Apparatus for deriving measurement signals from fibre optic distributed sensors

Abstract

The application describes methods and apparatus for distributed fiber sensing, especially distributed acoustic/strain sensing. The method involves launching at least first and second pulse pairs into an optical fiber, the first and second pulse pairs having the same frequency configuration as one another and being generated such that the phase relationship of the pulses of the first pulse pair has a predetermined relative phase difference to the phase relationship of the pulses of the second pulse pair. In one embodiment there is a frequency difference between the pulses in a pulse pair which is related to the launch rate of the pulse pairs. In another embodiment the phase difference between the pulses in a pair is varied between successive launches. In this way an analytic version of the backscatter interference signal can be generated within the baseband of the sensor.

Claims

1. A distributed fibre optic sensor apparatus comprising: an optical source, the optical source generating at least first and second optical pulse pairs for launching into an optical fibre, each pulse having a frequency, wherein the first pulses of the pulse pairs have the same frequency and the second pulses of the pulse pairs have the same frequency, and the pulses generated such that the phase relationship of the pulses of the first pulse pair and the phase relationship of the pulses of the second pulse pair are different by a predetermined relative phase difference wherein: the pulses of the first and second pulse pairs have a frequency difference between them; and the first and second pulse pairs are generated so as to be launched into the optical fibre at different times; such that the time between launch of the first and second pulse pairs corresponds to the time for a relative phase change in a signal at a frequency equal to the frequency difference between the pulses in a pair equal to said predetermined relative phase difference.

2. A distributed fibre optic sensor apparatus as claimed in claim 1 wherein said predetermined relative phase difference has a magnitude of 90.

3. A distributed fibre optic sensor apparatus as claimed in claim 1 further comprising: at least one detector configured to detect any radiation backscattered from said first and second pulse pairs; and processing circuitry coupled to said at least one detector, wherein the processing circuitry is configured to determine a phase value for at least one given section of optical fibre based on the detected backscatter radiation from said first and second pulse pairs.

4. A distributed fibre optic sensor apparatus as claimed in claim 3 wherein the optical source, detector and processing means are configured such that the bandwidth of the optical detection path is sufficient to represent substantially all of the frequency content of the backscattered radiation.

5. A distributed fibre optic sensor apparatus as claimed in claim 3 wherein said optical detection path bandwidth is sufficient to represent substantially all the pulse frequencies of the pulses of a pulse pair.

6. A distributed fibre optic sensor apparatus as claimed in claim 3 wherein the processing circuitry is configured to demodulate independent instantaneous detector samples from said first and second pulse pairs to determine said phase value.

7. A distributed fibre optic sensor apparatus as claimed in claim 3 wherein the processing circuitry determines said phase value using said predetermined relative phase difference.

8. A distributed fibre optic sensor apparatus as claimed in claim 1 wherein the optical source is configured to launch a series of first and second pulse pairs into the optical fibre at a launch rate such that the time between launch of any pulse pairs in said series corresponds to the time for a relative phase change in a signal at a frequency equal to said frequency difference equal to said predetermined relative phase difference.

9. A distributed fibre optic sensor apparatus as claimed in claim 8 wherein said frequency difference is lower than the launch rate.

10. A distributed fibre optic sensor apparatus as claimed in claim 9 wherein said frequency difference is less than half of the launch rate.

11. A distributed fibre optic sensor apparatus as claimed in claim 8 wherein said frequency difference is substantially equal to a quarter of the launch rate such that said predetermined phase difference is substantially 90.

12. A distributed fibre optic sensor apparatus as claimed in claim 8 wherein the optical source is configured to generate the pulse pairs at a launch rate such that there is only one pulse from said series in a sensing part of the optical fibre at any time.

13. A distributed fibre optic sensor apparatus as claimed in claim 1 wherein said frequency difference is 10 kHz or less.

14. A distributed fibre optic sensor apparatus as claimed in claim 1 where said frequency difference is within the baseband of the sensor.

15. A distributed fibre optic sensor apparatus as claimed in claim 1 where said frequency difference is within the Nyquist band of the sensor.

16. A distributed fibre optic sensor apparatus as claimed in claim 1 wherein said optical source comprises at least one laser and a modulator for modulating the light from said laser to produce the first and second pulse pairs.

17. A distributed fibre optic sensor apparatus as claimed in claim 16 wherein said modulator comprises at least one acousto-optic modulator.

18. A distributed fibre optic sensor apparatus as claimed in claim 1, further comprising: at least one detector configured to detect any radiation backscattered from said first and second pulse pairs; and processing circuitry coupled to said at least one detector, wherein the processing circuitry is configured to determine a phase value for at least one given section of optical fibre based on the detected backscatter radiation from said first and second pulse pairs wherein said processing circuitry is configured to: form a signal comprising samples of the backscatter radiation, the samples acquired at substantially the same time after launch of each pulse pair; and demodulate said signal at the frequency difference of the pulses in each pulse pair so to determine a phase value for at least one section of optical fibre.

19. A distributed fibre optic sensor apparatus as claimed in claim 1 wherein: the pulses in the first pulse pair and in the second pulse pair all have the same frequency as one another; and the phase difference between the pulses in the first pulse pair differs from the phase difference between the pulses in the second pulse pair by said predetermined relative phase difference.

20. A distributed fibre optic sensor apparatus as claimed in claim 19, further comprising: at least one detector configured to detect any radiation backscattered from said first and second pulse pairs; and processing circuitry coupled to said at least one detector, wherein the processing circuitry is configured to determine a phase value for at least one given section of optical fibre based on the detected backscatter radiation from said first and second pulse pairs, wherein said predetermined phase difference is substantially 90 and wherein the processing circuitry is configured to use the detected backscatter signals received the same time after launch of the first and second pulse pairs as in-phase and quadrature phase components respectively so as to determine a phase value for at least one section of optical fibre.

21. A distributed fibre optic sensor apparatus as claimed in claim 20 wherein the processing circuitry is configured to low-pass filter the detected backscatter signals received the same time after launch of the first and second pulse pairs to provide In-phase and Quadrature values and to convert said In-phase and Quadrature values to a phase value by rectangular to polar conversion.

22. A distributed fibre optic sensor apparatus as claimed in claim 19 wherein said optical source is configured to continually generate pulse pairs to be launched into the optical fibre at a launch rate wherein each successive pulse pair has a relative phase difference between the pulses of that pair that differs from that of the previous pair by said predetermined phase difference.

23. A distributed fibre optic sensor apparatus as claimed in claim 19 wherein said optical source comprises at least one laser and a at least one electro-optic modulator arranged to modulate the output of said laser so as to produce the first and second pulse pairs.

24. A distributed fibre optic sensor apparatus as claimed in claim 23 wherein said modulator comprises at least one electro-optic modulator.

25. A distributed fibre optic sensor apparatus as claimed in claim 1 wherein the optical source is configured to generate a first series of pulse pairs comprising said first pulse pair and said second pulse pair and a second series of pulse pairs wherein said first and second series of pulse pairs have different polarisation states to one another.

26. A distributed fibre optic sensor apparatus as claimed in 19 wherein the pulses of the first series of pulse pairs have an orthogonal polarisation state to the pulses of the second series of pulse pairs.

27. A distributed fibre optic sensor apparatus as claimed in claim 26 wherein the first series of pulse pairs and the second series of pulse pairs are polarised with orthogonal states to one another.

28. A distributed fibre optic sensor apparatus as claimed in claim 19 wherein said optical source is configured to generate the first and second series of pulse pairs to be launched into the optical fibre such that one pulse pair from both the first series of pulse pairs and the second series of pulse pairs are in a sensing part of the optical fibre at the same time.

29. A distributed fibre optic sensor apparatus as claimed in claim 19 wherein the optical source comprises at least one polarisation active element for generating a difference in polarisation between the first and second series of pulse pairs to be launched into the optical fibre.

