HIGH RESOLUTION ANALYSIS OF EIGENRAY GAIN PERTURBATIONS IN ULTRA-SHALLOW WATER S.M. SIMMONS, O.R. HINTON, A.E. ADAMS, B.S. SHARIF AND J.A. NEASHAM Department of Electrical and Electronic Engineering Merz Court, University of Newcastle upon Tyne, Newcastle upon Tyne, NE3 1DQ,, UK E-mail: [email protected] This paper presents results obtained from the acoustic analysis of an ultra-shallow channel of around 40 m in the North Sea over ranges of 0.9 km and 3.0 km. It is shown how the channel is characterised by severe multipath propagation and signal fading. A low frequency phenomenon is described that was apparent in all of the arrival paths to the receiver array over these distances. The article then demonstrates how these frequencies could arise as a result of an oscillation of the transmitter, possibly related to the swell of the channel. 1 Introduction The results and analysis presented in this paper are taken from the data obtained during the first LOTUS (Long range Telemetry in Ultra-Shallow channels) project sea trials held in the summer of 1999 [1]. The purpose of this project was to develop a system of digital communication between deployed underwater units. These units could either be mounted on the sea floor, free floating or attached to an un-tethered underwater vehicle. It was envisaged that the distance between the nodes would be far greater than the depth of the water column. The sea trials in 1999 were carried out in the North Sea off the North-East Coast of England over a period of five days. The average depth of the sea was around 40 metres for the chosen deployments, with transmission distances of 0.8–10 km. The shallow depth of the water combined with the medium to long ranges produced hostile multipath effects as well as variable direct paths due to the seasonal conditions. On each day of the trials, two transmitter units were deployed at different locations. The transmitter hydrophones were approximately 5 m above the sea floor and transmitted autonomously throughout the day. The receiver structure consisted of an array of 8 hydrophones. The bottom element of the array was 1 m above the sea floor, with an equal vertical spacing of 3 m between the remaining single hydrophones and a central horizontal structure of four of the hydrophones. The data was recorded on one of the support vessels that had a link to the receiver array, which was deployed at the beginning of the experiment. A representation of the transmitter and receiver structures is shown in Fig. 1. 303 N.G. Pace and F.B. Jensen (eds.), Impact of Littoral Environmental Variability on Acoustic Predictions and Sonar Performance, 303-310. © 2002 Kluwer Academic Publishers. Printed in the Netherlands. 304 S.M. SIMMONS ET AL. Command Link (not used) Receiver Transmitters Figure 1. Transmitter and receiver structures. 2 Chirp analysis During each day of the trials the two transmitters generated a wide range of data formats to assess the multi-user capability of the communications network. At the start of each day a chirp signal was transmitted for around half an hour. This was to provide data from which the statistical properties of the fading, multipath channel could be analysed. The chirp signal was designed to enable a high-resolution analysis of the channel and also to ensure that the current channel response is well separated from the echoes of the previous transmissions. The frequency of the chirp was 8–12 kHz over a period of 12.5 ms throughout the trials. The repetition rate of the chirp was 25 ms on the first two days (0.9 km and 3 km range) and 50 ms on the final three days (3 km, 5 km and 8 km range). The signal was transmitted at a level of 186 dB (re 1 µPa at 1m) and sampled at 48 kHz. Figure 2(a) shows the received signal at the bottom hydrophone element over a period of 100ms on the first day of the trials transmitted over a range of 0.9 km and Fig. 1(b) shows the signal after cross-correlation with the chirp. 