnam National University, Daejeon, Korea (Tel : ; eesjl@cslab.cnu.ac.kr) Abstract: As the low ower GPS signal is suscetible to interference, interference can seriously degrade the erformance of GPS. This aer designs a ANIEF(Adative Narrowband in domai filter that removes narrow band interferences with low signal loss. This filter uses the re-correlation technique and attemts to filter out the interference in the frequency domain. The interference excision erformance of the designed filter is evaluated for various interferences using the ANIEF filter inserted GPS software and the interference generator. s considered in this aer are single-tone CWI(Continuous Wave ), multi-tones CWI, ulsed CWI, and swet CWI. The narrowband interference excision filter in frequency domain is very effective against various interferences and the strong interference with a simle structure. However, the signal ower loss is unavoidable while transforming. In this aer, the hamming window and overla technique are adoted to reduce the signal ower loss. Finally, the interference excision erformance and the reduced signal ower loss of the ANIEF filter are shown. Keywords: suression, Adative narrowband filter, -domain excision, Anti-jamming 1. INTRODUCTION The low ower GPS signal is suscetible to the interference. Therefore, interference can seriously degrade the erformance of GPS. excision techniques can be classified into the time domain, frequency domain and amlitude domain technique. This aer designs the ANIEF(Adative Narrowband in domai filter that removes narrow band interference from the GPS sectrum. This filter attemts to filter out the interference before correlation in the GPS. The interference excision technique in the frequency domain transforms the received GPS signal containing interference into the signals in the frequency domain and comutes the signal statistically to determine an excision threshold. And then, it removes all sectrums exceeding the threshold, and restores the remaining signal without interference to the signals in the time domain. This aer rooses the ANIEF filter scheme that adots the hamming window and 5 % overla technique to reduce the signal ower loss which is unavoidable while transforming. To simulate the erformance of the roosed filter, the GPS IF signal generator and the software are imlemented. Section summarizes effects of RF interference on GPS s. Section 3 describes the structure of the roosed the interference excision filter. Section 4 shows some interference excision erformance results. Finally, summary and concluding remarks will be given in section 5.. EFFECTS AND MEASURES OF RF INTERFERENCE The interference can result in degraded navigation accuracy or comlete loss of tracking. Therefore, the requires countermeasure for removing imact of interference. Table 1 summarizes various tyes of RF interference[1]. Table 1 Tye of RF and Tyical Sources Tye Wideband-Gaussian Wideband hase / frequency modulation Wideband-sread sectrum Narrowband-ulse Tyical Sources Intentional noise jammers Television transmitter's harmonics or near-band microwave link transmitters overcoming front-end filter of GPS Intentional sread sectrum jammers or near-field of seudolites Radar transmitters Narrowband hase AM stations transmitter's harmonics or / frequency modulation CB transmitter's harmonics Narrowband-swet continuous wave Intentional CW jammers or FM stations transmitter's harmonics Narrowband-continuous Intentional CW jammers or near-band wave unmodulated transmitter's carriers Fig. 1 shows a basic GPS architecture with the interference effects on the. The interference can result in degraded navigation accuracy or comlete loss of tracking[]. Fig. shows all otential mitigation techniques which can be alied to the GPS would occur.

2 ICCAS5 Adative Antenna Harmonics AGC Degradation of the Quantification N 1 ( ) = X k x( e j π kn / N, k =,1,..., N 1 () Pre Amlifier and Filter Saturation and Intermodulation Conversion Oscillator ADC DSP Digital Channels Saturation, Over-Shot and Harmonics Loss of DDLL Loss of DPLL If the magnitude of sectrum of each frequency bin is larger than the magnitude of AWGN, it can be regarded as the interference and will be rejected. Because the ower of GPS signal is much smaller than that of AWGN, the GPS signal is not likely to be rejected. The remaining signal is restored in the time domain through I (see Eq. (3)). Adative Antenna Prefiltering BER Degradation Fig. 1 Effects on a GPS Pre Amlifier and Filter External Aid(INS. etc) Limiter and Filter Conversion Oscillator External Equiment AGC ADC DSP Digital Channels Jammer Detection Adative Coding Cycle Slis Pre-Correlation DSP Techniques Post-Correlation Mitigation Techniques Fig. Position of the Mitigation Techniques in a GPS The ANIEF filter belongs to the re-correlation technique and attemts to filter out the interferences in the frequency domain. 