30. A distributed fibre optic sensor apparatus as claimed in claim 19 further comprising at least one polarisation active element arranged to direct backscattered light of a first polarisation state to a first detector.

31. A distributed fibre optic sensor apparatus as claimed in claim 30 wherein said first polarisation state is varied over time.

32. A distributed fibre optic sensor apparatus as claimed in claim 30 wherein said at least one polarisation active element is arranged to direct backscattered light of a second polarisation state to a second detector.

33. A distributed fibre optic sensor apparatus as claimed in claim 19 further comprising at least one polarisation modulator for compensating for any polarisation drift within said optical fibre.

34. A distributed fibre optic sensor apparatus as claimed in claim 19 wherein said different polarisation states are such so that backscatter radiation received at the proximal end of the optical fibre from the first series of pulse pairs has an orthogonal polarisation state to backscatter radiation received at the proximal end of the optical fibre from the second series of pulse pairs.

35. A distributed fibre optic sensor apparatus as claimed in claim 1 wherein the first and second pulse pairs comprise a first pulse followed by, and temporally separated from, a second pulse.

36. A distributed fibre optic sensor apparatus as claimed in claim 35 wherein the time between end of the first pulse and start of the second pulse is equal to or greater than the pulse duration of the second pulse.

37. A distributed fibre optic sensor apparatus as claimed in claim 36 wherein the time between end of the first pulse and start of the second pulse is equal to or greater than twice pulse duration of the second pulse.

38. A distributed fibre optic sensor apparatus as claimed in claim 35 wherein the pulse widths of the first and second pulses are relatively narrow compared to the gap between the pulses.

39. A distributed fibre optic sensor apparatus as claimed in claim 35 wherein the first and second pulses have substantially equal durations.

40. A distributed fibre optic sensor apparatus as claimed in claim 35, further comprising: at least one detector configured to detect any radiation backscattered from said first and second pulse pairs; and processing circuitry coupled to said at least one detector, wherein the processing circuitry is configured to determine a phase value for at least one given section of optical fibre based on the detected backscatter radiation from said first and second pulse pairs wherein the detector and processing circuitry are configured so as to acquire a plurality of diversity samples of the backscatter signal from a pulse pair in a duration equal to half the time between the pulses in that pulse pair.

41. A distributed fibre optic sensor apparatus as claimed in claim 40 wherein detector and processing circuitry are configured so as to acquire at least five diversity samples.

42. A distributed fibre optic sensor apparatus as claimed in claim 40 wherein the time between acquiring diversity samples of the backscatter signal is about the same as half the pulse duration.

43. A distributed fibre optic sensor apparatus as claimed in claim 40 wherein the processing circuitry is configured to analyse a plurality of successive diversity samples of the backscatter signal in an analysis bin to determine a resultant phase value for a corresponding section of optical fibre.

44. A distributed fibre optic sensor apparatus as claimed in claim 43 wherein the processing circuitry is configured such that: each analysis bin may comprise a plurality of diversity channels, wherein each diversity channel receive a diversity sample from launch of a pulse pair that corresponds to the same position in the optical fibre; and each diversity channel is processed to determine a phase value for that diversity channel.

45. A distributed fibre optic sensor apparatus as claimed in claim 44 wherein the phase values determined for the diversity channels of an analysis bin are analysed before being combined.

46. A distributed fibre optic sensor apparatus as claimed in claim 45 wherein the processing circuitry is configured to apply a quality metric to the processed phase values and to combine the phase values based on said quality metric.

47. A distributed fibre optic sensor apparatus as claimed in claim 46 wherein said quality metric comprises the degree of similarity between the determined phase values.

48. A distributed fibre optic sensor apparatus as claimed in claim 46 wherein said combination comprises applying a weighting to at least some phase values in said combination.

49. A distributed fibre optic sensor apparatus as claimed in claim 46 wherein said combination comprises omitting some phase values from the combination based on the quality metric.

50. A distributed fibre optic sensor apparatus as claimed in claim 1 wherein said optical source is configured to generate at least one additional pulse sequence having a frequency configuration which is different to that of the first and second pulse pairs.

51. A distributed fibre optic sensor apparatus as claimed in claim 50 wherein the optical source is configured to generate a first series of pulse pairs comprising said first and second pulse pairs and one or more additional series of pulse sequences wherein the first series of pulse pairs and the or more additional series of pulse sequences are generated for launch such that at least one pulse pair of the first series and a pulse sequence of the or more additional series are in the sensing optical fibre at the same time.

52. A distributed fibre optic sensor apparatus as claimed in claim 51 wherein at least one of said additional series of pulse sequences comprises a series of frequency difference pulse pairs, the frequency difference pulse pairs comprising two pulses having a predetermined frequency difference between them, and wherein the frequency difference pulse pairs of a series are generated so as to be launched into the optical fibre at a launch rate such that the time between launches of successive frequency difference pulse pairs of that series corresponds to the time for a predetermined relative phase change in a signal at a frequency equal to said predetermined frequency difference of said series.

53. A distributed fibre optic sensor apparatus as claimed in claim 52, wherein the pulses of the first and second pulse pairs have a frequency difference between them; and the first and second pulse pairs are generated so as to be launched into the optical fibre at different times; such that the time between launch of the first and second pulse pairs corresponds to the time for said predetermined relative phase change in a signal at a frequency equal to the frequency difference between the pulses in a pair and wherein said predetermined frequency difference between the pulses of said at least one additional series is the same as the frequency difference between the pulses of the first and second pulse pairs.

54. A distributed fibre optic sensor apparatus as claimed in claim 53 wherein the launch rates of the pulse pairs of the first series and said at least one additional series are the same.

55. A distributed fibre optic sensor apparatus as claimed in claim 52, wherein the pulses of the first and second pulse pairs have a frequency difference between them; and the first and second pulse pairs are generated so as to be launched into the optical fibre at different times; such that the time between launch of the first and second pulse pairs corresponds to the time for said predetermined relative phase change in a signal at a frequency equal to the frequency difference between the pulses in a pair and wherein said predetermined frequency difference between the pulses of said at least one additional series is different to the frequency difference between the pulses of the first and second pulse pairs.

56. A distributed fibre optic sensor apparatus as claimed in claim 55 wherein the launch rates of the pulse pairs of the first series and said at least one additional series are different and arranged such that said predetermined relative phase change for the first series is the same as said predetermined relative phase change for the additional series.

57. A distributed fibre optic sensor apparatus as claimed in claim 51 wherein the optical source is configured to generate a plurality of additional series of pulse sequences.

58. A distributed fibre optic sensor as claimed in claim 57 wherein two or more of said plurality of additional series comprises series of pulse pairs wherein the pulse pairs of each series have the same pulse configuration in relation to a base wavelength and wherein the base wavelength is different for each series.

59. A distributed fibre optic sensor apparatus as claimed in claim 57 wherein the launches of the pulse pairs of the additional series are staggered from one another at irregular intervals.

60. A distributed fibre optic sensor apparatus as claimed in claim 51 wherein at least one additional series of pulse sequences comprises a series of pulse pairs that provides at different sensing characteristic.

61. A distributed fibre optic sensor apparatus as claimed in claim 50 wherein at least one of said additional series of pulse sequences comprises a series of phase difference pulse pairs, said phase difference pulse pairs comprising two pulses of the same frequency, wherein the phase difference between the pulses in successive phase difference pulse pairs of said series differs by a predetermined relative phase difference.