0.1 8 0.08 6 0.06 4 Correlation Amplitude Signal Amplitude 0.04 0.02 0 -0.02 -0.04 2 0 -2 -4 -0.06 -6 -0.08 -0.1 0 0.02 0.04 0.06 Time (s) (a) 0.08 0.1 -8 0 0.02 0.04 0.06 0.08 0.1 Time (s) (b) Figure 2. (a) Received signal and (b) cross-correlation with the chirp. It was assumed that the peak of the cross-correlation in the periodicity of the chirp represents the direct path gain at that instant. Hence by taking the values of the maximum 305 ANALYSIS OF EIGENRAY GAIN PERTURBATIONS peaks it is possible to infer the variation of the direct path gain with time. The sampling period is the inverse of the chirp repetition rate; giving a high resolution of 40 Hz. Figure 3(a) shows the direct path correlation amplitude over a period of 25 s for the bottom receiver element. Low frequency periodic behaviour at around 10 s can be observed in the direct path gain. Also present is higher frequency periodic behaviour, which was analysed using the Wigner-Ville transform with a Choi kernel. The transformation is shown in Fig. 3(b) corresponding to the direct path gain between 3 s and 15 s from Fig. 3(a) after filtering of the low frequency oscillations. The periodicity varies approximately sinusoidally between 2 Hz and 10 Hz. The period of this change in frequency correlates with the low frequency oscillation of the direct path gain of around 10 s periodicity. During the dry tests prior to the sea trials, oscillations were observed in the peak of the correlation due to the difference in the transmitter and receiver clock crystals. It is this phenomenon that is responsible for the oscillations in the frequency band 2–10 Hz shown in Fig. 3. 9.5 12 10 8.5 8 8 Time (s) Direct Path Correlation Amplitude 9 7.5 4 7 2 6.5 6 0 6 5 10 15 Time (s) (a) 20 25 1 2 3 4 5 6 Frequency (Hz) 7 8 9 10 (b) Figure 3. (a) Direct path gain of the bottom element of the receiver array and (b) the Wigner-Ville time-frequency transformation of the gain. The impulse response estimates were filtered above 2 Hz and squared to separate the effects induced the channel properties from the artefact effects of the difference in the two crystal oscillator frequencies. Figure 4(a) shows a 100 ms series of the filtered chirps. The impulse responses were then time aligned, assuming that the transmitter and receiver crystals were operating at exactly 48 kHz. Figure 4(b) shows a contour plot of a section of the aligned responses over a period of 12.5 s for the lower receiver element. It can be seen from the plot that the responses are misaligned due to the differences in the sampling frequencies. Further to the gradient of the peak values, an oscillation can be observed with a period of around 10 s. The oscillation of the position of the peak values was extracted and is shown in Fig. 5. This oscillation appears to be a variation in the time of arrival of the direct path. As can be seen from Fig. 4(b) this oscillation also occurs in the multipath arrivals. Although not shown here, the oscillations were also present in the responses obtained from the other 7 elements of the array. 306 S.M. SIMMONS ET AL. 0.8 Normalized Impulse Response 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.02 0.04 0.06 0.08 0.1 Time (s) (a) (b) Figure 4. (a) Normalized impulse responses after squaring and filtering and (b) a contour plot of the responses after time alignment for the bottom receiver element. 0.2 Variation in Time of Arrival (ms) 0.15 0.1 0.05 0 -0.05 -0.1 -0.15 -0.2 0 20 40 60 80 Time (s) 100 120 Figure 5. The apparent variation of the time of arrival of the lowest receiver element direct path. The gradient of the time of arrival shown in Fig. 4(b) corresponds to a difference in the transmitter and receiver crystal oscillators of just under 5 Hz. The combination of this frequency with the variation in the time of arrival shown in Fig. 5 accounts for the variation in the artefact frequency shown in Fig. 3(b). The plots of Figs. 6(a) and 6(b) show the variation in the direct path for the bottom receiver element for a period of over 2 minutes at a range of 0.9 km and 3.0 km respectively. At the shorter range an oscillation in the amplitude can be observed with a period of around 10 s. At the longer range there still appears to be low frequency oscillations, although not as regular and pronounced as at the shorter range. Figures 7(a) and 7(b) show the power spectra of the direct path gain at a range of 0.9 km and 3.0 km respectively, estimated using the data shown in Fig. 6. The spectrum shows the presence of the 0.1 Hz period as well as what appears to be a second harmonic at 0.2 Hz for the shorter range. The power spectrum at the larger range displays similar behaviour in terms of low frequency components, although at a slightly lower frequency and lower power due to the greater attenuation. 