3. ANIEF FILTER STRUCTURE AND INTERFERENCE EXCISION TECHNIQUE The structure of ANIEF is shown in Fig. 3. x ( X (k ) x ( I Fig. 3 The structure of ANIEF Filter The received GPS signal is transformed to the frequency domain one by the (Fast Fourier Transform) of 56-samles and analyzed by the block to determine an excision threshold. The interference excision block removes all sectrums exceeding the excision threshold. The received GPS signal, x ( is comosed of the transmitted GPS signal, the AWGN(Adative White Gaussian Noise) and the interference (see Eq. (1)). x ( s( + w( + j(, n =,1,..., N 1 (1) = In equation, s ( is the transmitted GPS signal, w ( is AWGN, and j ( is interference signal. The outut of N samles, X (k), is given in Eq. ()[3][4]. N 1 1 ( ) = x n X ( k) e N j π kn / N, k =,1,..., N 1 (3) The rocessing without window brings the sectral leakage[3]-[5]. ing smooths the discontinuities at the block boundary and lessens the effect of sectral leakage. The signal energy will be sread across the sectrum roortional to the width of the main-lobe and the height of the side-lobes of the window. Selecting a window with lower side-lobes will reduce the amount of sectral leakage. However, a window with lower side-lobes usually has a wider main-lobe. For the GPS alication, the objective is to minimize the frequency sreading of each CW tone in order to minimize the number of frequency bins that will be excised[6]. At the same time, it is also required to minimize the degradation of the GPS signal when the interference is not resent. The window selection requires a tradeoff between the reduction in SNR due to the signal attenuation incurred by multilying the data sequence by a window and the effectiveness of the sectral containment for a CW tone. The degradation of the SNR can be exressed as Eq.(4). The SNR loss may exceeds high as 3dB for some window functions with extremely low side-lobes. N 1 w( SNR degradation = N 1 N w (, w ( window (4) In Table, the characteristics of Rectangular window, Blackman-Harris window, window are comared [3][4]. Table Weighting Characteristics in Analysis Worst Case Side-lobe Processing Height Containment Loss Rectangular -13dB 3.9dB 5 bins Blackman-Harris -9dB 3.47dB 6 bins -43dB 3.1dB 4 bins The most filters with O(Overla Fast Fourier Transform) structure use Blackman-Harris window. This aer will use the window since it has the smallest the frequency containment and the least rocessing loss are shown in table. The window can be exressed as Eq. (5). w ( =.54.46[cos(π n /( N 1))]( n < N) (5)

3 ICCAS5 In this aer, the hamming window and 5% overla technique are adoted to reduce the signal ower loss. 5% overla rocessing is of comlex structure, but it reduces the effect of the signal attenuation from the window on the outut SNR [6][7]. The SNR loss is reduced from.6db to 3dB using the hamming window and overla technology. Each ath in the rocessing chain roduces one-half of the usable outut sequence as shown in Fig. 4. This figure shows the contribution of each data ath to the overall result. CW interferences occuy relatively few frequency bins, and their amlitudes are above the noise floor[6][8]. Therefore, interference excision block determines the excision threshold and then removes all sectrums exceeding the excision threshold every 56 samles. The block firstly transforms the magnitude of the outut into the magnitude in decibels for each frequency bin to determine the excision threshold which is the boundary between the interference and the AWGN. Then, it calculates the standard deviation ( σn block ) and mean ( µ n block ) of the magnitudes of 56 samles. The standard deviation and the mean are arameters that determine the excision threshold. The excision threshold can be exressed as Eq. (6). Threshold = N (6) µ + σ Where N is to be chosen are one of five reference oints considering the mean and the standard deviation. The larger standard deviation indicates the resence of interferers and results in a smaller scale factor N to maintain the threshold at the level of the noise floor. If the magnitude of a samle is larger than this threshold, then the magnitudes of 4-samles is set to zero. The 4-samles is chosen by the characteristics of the hamming window The functional descrition of the block is shown Fig. 6. mag Fig. 4 5% Overla Processing The overla function unfortunately doubles the rocessing since it requires a second rocessing ath that includes the,,, and I, as shown in Fig. 5. x( 1log1 ( mag) µ & σ ( µ + N σ Threshold Select ) Delay 1/ block 1log1( mag ) > Threshold NO YES 4 frequency bins return() return(mag) Fig. 6 Functional Descrition of the Block 4. PERFORMANCE EVALUATION I Mux I x ( Fig. 5 Block Diagram of the Narrow band Filter This section shows simulation results using the test environment shown in Fig. 7. The simulated GPS signal consists of IF GPS signal, AWGN and interference. The GPS signal generator oututs the signal with 1.45MHz center frequency which is samled by 5.714MHz and quantized for 1bits. The outut signal is alied to the GPS software. The ANIEF filter has been inserted in front of the correlator in the software GPS.