62. A distributed fibre optic sensor apparatus comprising: an optical source that launches at least first and second optical pulse pairs of a series pulse pairs into an optical fibre, each pulse having a frequency, wherein each pulse pair has a frequency configuration such that the first pulses of the series of pulse pairs have the same frequency as one another and the second pulses of the series of pulse pairs have the same frequency as one another, the pulses generated such that the phase relationship of the pulses of the first pulse pair and the phase relationship of the pulses of the second pulse pair are different by a predetermined relative phase difference; and a detector that detects radiation which is backscattered within the optical fibre and a processor coupled to said detector to form at least one channel signal comprising samples of the backscatter radiation, the samples acquired at substantially the same time after launch of each pulse pair; and to demodulate said channel signal to determine a phase value for the relevant section of optical fibre wherein: the pulses of the first and second pulse pairs have a frequency difference between them; and the first and second pulse pairs are generated so as to be launched into the optical fibre at different times; such that the time between launch of the first and second pulse pairs corresponds to the time for a relative phase change in a signal at a frequency equal to the frequency difference between the pulses in a pair equal to said predetermined relative phase difference.

Description

DESCRIPTION OF THE DRAWINGS

(1) The invention will now be described by way of example only with reference to the following drawings, of which:

(2) FIG. 1 illustrates the general components of a distributed acoustic sensor;

(3) FIG. 2 illustrates an interrogating pulse pair configuration used in embodiments of the preset invention;

(4) FIG. 3 illustrates how such a pulse pair sets the maximum spatial resolution of the sensor;

(5) FIG. 4 illustrates the origin of the backscatter signal from within the fibre;

(6) FIG. 5 illustrates the returns from launch of a series of pulse pairs according to an embodiment of the invention;

(7) FIG. 6 illustrates one embodiment of processing of the detected signal for a sensor channel;

(8) FIG. 7 illustrates an embodiment for producing pulse pairs having different polarisation states;

(9) FIG. 8 illustrates an embodiment of a detector arrangement for independently detecting the backscatter signal from pulse pairs of different polarisation states;

(10) FIG. 9 illustrates two examples of series of pulse pairs used in another embodiment of the invention;

(11) FIG. 10 illustrates the effect of a 90 relative phase change between the pulses in a pulse pair on the signal received at the detector;

(12) FIG. 11 illustrates an embodiment of a modulator arrangement for generating a controlled phase modulation between the pulses of a pulse pair;

(13) FIG. 12 illustrates two pulse pairs having a variation in phase difference between the pulses being generated simultaneously at different polarisation states;

(14) FIG. 13 illustrates one embodiment for generating simultaneous pulse pairs as shown in FIG. 12;

(15) FIG. 14 illustrates different series of pulse pairs being launched at different wavelengths;

(16) FIG. 15 illustrates three examples of pulse pair configurations;

(17) FIG. 16 illustrates the principle of temporal oversampling to generate spatial diversity samples;

(18) FIG. 17 illustrates an example backscatter intensity signal against sampling time for a single pulse pair;

(19) FIGS. 18a and 18b illustrate how spatial diversity samples can avoid the problem of fading;

(20) FIG. 19 illustrates the processing a diversity samples in one analysis bin to generate phase values for each channel;

(21) FIG. 20 illustrates how a selected number of channels in an analysis bin may be selected for combination;

(22) FIG. 21 shows an example of phase values calculated for the channels of an analysis bin and a resulting quality metric determined for similarity of one channel with the other channels;

(23) FIG. 22 shows another example of phase values calculated for the channels of an analysis bin;

(24) FIG. 23 shows an example of a combined waveform for the analysis bin;

(25) FIG. 24 illustrates the returns from launch of a series of pulse pairs such as shown in FIG. 9a; and

(26) FIG. 25 illustrates one embodiment of processing of the detected signal such as illustrated in FIG. 24 for a sensor channel.

DESCRIPTION OF THE INVENTION

(27) The general components of an interrogator unit of a distributed acoustic sensor are illustrated with respect to FIG. 1. In use the interrogator unit 100 is connected to an optical fibre 101 which acts as the sensing fibre. The sensing fibre is coupled to an output/input of the interrogator using conventional fibre optic coupling means. The interrogator unit is arranged to launch pulses of coherent optical radiation into the sensing fibre 101 and to detect any radiation from said pulses which is Rayleigh backscattered within the optical fibre. To generate the optical pulses the interrogator unit 100 comprises at least one laser 102. The output of the laser is received by an optical modulator which generates the pulse configuration as will be described later. The pulses output from the optical modulator 103 are then transmitted into the sensing fibre 101, for instance via a circulator 104.

(28) Any optical radiation which is backscattered from said optical pulses propagating within the sensing fibre is directed to at least one photodetector 105, again for instance via circulator 104. The detector output is sampled by an analogue to digital converter (ADC) 106 and the samples from the ADC are passed to processing circuitry 107 for processing. The processing circuitry 107 processes the detector samples to determine a phase value for each of a plurality of analysis bins, each analysis bins corresponding to a different longitudinal sensing portion of optical fibre. It will be noted that the interrogator unit may comprise various other components such as amplifiers, attenuators, filters etc. but such components have been omitted in FIG. 1 for clarity in explaining the general function of the interrogator.

(29) In embodiments of the present invention the laser 102 and modulator 103 are configured to produce at least one series of pulse pairs at a particular launch rate. Each pulse pair comprises at least a first pulse and a second pulse and preferably the first and second pulses are separated in time from one another as illustrated in FIG. 2. FIG. 2 shows a first pulse 201 at a first frequency F1 and having a duration d1 followed a short time later by a second pulse 202 having a second frequency F2 and having a second duration d2. In some embodiments the frequencies of the two pulse F1, F2 are equal whereas in other embodiments they are different as will be explained later. Preferably the durations (and hence the spatial widths) of the two pulses d1, d2 are equal to one another although this need not be the case. The two pulses 201, 202 have a separation in time equal to Ts (as shown Ts represents the time separation between the leading edges of the pulses).

(30) When a pulse pair having this type of configuration propagates within the optical fibre some light will be scattered from each of the pulses from the intrinsic scattering sites within the optical fibre. At least some of this backscattered light will be guided back to the beginning of the optical fibre where it can be detected. At any instant the light arriving at the detector may comprise light scattered from the first pulse from a first range of scattering sites and light scattered from the second pulse from a second range of scattering sites.

(31) FIG. 3 illustrates the propagation of the pulse pair in the optical fibre and shows distance along the fibre against time. Lines 301 and 302 illustrate the leading and trailing edges of the first pulse respectively and lines 303 and 304 the leading and trailing edges of the second pulse respectively. Thus at time t.sub.0 the leading edge of the first pulse enters the optical fibre and at t.sub.1 the trailing edge of the first pulse enters the fibre. The time between t.sub.0 and t.sub.1 therefore corresponds to the duration of the first pulse, i.e. d1. At a time t.sub.2 the leading edge of the second pulse enters the fibre and at t.sub.3 the trailing edge of the second pulse enters the optical fibre. Thus the time between t.sub.2 and t.sub.3 is equal to the second pulse duration, d2, and the time between t.sub.0 and t.sub.2 is equal to the pulse separation time, Ts. The pulses propagate in the fibre at a velocity equal to c/n where c is the speed of light and n is the effective refractive index of the optical fibre. Thus the gradient of the lines 301, 302, 303 and 304 are equal to c/n. This means that in the fibre the first and second pulses will have widths equal to W.sub.1 and W.sub.2 respectively, represented by the vertical distance between lines 301 and 302 and between lines 303 and 304.

(32) As the pulses propagate in the optical fibre some light will be backscattered towards the start of the fibre. This backscattered light will also travel at a velocity equal to dn. Consider the light reaching the detector. Line 305 represents the trajectory of light which could possibly be received at the start of the optical fibre at instant t.sub.4. Any backscattering which occurs at a time and distance into the fibre that lies on line 305 could be received at the start of the fibre at the same instant t.sub.4. Thus it can be seen that light from a first section of fibre illuminated by the first pulse at a first range of times will be coincident with light scattered from a second, different, section of fibre at a different range of times. It can also be seen that, as any light scattered from a pulse travels backwards to the start of the optical fibre at the same rate as the pulses themselves propagate forwards, the width of the first section of fibre is equal to half the width of the first pulse in the fibre, i.e. W.sub.1/2 and likewise the width of the second section of fibre is equal to half the width of the second pulse in the fibre, i.e. W.sub.2/2. Also the physical separation between the first and second sections of fibre is equal to half the physical separation of the pulses in the fibre.