307 ANALYSIS OF EIGENRAY GAIN PERTURBATIONS 55 2.5 Direct Path Correlation Amplitude Direct Path Correlation Amplitude 50 45 40 35 30 2 1.5 1 0.5 25 20 0 20 40 60 80 Time (s) 100 0 0 120 20 40 60 80 Time (s) (a) 100 120 (b) 300 0.06 250 0.05 Power Spectrum Amplitude Power Spectrum Amplitude Figure 6. Direct path amplitude variation over 130 s for the bottom receiver element at a range of (a) 0.9 km and (b) 3.0 km. 200 150 100 50 0 -0.8 0.04 0.03 0.02 0.01 -0.6 -0.4 -0.2 0 0.2 Frequency (Hz) 0.4 0.6 0.8 0 -0.8 -0.6 -0.4 -0.2 0 0.2 Frequency (Hz) (a) 0.4 0.6 0.8 (b) Figure 7. The power spectrum of the direct path for ranges of (a) 0.9 km and (b) 3.0 km. 200 100 180 90 160 Power Spectrum Amplitude 80 140 70 120 60 100 50 40 80 30 60 20 40 10 20 0 -0.8 -0.6 -0.4 -0.2 0 0.2 Frequency (Hz) (a) 0.4 0.6 0.8 0 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 (b) Figure 8. (a) The Fourier transform of the cross-covariance of (a) the direct path at the bottom and top receivers and (b) the direct path and second dominant arrival at the bottom element. Figure 8(a) shows the Fourier transforms of the cross-covariance of the direct path at the bottom and top receivers at a range of 0.9 km. The plot is very similar to the power 308 S.M. SIMMONS ET AL. spectrum of Fig. 7(a). Hence there is a high degree of correlation between the received signals at the different elements due to the large wavelength. The same correlation can be seen in the Fourier transform of the cross-covariance of the direct path at the bottom receiver with the second significant arrival. All of the direct paths and later arrivals were found to be highly correlated for all of the arrays. Figure 9 shows the evolution of the multipath signals over a period of 500 s for four of the receiver elements. The contour plots were obtained by aligning the direct path at each element. The direct path appears at around a 2 ms (aribitrary) delay from the beginning of the sequence. This direct path is not visible on Fig. 9(b) as the amplitude is too small to show up on the plot. Further to the direct paths there are two distinguishable arrivals visible. The delay of the first of these arrivals decreases as the height above the sea floor increases, indicating that it is a surface reflection. The delay of the second dominant arrival after the direct path behaves in the opposite manner, indicating it has been reflected off the sea floor. Closer inspection of Fig. 9(d) for the top element shows the presence of an arrival between the direct path and the first dominant arrival from the other plots. This arrival has been reflected from the sea floor and has been lost on the other plots as it is too close to the direct path and the resolution is not fine enough. The most noticeable feature of the plots shown in Fig. 9 is the large fading of the later arrivals over long periods. 500 500 0.8 450 0.8 450 0.7 400 0.7 400 0.6 0.6 350 350 0.5 250 0.4 200 0.3 150 0.5 300 Time (s) Time (s) 300 250 0.4 200 0.3 150 0.2 100 0.2 100 0.1 50 1 2 3 4 5 Delay (ms) 6 7 8 0.1 50 1 2 3 4 5 Delay (ms) (a) (b) (c) (d) 6 7 8 Figure 9. Evolution of the multipath propagation paths over 500 s for receiver elements at (a) 1 m, (b) 4 m, (c) 8 m and (d) 14 m above the sea floor. 309 ANALYSIS OF EIGENRAY GAIN PERTURBATIONS Figure 10 shows the direct path over 500 s for the bottom elements at a range of 0.9 km and 3.0 km. The amplitude of the direct path tended to remain fairly constant for each element at the shorter range and varied substantially at the longer range. The amplitudes at the different elements however tended to vary a great deal from one another. The large degree of fading at the 3.0 km range suggests that the arrival is in fact a reflection, as it exhibits similar fading behaviour to the reflected paths at 0.9 km. It is possible for no direct path exists between the transmitter and the receiver at the time of year of the trials due to the downward refraction caused by the sound speed profile. However, this is not verifiable as no sound speed recordings were taken. 