4 ICCAS5 s considered in the test are single-tone CWI, multi-tones CWI, ulsed CWI and swet CWI. Against these interferences, it is checked whether the GPS with ANIEF can acquire the GPS signal. In additional, the maximum J/S is also tested. Condition 1 : Lock indicator condition ( I + Q ) Condition : Loss of lock threshold condition (3dB above the noise floor) GPS Relica Signal Generator The signal sectrum of single-tone CWI in 9kHz bandwidth is shown in Fig 8. Then the result that the single-tone CWI is rejected after ANIEF filtering is shown in Fig. 9. At the frequency domain aears ower distribution of signal. Therefore it is easy to detect the CWI. So the ANIEF filter will be shown high erformance that excises single-tone CWI and multi-tones CWI of high ower level Before the 14 S( W( J( 1 db 1 x( Fs=5.714MHz Samling 1bit Quantization Software (Hz) x 1 6 Fig. 8 The ower sectral density before the ANIEF filter Narrow band Filter 18 After the x'( db 1 Monitor Dislay ( result) Fig. 7 The test environment setu for simulation First check u on SNR loss after filter is inserted in software GPS. Table 3 is result that comare to outut SNR after rocessing that rocess the simulative GPS signal not included interference with the without the ANIEF filter, with the ANIEF filter excet the overla rocessing and with the ANIEF filter. SNR loss is 3.4dB when the signal transforms frequency domain for the interference excision. At this time gain is.3db which use 5% overla for minimize SNR loss. Table 4 comare to SNR loss of software for various windows to reduce signal loss by window function. Because of the result in Table 4, the ANIEF filter structure used the hamming window that it has small SNR loss. Table 3 The SNR loss due to ANIEF No the The filter excet filter overla inserted inserted The filter inserted SNR 14.6dB 11.dB 13.5dB Table 4 The SNR loss due to window Rectangular Blackman-Harris SNR 5.8dB 6.9dB 7.dB (Hz) x 1 6 Fig. 9 The ower sectral density after the ANIEF filter This aer evaluates the filter erformance against interference ower level of single-tone CWI, multi-tones CWI, ulsed CWI and swet CWI. The acquisition of signal to evaluate the filter erformance is determined according to conditions. The interference removal erformance evaluates by comaring maximum J/S. Table. 5 defines arameter of interference used this test. Table 5 s of the interferences. Duration Single-tone CWI Multi-tones CWI Pulsed CWI Swet CWI.9MHz -.5MHz, 1MHz, 1.5MHz 1.5MHz 1Hz ~ 1MHz - 4,samles : On 1,samles : Off Increase 1Hz every 1samles

5 ICCAS5 Fig are an examle to evaluate filter erformance when the single-tone CWI(J/S=3.837dB) is inutted like Fig Some ower loss is shown in Fig. 11, but the acquisition of signal is ossible. Above the test checked u the effective erformance about various interferences of the ANIEF filter. 5. SUMMARY AND CONCLUSION I +Q 16 x 15 < J/S=3.837dB, Single-Tone CWI > This aer shows that the erformance of designed anti-jamming filter is excellent against CW interferences and that such time-variant interferences as ulsed CWI and swet CWI are also excised efficiently. Although the erformance of the designed filter varies a little bit according to the tye of interference, the designed filter excises all of the interferences with the minimum J/S of db so that the signal acquisition and tracking are maintained in site of the resence of such interferences. Furthermore, the designed filter reduces the signal ower loss of at least.3db using overla technique with hamming window. REFERENCES samle x 1 4 Fig. 1 The software GPS without the ANIEF filter I +Q 16 x 15 < J/S=3.837dB, Single-Tone CWI > samle x 1 4 Fig. 11 The software GPS with the ANIEF filter Fig. 1 comares erformance of filter about various interferences. This is the maximum J/S to be able to excise the interference in the ANIEF filter inserted and the ANIEF filter not inserted. Fig. 1 can check u high the excision erformance of the narrowband interference in frequency domain. [1] Elliott D. Kalan, Understanding GPS Princiles and Alications, , [] Rene Jr. Landry, Vincent Calmettes, Alain Ducasse, Imact of interference on the new COSSAP GPS and mitigation techniques evaluation,, ION National Technical Meeting, Long Beach, CA, 1-3 Janvier, [3] Alan V. Oenheim, Ronald W. Schafer, Discrete-Time Signal Processing, [4] Paul A. Lynn, Wolfgang Fuerst, Introductory Digital Signal Processing With Comuter Alications [5] F. Harris, On the use of windows for harmonic analysis with the discrete Fourier transform, Proc.IEEE, vol.66,.51-83, Jan [6] P. Caozza, T. Hokinson, B. Holland, and R. Landrau, A Single-Chi Narrowband Domain Excisor for a Global Positioning System(GPS), Custom Integrated Circuits Conference 1999, San Diego, CA, May [7] R. Rifkin, J. Vaccaro, Comarison of Narrowband Adative Filter Technologies for GPS, IEEE,. [8] J. Young, Analysis of DFT-based frequency excision algorithms for direct-sequence sread-sectrum communications, IEEE Trans. Commun., vol.46, , Aug Fig. 1 Performance comarison according to the interference