(33) This means that, as illustrated in FIG. 4, at any instance the backscattered light received at the start of the optical fibre corresponds to backscattering in the fibre from a first section 401 of scattering sites illuminated by the first pulse and also from a second section 402 of scattering sites illuminated by the second pulse. The distance between these sections of scattering sites is referred to as the gauge length, L.sub.G. As illustrated the gauge length may be measured between the middles of the sections 401 and 402 of scattering sites.

(34) The backscatter signal received at the detector at any instant is therefore an interference signal resulting from the combination of the scattered light from all of these scatter sites. Effectively the backscatter signal at any instant corresponds to signal from a travelling interferometer defined by the positions of the scatter sites within sections 401 and 402. It will be appreciated that the backscatter from all the scatter sites of the second position 402, which will all be at frequency F2, can be considered to interfere to produce a composite signal from the second scatter sites and likewise the backscatter from all the scatter sites of the first section 401, which will all be at frequency F1, can be considered to interfere to produce a composite signal from the first scatter sites. These two composite signals will then also interfere.

(35) The distributed acoustic sensor of the present invention relies on the fact that any acoustic waves incident on the optical fibre may cause a disturbance, e.g. strain, in the optical fibre which can therefore phase modulate the interference signal generated by the travelling interferometer. As will be understood any phase changes occurring on the optical fibre up to the position of the second section 402 of scattering sites will affect the light from the first section 401 and the second section 402 of fibre equally. However any phase modulation (t) on the section of fibre between the first section 401 and second section 402 will affect the light from the first section 401 only. This can cause a change in phase of the interference signal. Detecting a phase change in the interference signal from a particular section of fibre can thus be used as an indication of an acoustically induced strain upon the optical fibre and hence as an indication of the acoustic waves incident at that section of fibre.

(36) Embodiments of the present invention ensure that the pulse duration and detector and processing arrangements are such that the sensor bandwidth is sufficient to detect all of the frequency components of the backscattered signal in an instantaneous sample. This allows the phase modulated signal to be detected with good signal to noise and good spatial resolution. An analytic version of this signal can be generated in various ways.

(37) In one embodiment the pulse configuration is such that the frequency of the first pulse is different to that of the second pulse, i.e. F1F2. This will result in the backscatter interference signal having a component at a carrier frequency equal to the frequency difference between the pulses (|F1-F2|). By monitoring the phase of this carrier frequency any phase modulations due to acoustic disturbances can be detected.

(38) In this embodiment of the invention therefore a series of pulse pairs are launched into the optical fibre, wherein each pulse pair in the series has the same frequency configuration, i.e. a pulse of frequency F1 and duration d1 followed a time Ts later by a pulse of frequency F2 and duration d2. The launch rate of the pulse pairs into the optical fibre (also called the ping rate) is related to the carrier frequency of these pulse pairs so that the time between successive launches is equal to the time taken for a signal at the carrier frequency to evolve by a predetermined phase amount.

(39) This embodiment of the present invention relies on a carrier within the acoustic bandwidth of interest being preserved between successive launches of pulse pairs. Thus a pulse pair is launched into the optical fibre and the backscatter signal received at the detector is sampled at intervals to provide a plurality of channels each corresponding to a different location of scattering sites within the optical fibre. After a certain interval, which corresponds to the time taken for a signal at the carrier frequency to change by a predetermined phase amount, another pulse pair is launched into the optical fibre and another plurality of samples acquired at the same intervals after launch. This is repeated for as long as it is wished to monitor the optical fibre.

(40) For each channel (defined by a certain sampling time after launch of a pulse pair) the successive detector outputs will provide a phase modulated carrier signal at the carrier frequency defined by the frequency difference between the pulses of a pulse pair.

(41) Referring back to FIG. 1 the modulator 103 may therefore comprise an acousto-optic modulator (AOM) for modulating the frequency of the optical radiation generated by the laser 102. As will be understood by one skilled in the art an AOM may be driven by a drive frequency (F.sub.d) and in operation shifts the optical frequency of output light by a frequency equal to the drive frequency. Therefore by driving the AOM at a first drive frequency for the duration of the first pulse and at a second different drive frequency for the duration of the second pulse the light in the first pulse will have a different resultant frequency to that of the light in the second pulse and the frequency difference between the pulses will be equal to the frequency difference between the respective drive frequencies used for the AOM. In other words the carrier frequency of the pulse pair can be controlled by controlling the drive frequencies used to drive the AOM. The laser and AOM are phase coherent and so the pulses in each pulse pair have the same properties as if they were derived from two stable sources running at the relevant frequencies.

(42) In one arrangement the carrier frequency is related to the launch rate such that the carrier frequency is less than half the launch rate such that a signal at the carrier frequency can only change in phase by less than 180 between successive launches. In effect this means that the launch-to-launch sample rate for each channel is above the Nyquist limit needed to accurately recreate the signal at the carrier frequency.

(43) In one arrangement the carrier frequency is arranged to be one quarter of the launch rate such that a signal at the carrier frequency evolves by 90 in phase between launch of successive pulse pairs.

(44) Using a phase change of 90 can ease processing as will be described but also allows for efficient use of modulation bandwidth. As the skilled person will appreciate, in the frequency domain any modulation may broaden the spectrum around the signal of interest at the carrier frequency. A similar effect will also be observed around DC and also at the image frequency (twice the carrier frequency). By ensuring that the carrier signal is at half the Nyquist limit the available bandwidth before crossover with the DC component or image frequency component is maximised.

(45) FIG. 5 illustrates how this embodiment of the invention operates. FIG. 5 illustrates the sampled detector output from a series of successive launches of a pulse pair having the frequency configuration as described above wherein the frequency difference between the pulses in the pair is equal to a quarter of the ping rate, i.e. launch rate of the pulse pairs. In the example shown the ping rate is 20 kHz, which may be a typical ping rate used with a length of optical fibre of the order of 5 km in length or less so as to ensure that only one pulse pair is propagating within the fibre at any one time.

(46) Thus, in this example pulse pairs are launched into the optical fibre every 50 s and the backscatter return signal generated by the pulse pair as it propagates through the fibre is detected. Following every launch of a pulse pair the detector output is sampled at a relatively high rate, for instance of the order of 80-100 MHz, to detect the backscatter signal from a plurality of different locations within the fibre. In this sample clock domain the random phase variations of the scatter sites lead to a random variation in intensity which destroys the carrier frequency and no useable signal may be detected.

(47) However, at any given position in the optical fibre, a variation in intensity may still be observed at the carrier frequency from launch to launch. Thus by comparing the backscatter interference signal from the same section of optical fibre from different pulse pairs a carrier signal may be detected. FIG. 5 illustrates that by taking the appropriate sample acquired the same time after launch of each pulse pair a carrier signal at frequency equal to one quarter of the launch rate may be detected. This carrier signal will be phase modulated by any acoustic signals affecting the relevant section of fibre as described above and thus by monitoring the phase of this carrier signal over time any acoustic signals incident on the relevant section of optical fibre can be detected.

(48) FIG. 6 illustrates one embodiment of how this modulated carrier signal may be processed by processing circuitry 107 to determine the phase of the carrier signal. The samples representing the modulated carrier signal for one channel of the sensor may be high pass filtered 601 to pass the signal at the relevant carrier frequency only. The filtered signal may then be divided into two processing channels and the signals in each channel may be multiplied by sine 602 and cosine 603 functions at the carrier frequency in order to generate In-phase (I) and quadrature (Q) components as is known in complex demodulation schemes. Where the known phase difference is 90 this simply comprises multiplying by +1 and 1. The resultant I and Q signals may then be low pass filtered 604 and 605 and converted to a phase value by rectangular to polar (RP) conversion 606. The RP conversion may optionally also generate an amplitude value. The phase value produced, .sub.0, may optionally be high pass filtered 607.