60 3.5 3 Correlation Peak Amplitude Correlation Peak Amplitude 50 40 30 20 10 0 0 2.5 2 1.5 1 0.5 50 100 150 200 Time (s) 250 300 350 0 0 50 100 150 (a) 200 Time (s) 250 300 350 (b) Figure 10. Variation of the direct path gain for the lowest receiver element over 375 s for (a) day 1 of the trials at 0.9 km range and (b) day 2 at 3.0 km range. 3 Discussion Figure 11 shows the normalised power obtained from a simulation of the correlation of a signal consisting of two chirps with different delays. It can be seen that for certain differences in delays between the two chirps a considerable amount of signal fading occurs. The maximum fading occurs when there is a lag of about 0.05 ms corresponding to an underwater path difference of around 70 mm. Such a delay corresponds to the geometry of the transmitter and receiver arrays used in the trials. Hence it is thought that the fading seen in the direct path of Fig. 9(b) is attributable to this phenomenon. 1 0.9 0.8 Normalized Power 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.2 0.4 0.6 Delay Difference (ms) 0.8 1 Figure 11. The normalized peak correlation amplitude of the sum of two chirps with different delays. 310 S.M. SIMMONS ET AL. Low frequency oscillations appear in the amplitude of all of the arrivals for both ranges examined in this paper. The frequency of the oscillations remains virtually constant at the range of 0.9 km and less so at the longer range of 3.0 km. The period of the major component of the oscillations correlates with what appears to be a variation in the travel time delay of all of the paths. Possible mechanisms for the generation of the travel time delay perturbations include the crystal oscillator behaviour, a change in the sound speed and travel path by mechanisms such as the swell and finally an oscillation of either the transmitter and/or the receiver array. The sea state at the time of the trials was Beaufort state 1 with a 1 to 1.5 m swell having a 10 s to 15 s period. Hence the variation in the travel time could be associated with the swell. One possible mechanism for the generation of the travel-time delays is an oscillation of the transmitter. A peak-peak oscillation of around 0.3 m of the transmitter would produce travel time delays equivalent to those observed. It is unlikely the receiver is oscillating, as the frequency shift of the artefact frequency was the same for all of the elements [2]. The effect of the increase of depth due to the swell is not enough to account for a sufficient change in sound speed on its own. The most overriding piece of evidence is the correlation between the travel-time delay and the amplitude oscillations in the paths. As demonstrated in Fig. 11 a small change in the time delay between the arrivals of two chirps can result in large fading. If all of the travel-time delays oscillated with exactly the same amplitude then no amplitude oscillations would occur. Hence it is necessary for at least one of the paths to have a travel-time oscillation that is different in magnitude to the others. This difference in the time of arrival variation magnitude is possible from the geometry of the system of the trials, especially if it was correlated with the swell (resulting in a slightly greater oscillation amplitude for surface reflections). If the crystal oscillators had frequency variations that create an impression of a variation in the traveltime delay then there would be no change in the amplitude of the paths as all the delays would be equal. 4 Conclusions It has been shown from experimental data that severe multipath and fading occur in the channel used in the LOTUS sea trials for the chirp signals used to infer the channel properties. It is postulated that the fading of the chirp combined with a low frequency oscillation of the transmitter unit, possibly related to the swell, creates the low frequency phenomena observed in the arrival paths. References 1. Adams, A.E., Hinton, O.R., Sharif, B.S., Salles, G., Orr, N. and Tsiminedis, C., An experiment in sub-sea networks – The LOTUS sea trials. In Proc. 5th European Conference on Underwater Acoustics, Lyon, France, 10–13 July, 2000. 2. Salles, G., Adams, A.E., Hinton, O.R., Sharif, B.S., Orr, N. and Tsiminedis, C., High resolution analysis of ultra-shallow water acoustic travel time and gain perturbations. In Proc. Oceans'2000, Rhode Island, USA, 17–20 Sep., 2000.
© Copyright 2025 Paperzz