(49) Thus, using the samples acquired for each channel from successive launches of a series of pulse pairs, a phase modulated carrier signal can be detected and demodulated and thus any phase changes induced by acoustic signals acting on the relevant section of optical fibre can be detected.

(50) It will be understood that as each pulse pair in the series of pulse pairs has the same frequency configuration, that is a first pulse at a frequency F1 and a second pulse at a frequency F2, then if the backscatter signal from two different pulses pairs was incident on a detector at the same time the backscatter signals would interfere with another at the relevant carrier frequency. Hence the launch rate of the pulse pairs may be limited so as to ensure that radiation from only one pulse pair is present in the optical fibre at any one time. Thus the time between launches should be sufficient to allow all the radiation from the pulse pair to reach the end of the optical fibre and any radiation reflected or scattered from the end of the optical fibre to be reach the detector before another pulse pair is launched. The maximum launch rate of the pulse pairs may therefore be limited by the length of the optical fibre. An optical fibre having a length of the order of 5 km may operate with a launch rate up to about 20 kHz (assuming a refractive index of about 1.48). An optical fibre having a length of the order of 40 km may require the launch rate to be lower than 2.5 kHz. Where a much longer length of optical fibre is used there may be a maximum range into the fibre from which backscatter signals may be detected and in this case it may be possible to launch pulses at a rate that allows time for radiation from each pulse pair to reach this point of maximum range and return to the detector between successive launches. After this time from launch although there may still be some radiation from the original pulse pair in the optical fibre any such scattered radiation which reaches the detector will be insignificant.

(51) In one embodiment however the overall launch rate of pulse pairs can be increased by using polarisation to discriminate between successive pulse pairs. In this embodiment one pulse pair is launched into the fibre at first polarisation state. The next pulse pair is launched into the optical fibre with a second, different polarisation state. The first and second polarisation states are chosen such that the backscatter signal from the first pulse pair is received at the detector with a polarisation state which is orthogonal to the polarisation state of the backscatter signal from the second pulse pair. In this way the relevant backscatter signals can be separated by using polarisation sensitive elements.

(52) Thus the optical path for generating the interrogating pulses may comprise a variable polariser element as shown in FIG. 7. FIG. 7 shows that a series of pulse pairs may be output from modulator 103, which may as described above comprise an AOM, and incident on a variable polariser element 701. Variable polariser element acts on the pulse pairs to ensure that all light comprising a pulse pair is polarised to one polarisation state and ensures that successive pulse pairs have different polarisation states. In the example shown the polarisation states are orthogonal linear polarisation states, such as vertical and horizontal polarisations, although any orthogonal polarisation states, such as right and left circular polarisations, may be used instead.

(53) FIG. 8 illustrates one embodiment of a detector arrangement that could then used to discriminate between the backscatter signals. Backscatter light received from the optical fibre is directed toward a polarisation element 801 which, in this example is a polarisation dependent beam splitter. This beamsplitter 801 directs light having one polarisation state, say horizontally polarised, to a first detector 802 and light having an orthogonal polarisation state, vertically polarised, to a different detector 803. Thus each detector receives light of a single polarisation state only and hence light backscattered from one of the pulse pairs only. Thus the signal from each detector can be processed as discussed above to generate a series of phase values from each pulse pair independent of the other pulse pair.

(54) It will be appreciated that this technique relies on the input polarisation difference between the pulse pairs being maintained within the optical fibre. In conventional optical fibre the input polarisation difference will be maintained for a certain range in the fibre. Further any factors resulting in change in polarisation in the fibre may affect both polarisation states. Thus the detector arrangement may comprise a polarisation modulator 804 to compensate for any general polarisation drift that occurs within the optical fibre. The polarisation modulator may be adjusted in use to match any changes in polarisation drift, for instance by periodically illuminating the optical fibre only with light of a known polarisation and detecting the response at both detectors. Alternatively a polarisation modulator could be included in the output optical arrangement to apply pre-compensation to the pulses launched into the optical fibre.

(55) Instead of using a polarising beam splitter and two detectors element 801 could comprise a variable polariser element which transmits light to a single detector (say detector 801) but which rapidly varies the polarisation between the two orthogonal states. Thus the detector 802 could be arranged to take one sample at horizontal polarisation say and then another sample at vertical polarisation. Processing could therefore separate every other sample into a different stream from appropriate processing based on the time of launch of the appropriate pulse pair.

(56) In another embodiment, instead of using two pulses at different frequencies, the pulses of the pulse pair may be at the same frequency as one another. In this embodiment F1=F2. However the pulse pairs are generated such that the phase difference between pulses in the pair vary by a predetermined amount between successive pulse pairs.

(57) FIG. 9 illustrates some examples of series of pulse pairs that may be generated according to this embodiment of the invention. Four pulse pairs in each series are shown, each comprising two pulses at the same frequency (F1=F2).

(58) Series (1) shows a first example of a series of pulse pairs. In a first pulse pair there is a relative phase difference of P between the pulse in that pair. This will be an arbitrary phase difference that may arise because of the arrangement of the optical components used to generate the pulse pair. In some arrangements the two pulses of the first pulse pair may be in phase and thus the value of P may be zero.

(59) In the next pulse pair the phase difference between the pulses in the pair is controlled to vary by a predetermined amount from that of the previous pulse pair, in this example 90 (although other amounts of phase variation may be used if preferred). In other words the phase difference between the pulses in the second pulse pair is equal to P+90. In the third pulse pair the phase difference between the pulses in the pair is again controlled to be P and in the fourth pulse pair the phase difference is again P+90. In this way there is a relative phase difference between successive pulse pairs of 90.

(60) This variation in phase difference between the pulse pairs will result in a variation in the backscatter interference signal from a given section of optical fibre that can be used to determine a phase value for the relevant section of fibre. Referring back to FIG. 4 it will be recalled that the backscatter signal received at the detector at any instant is the combined signal from a range of scatter sites from a first section of optical fibre 401 illuminated by the first pulse and a range of scatter sites from a second section 402 illuminated by the second pulse. As mentioned above the returns from all of the scatter sites of the first section 401 will interfere together, as will the returns from all the scatter sites of the second section. For the purposes of explanation however consider a single (first) scatter site from the first section 401 and a single (second) scatter site from the second section 402.

(61) The phase of light reaching the detector from the second scatter site will depend on the initial phase of the light of the second pulse of the pulse pair and the effective path length to the second scatter site and back to the detector. The phase of light reaching the detector from the first scatter site will depend on the initial phase of the light of the first pulse of the pulse pair and the effective path length to the first scatter site and back to the detector. The path length to and from the first scatter site is equal to the path length to and from the second scatter site plus the additional path length from the second scatter site to the first scatter sight and back again. Thus the difference in phase between the signal scattered from the first and second scatter sites depends on the path length between the scatter sites and any phase difference between the light of the first and second pulses.

(62) If the first and second scatter sites are illuminated by two successive pulse pairs, at a rate sufficient that any path length changes occurring in the optical fibre between successive pulses are very small, then it will be clear that the only factor which influences any change in phase difference of the backscattered signals will be a change in phase difference of the pulses of the pulse pair. Thus if the first pulse pair has an initial phase difference P between the pulses in that pulse pair, this will lead to a certain phase difference (P) in the backscatter signals received at the detector. If the second pulse pair then has a phase difference P+90, this will lead to a phase difference of P+90 in the backscattered signals from the first and second scatter sites.

(63) As the skilled person will appreciate where two signals of the same frequency interfere with one another the result is a signal of the same frequency but with an amplitude that depends on the phase difference between the interfering signals (and their initial amplitudes) as illustrated in FIG. 10. FIG. 10 illustrates the backscatter signals that may be received from the first and second scatter sites and the resultant interference signal that would be detected. For simplicity the returns from the first and second scatter sites are illustrated with same amplitude although it will be appreciated that this may not actually be the case. The top plot shows the backscatter signals when illuminated with a first pulse pair having a phase difference P between the pulses. Curve 1001 illustrates the signal from the backscatter signal from the first scatter site and curve 1002 the signal from the second backscatter site. There is a certain phase difference between these signals which is determined by the factors mentioned above. Curve 1003 represents the resultant interference signal.

(64) The lower plot shows the backscatter signal when illuminated with a second pulse pair which has a phase difference of P+90 between the pulses. Curve 1001 again shows the backscatter signal from the first scatter site. Curve 1004 shows the backscatter signal from the second scatter site and curve 1005 the resultant interference signal. It will be noted that for ease of comparison curve 1001 is shown in the same phase position in both plots. This of course may not actually happen in practice and there may be an arbitrary phase change. However what is important is the relative phase difference between the backscatter signals from the first and second scatter siteswhich is determined solely by the variation in phase change between the pulses in the successive pulse pairs.

(65) The variation in phase difference between the pulses in the first and second pulse pairs leads to a phase variation in the backscattered signal which results in an overall amplitude variation in the resultant interference signal which can be used to determine a phase value for the section of fibre between the first and second scatter sites. Where the phase variation between successive pulse pairs is 90, as in this example, these interference signals represent In-phase (I) and quadrature (Q) components and low pass filtering of the signals can be used to derive I and Q values which can then be used to derive a phase value through conventional rectangular to polar conversion. Of course other phase values could be used for the variation in phase difference between successive pulse pairs and other conversion techniques used.

(66) The analysis above has focussed on scattering from a single scattering site from each of the first and second sections for ease of explanation. However it will be understood by one skilled in the art that the overall effect is generally the same when considering all the scattering sites together.

(67) FIG. 24 illustrates how this embodiment of the invention operates. FIG. 24 illustrates the sampled detector output from a series of successive launches of a pulse pairs such as shown in FIG. 9(a) where successive pulse pairs have a phase difference between the pulses of the pair that varies by 90.

(68) Pulse pairs are launched into the optical fibre at a launch (ping) rate suitable for the sensor, e.g. 20 KHz. The backscatter return signal generated by the pulse pair as it propagates through the fibre is detected. Following every launch of a pulse pair the detector output is sampled at a relatively high rate, for instance of the order of 80-100 MHz, to detect the backscatter signal from a plurality of different locations within the fibre.

(69) At any given position in the optical fibre, there is consequently a 90 phase difference in the relevant measurement signal from launch to launch. The signals from a given section of fibre from launch to launch can therefore be used to determine any phase modulation effecting the fibre by using the successive samples as I and Q signals.

(70) FIG. 25 illustrates one embodiment of how the backscatter signal may be processed by processing circuitry 107 to determine a phase value.

(71) The backscatter samples from a given section of fibre, i.e. a given channel, are received and alternating samples are switched, by switch 2501, into a In-phase (I) and Quadrature (Q) signal paths respectively. In each signal path high pass filter 2502 may remove unwanted signal components and the I and Q signals may then be low pass filtered 2503 and converted to a phase value by rectangular to polar (RP) conversion 2504. The RP conversion may optionally also generate an amplitude value. The phase value produced, .sub.0, may optionally be high pass filtered 2505.

(72) Thus, using the samples acquired for each channel from successive launches of a series of pulse pairs, with an induced variation in phase difference between the pulses of successive pulse pairs, an analytic version of a measurement signal can be recovered and used to derive a phase value such that any phase changes induced by acoustic signals acting on the relevant section of optical fibre can be detected.

(73) Referring back to FIG. 9, the first series of pulse pairs has an alternating phase difference between successive pulse pairs. Series (2) shows an example of an alternative series of pulse pairs where the same relative phase change is applied between successive pulse pairs.

(74) For explanation the relative phase of each pulse will be defined with reference to the phase of the first pulse in the series. Thus the first pulse of the first pulse pair will be taken to have a relative phase of 0. The second pulse in this first pulse pair will have an arbitrary phase difference, P1, from this first pulse which arises from the arrangement of the optics used to generate the pulses. In some arrangement the phase difference between the first and second pulses in the first pulse pair may be zero, i.e. P1=0.

(75) For the second pulse pair the first pulse will have another arbitrary phase difference, P2, from the first pulse of the first pulse pair which again may be a function of the optics used to generate the pulses. As the same general optical arrangement is used to generate the pulses in each pulse pair it would be expected that the same phase difference between the pulses, P1, would also exist between the pulse of the second pulse pair.

(76) In the absence of any external modulation (or thermal drift) one may therefore expect the first pulse in each pulse pair to differ in phase from the first pulse of the previous pulse pair by a relative phase difference of P2 and one would also expect a relative phase difference between the pulses of a pair equal to P1.

(77) This embodiment however deliberately introduces a phase difference between the pulses in each pulse pair which varies between successive pulse pairs. In the example shown the phase difference introduced is a positive phase difference of 90. The first pulse pair is produced with the inherent phase difference between the pulses, i.e. P1. For the second pulse pair however there is a phase modulation applied to the second pulse to introduce an additional 90 phase change. Thus the phase difference between the pulses of the second pulse pair is equal to P1+90. For the third pulse pair the phase modulation applied to the second pulse is increased by a further 90. Thus the phase difference between the pulses of the second pulse pair is equal to P1+180. For the fourth pulse pair the phase modulation is increased by a further 90 so that the phase difference between the pulses of the fourth pulse pair is equal to P1+270. For a fifth pulse pair, not shown the phase modulation could be reset to 0.

(78) Thus between each successive pulse pair the phase difference between the pulses of the pair varies by 90.

(79) It will be appreciated that FIG. 9 illustrates a phase modulation being applied to modulate the phase of the second pulse in each pulse pair but the phase difference could instead by applied to the first pulse in each pulse pair, or both pulses in a pair could be phase modulated to provide the required phase difference.

(80) The phase variation between the pulses could be generated in a number of ways. For instance referring back to FIG. 1 modulator 103 could comprise an electro-optic modulator (EOM) such as a pockels cell or liquid crystal electro-optic modulator to apply a controlled phase variation to at least one of the pulses. Modulator 103 may also therefore comprise an intensity modulator for forming the pulses. As an example, for one pulse pair the EOM could be controlled to apply no phase variation to the pulse pair which will therefore be transmitted with any phase difference between the pulses being due to the inherent phase difference of the generation optics. For a subsequent pulse pair the EOM may apply a phase difference to one of the pulses, or to both of the pulses, to generate the required phase difference.

(81) FIG. 11 shows an alternative example of modulator 103. Incoming light is divided into two channels 1101 and 1102 for generating the first and second pulses respectively. Light is channel 1101 is modulated by an intensity modulator 1103 to form the first pulse. Light in channel 1102 is incident on EOM 1104 which applies a required phase modulation (which varies from pulse pair to pulse pair). The phase modulated light is then intensity modulated by intensity modulator 1105. The timing of intensity modulators 1103 and 1104 are controlled so as to provide the required duration of the pulses and also the required time separation between the pulses. Additionally or alternatively a delay 1106 in channel 1102 may provide at least some of the required time separation (although in some embodiments it is preferable that the time separation of the pulses is variable and use of a fixed delay may not be appropriate) The two pulses, having the required phase difference, are then combined to an output line for transmission.

(82) As successive pulse pairs have the same frequency as one another the launch rate of the pulse pairs may be arranged so that radiation for only one pulse pair at a time can reach the detector. Again, as with the embodiment described above, polarisation discrimination may also be used between successive pulse pairs to allow a faster update rate.

(83) In another embodiment first and second pulse pairs of a first series having a variation in the phase difference between the pulses in the pair may be launched into the optical fibre simultaneously or near simultaneously with pulse pairs of a second series, with the first and second series having different polarisation states. Thus, as illustrated in FIG. 12 a first series of pulse pairs, with a varying phase difference between the pulses in successive pulse pairs, may be launched into the optical fibre with a horizontal linear polarisation state. At, or near the same time a second series of pulse pairs having with phase variations between the successive pulse pairs may be launched into the optical fibre with a vertical linear polarisation state. The two pulse pairs will therefore propagate down the optical fibre together and will generate backscatter signals from the same sections of optical fibre at the same time. These backscatter signals can be passed independently to different detectors, for instance using a detector arrangement as shown in FIG. 8, to allow for simultaneous or near simultaneous measurements from a given section of fibre. As the backscatter signal exhibits a polarisation dependence such simultaneous or near simultaneous interrogation could help avoid problems with fading as the signals from both series may not fade at the same time.

(84) Two pulse pairs having different polarisation state could be produced in a number of different ways. For instance, FIG. 13 shows one possible arrangement where a pulse pair output from intensity modulator 1103 is separated by polarising beam splitter 1301 into horizontally and linearly polarised components into two different channels. Both channels are incident on an EOM 1302 which imposes appropriate phase changes to at least one of the pulses to introduce relative phase differences in successive pulse pairs and the two channels are then recombined so that the pulses are temporally coincident for launching into the optical fibre. When two pulses are coincident however the overall optical power of both pulses should be below the non-linear threshold for the optical fibre and hence, to increase the power available in each pulse the actual pulses may be slightly staggered as illustrated in FIG. 12.

(85) Wavelength division multiplexing techniques can also be used to allow more than one pulse pair to be in the fibre at the same time, for instance to increase the update rate of the sensor and/or to provide different sensor resolution or sensitivity.

(86) It will be appreciated that the techniques described above rely on two different pulse pairs having the same frequency configuration as one another being transmitted into the optical fibre with a predetermined phase relationship between the pulse pairs, either through the timing of launch in relation to a carrier frequency or direct control of the relative phase difference between pulses in the pulse pair. Thus a series of pulse pairs according to embodiments of the present invention should have the same frequency configuration. However wavelength division techniques can be used to launch other series of interrogating pulses into the optical fibre at the same time.

(87) FIG. 14 illustrates a series of pulses being launched at different times and at different wavelengths. Thus a first series of pulse pairs may be launched at a first wavelength, 1, with a period T.sub.1 between each pulse pair. A second series of pulse pairs may also be launched into the optical fibre at a different wavelength, 2, with a period T.sub.2 between each pulse pair in this second series and a third series of pulse pairs may also be launched into the optical fibre at a different wavelength again, 3, with a period T.sub.3 between each pulse pair. The time between launches of the pulses in each series need not be the same, i.e. T.sub.1 may be different to at least one of T.sub.2 and T.sub.3 but in some embodiments the launch rate of at least some of the pulse pairs are the same (i.e. T.sub.1 may be equal to T.sub.2 and/or T.sub.3) so that the pulse pairs in the relevant series maintain a certain temporal relationship.

(88) In such an application the relative launch times of pulse pairs may be arranged so that there is a defined time, T.sub.4, between launch of a pulse pair of the first series and a pulse pair of the second series and a defined time, T.sub.5, between a pulse pair of the second series and one of the third series. The time between launches may be arranged regularly so that, in this example, T.sub.4=T.sub.5=T.sub.1/3. If each pulse pair of each series allows a measurement from the same sections of optical fibre this arrangement effectively increases the update rate from each section of fibre by a factor of three as compared to using a single pulse pair. This could be particularly advantageous when monitoring relatively long lengths of optical fibre where the launch rate for a single series of pulse pairs would provide too slow an update rate for the acoustic events of interest.

(89) In another embodiment however the timing of the pulses in each series of pulse pair may be irregularly spaced, i.e. T.sub.4T.sub.5 and both T.sub.4 and T.sub.5 may be different to the time between a pulse of the third series and a subsequent pulse of the first series. By spacing the timing of the pulse pairs in each series irregularly the time between updates for each sensing portion of fibre also varies. A certain update rate may allow better detection of some event than others. For instance if a certain acoustic event has a frequency equal to a multiple of the update rate of the sensor it may be that each time the relevant sensing portion of fibre is sampled the same return is detectedwhich may lead to the acoustic signal being missed. However if the relevant section of fibre is sampled at irregular intervals in at least one of the samples the variation due to the acoustic signal will be detectable. Thus the use of irregular spacing may aid in detection of events. Further processing of measurement signals acquired at such irregular interval may be able to provide a synthetic sampling rate higher than that achievable through use of three regularly spaced series of pulse pairs.

(90) One or more of these series of pulse pairs may consist of pulse pairs which comprise pulses at different frequencies to one another with a carrier frequency related to the launch rate as in an embodiment above, in which case the reference to a wavelength for the pulse pair may refer to a base wavelength.

(91) For instance, consider the first series of pulse pairs. A laser may produce light of wavelength 1 (e.g. 1550 nm), which corresponds to a base frequency of 1 (about 193.4 THz), which is then modulated by an AOM to provide two pulses of frequency 1+f1 and 1+f2 respectively (f1 and f2 being the relevant driving frequency for the AO for each pulse). The driving frequencies of the AOM are arranged so that |f1-f2| (i.e. the carrier frequency) is related to the launch rate (1/T.sub.1) to provide the required phase difference in the carrier signal from launch to launch, i.e. for a phase difference of 90 the carrier frequency may be equal to a quarter of the launch rate. Thus the carrier frequency may be of the order of a few kHz.

(92) The second series (and/or third series) may also comprise pulse pairs according to this embodiment, i.e. having pulses with a frequency difference between them related to the launch rate. Where the launch rates for the relevant series are the same this may mean that the carrier frequency for each series of pulse pairs is the samebut the base frequency of the pulses is different. For example, for series two, the base wavelength 2 (e.g. 1530 nm) will corresponds to a base frequency of 2 (about 195.9 THz). Light at this base frequency may by modulated by an AOM operated at the same driving frequencies to provide two pulses of frequency 2+f1 and 1+f2 respectively. The carrier frequency for series two will therefore also be |f1-f2|. Of course if different launch rates were used for the different series the carrier frequency for each series may also vary accordingly.

(93) It will be noted that even though the pulses of the series of pulse pairs are modulated to a different frequency from the base frequency this modulation is relatively small compared to the frequency separation employed in wavelength division multiplexing schemes (for example the wavelengths 1550 nm and 1530 nm discussed above lead to base frequencies of 193.4 THz and 195.9 THz respectively). Thus the different wavelengths typically used even in DWDM (dense wavelength division multiplexing) schemes could be used as the base wavelengths for the different series of pulse pairs and still allow separation of the backscatter signals from each series of pulse pairs.

(94) At least one of the series of pulse pairs may also comprise pulses of the same frequency but having a variation in the phase difference between each of the pulses as discussed above in relation to FIG. 9. Additionally or alternatively at least one series of pulse pairs could comprise a series of simultaneous launches of two pulses pairs at different polarisations and having a variation in the phase difference between the pulses, i.e. a series of launches of pulse pairs as shown in FIG. 12.

(95) It will be noted that whilst the discussion above has focussed on pulse pairs being launched into the optical fibre there may be other pulse configurations launched into the optical fibre as well. For instance a series of single pulses could be launched into the optical fibre at a different wavelength to a series of pulse pairs.

(96) As mentioned the use of multiple series of pulse pairs can be used to provide a faster update rate than would be possible using a single pulse pair alone. Additionally however the use of multiple pulse pairs may be used to provide sensing at different spatial resolutions and/or sensitivity.

(97) Referring back to FIGS. 3 and 4 it will be recalled that when interrogating the optical fibre using a pulse pair comprising temporally distinct first and second pulses the backscatter signal received at the detector at any instant is that received from a first section of fibre 401 illuminated by the first pulse and from a second section 402 of fibre illuminated by the second pulse. Each of these sections of optical fibre is equal in width to half of the relevant pulse width and the separation of (the leading edges of) the sections, referred to as the gauge length, is equal to half the time separation of the interrogating pulses.

(98) As mentioned previously any strain on the optical fibre that affects the fibre between the first and second sections of optical fibre will therefore lead to a detectable change in phase. It will also be appreciated that a change in path length occurring within either of the first or second sections of fibre 401 and 402 will result in a change of phase of the backscatter signal from at least some of the scattering sites. Thus any acoustically induced phase changes occurring within the area of the first and second sections 401 and 402 may be detectable (with a varying degree of sensitivity). Thus the maximum achievable spatial resolution of the distributed acoustic sensor depends on the separation of the pulses and also on the width of the pulses used.

(99) The configuration of the pulse structure also impacts on other aspects of system performance however. For instance the relative width of the pulses can influence the signal to noise ratio of the sensor and also the effective length of sensing fibre that can be reliably monitored.

(100) For a distributed acoustic sensor based on Rayleigh backscatter it is desirable to avoid non-linear optical effects within the optical fibre. This means typically that for a particular length of optical fibre there is a maximum power threshold, or in other words a limit to the intensity of the radiation of the pulses. The amount of backscatter received at any time is related to the intensity of the pulses but also to the width of the pulses. Simplistically speaking the more photons within a pulse the greater the amount of backscatter and if the intensity is capped the only way to increase the number of photons is to increase the pulse length. Thus using wider pulses, i.e. longer duration pulses, can improve the signal to noise ratio for any given section of optical fibre.

(101) However using wider pulses can lead to the phase centre of the relevant sections of fibre being less well defined.

(102) For example FIG. 15 illustrates three examples of different pulse configurations having pulses of different widths. FIG. 15 shows the scattering sites within the fibre that would contribute to the instantaneous signal arriving at the detector. In each case the time difference between the start of the first pulse and end of the second pulse is the same. Hence, within the fibre, the outer edges of the sections of fibre which may contribute to scattering at any instance are located the same distance apart. As the intensity of the backscatter varies on the scale of the pulse width FIG. 15 also illustrates (dotted line) an idealised maximum intensity variation in terms of spatial length along the fibre.

(103) FIG. 15a shows an example with pulses that immediately follow one another. Hence the sections of fibre 1501 and 1502 from which scattering may occur are also adjacent. FIG. 15b illustrates an example in which the pulses are temporally separated and the pulses are relatively wide compared to the gap between pulses (which leads to a similar configuration of the sections of fibre 1502, 1504 which contribute to the scattering signal at any instant). FIG. 15c shows an example in which the pulses are temporally separated and the width of the pulses is relatively narrow compared to the gap between pulses and hence the sections of fibre 1505 and 1506 (from which scattering from the first and second pulses respectively may be received at the same time) are similarly relatively narrow compared to the gap between them.

(104) The example shown in FIG. 15a would provide the best signal to noise ratio of any of these pulse configurations. However the configuration illustrated with regard to FIG. 15 would have a variable phase centre.

(105) It will be appreciated that scattering may occur from any of the scattering sites within section 1501 when illuminated by the first pulse and likewise scattering may occur from any of the scattering sites within section 1502. As the distribution of scattering sites is effectively random within the optical fibre it is entirely possible that in one section of fibre the distribution of scattering sites is not even throughout the section. The scattering sites may be concentrated in certain areas. Consider a path length variation that occurs at about the middle of section 1501 of optical fibre. Any scattering from within section 1501 that occurs before this disturbance will clearly not experience the path length variation and thus will not lead to a detectable phase change in the interference signal. Any scattering from section that 1501 that occurs after the path length variation clearly have a relative phase difference that may lead to a detectable phase change in the interference signal. If the scattering sites are distributed evenly throughout section 1501 then about half of the scattering from section 1501 will have the relevant phase change. If however the distribution of scattering sites are such that most of the scattering occurs before the position of the disturbance then the majority of scattering from section 1501 may not exhibit the relevant phase change and the sensitivity to such a disturbance is therefore relatively low. Thus a pulse configuration of the type illustrated with reference to FIG. 15 will tend to have a sensitivity that varies greatly based on the distribution of scattering sites.

(106) The pulse configuration used in the example shown in FIG. 15b has a reduced SNR compared to that of FIG. 15a but has a better defined phase centre. The pulse configuration used in the example shown in FIG. 15c has a further reduced SNR but even better defined phase centre.

(107) In addition the pulse configuration example illustrated in FIG. 15c, and to a lesser extent FIG. 15b, provides opportunity for multiple diversity samples to be acquired which can reduce the problem of fading.

(108) Referring back to FIG. 4 it will be recalled that the signal received at the detector at any instant is the result of the interference of all the scattered light from section 401 of the fibre and from section 402 of the fibre. As mentioned previously one can effectively consider all the scattering from section 401 interfering to provide a composite signal at frequency F1 and all the scattering from section 402 interfering to provide a composite signal at frequency F2. It will be appreciated that as the scattering sites are effectively randomly distributed there is a finite chance that at any location in the optical fibre the scattering from either section 401 or section 402 destructively interferes so that no, or only a very low intensity, composite signal is produced. In this event, there will be no appreciable interference with the composite signal from the other section and it will not be possible to determine any phase variations at that position in the optical fibrethe signal has faded.

(109) Embodiments of the present invention overcome the problem of fading by using a pulse configuration that comprises temporally separated pulses wherein the gap between the pulses is equal to or greater than the pulse width (of at least the second pulse), i.e. by using a pulse pair of the type illustrated with reference to FIGS. 15b and 15c.

(110) Consider how the backscatter signal generated by such a pulse pair changes as the pulses propagate through the optical fibre. FIG. 16 shows that at a first sample time, T.sub.1, the signal received at the detector is due to the scattering from a first section 1601 of fibre illuminated by the first pulse and a second section 1602 illuminated by the second pulse. At a second sample time, T.sub.2, the signal at the detector will be due to a third section of fibre 1603 illuminated by the first pulse and a fourth section 1604 of fibre illuminated by the second pulse. If the sample time between T.sub.1 and T.sub.2 is equal to or greater than half the duration of an individual pulse then the sections of fibre 1602 and 1603 illuminated by the first pulse at the different sampling times will be different (i.e. there will be no overlap). Likewise the sections 1602 and 1604 of fibre illuminated by the second pulse will also be completely different. As the scattering from these different sections of fibre are uncorrelated, the effective bias position of the interferometer formed by the pulses are also uncorrelated. The backscatter signal intensity (from a single pulse pair travelling through the optical fibre) will therefore vary over a temporal length of the order of half of the pulse length, i.e. of the order of the width of the sections of scattering sites within the optical fibre. FIG. 17 illustrates the backscatter intensity pattern and how the intensity may therefore vary between the two sample times, T.sub.1 and T.sub.2. Thus the samples acquired at the different sample times can be seen as diversity samples.

(111) In effect at each different sample time the phase bias position of the interferometer defined by the scattering sites varies. FIG. 18 shows an illustrative curve 1801 of backscatter signal intensity against phase bias positionwhen the scattering sites define a interferometer with zero phase bias, i.e. the backscatter signals are in phase, the intensity is at a maximum and when the phase bias position is 180, i.e. the signals are out of phase, the intensity is at a minimum.

(112) FIG. 18a illustrates the phase bias position 1802 at the first sample time, T.sub.1. The distribution of scattering sites at sections 1601 and 1602 of fibre will determine the relevant phase bias position. FIG. 18b illustrates the phase bias position 1804 at the second sample time, T.sub.2.

(113) Ignoring thermal drift for a moment, for each successive launch of pulse pairs a sample acquired at the same sample time, i.e. T.sub.1 or T.sub.2 after launch (which will be referred to as a channel) will give rise to the same phase bias position. Thus, in the absence of any other modulation the same general intensity of backscatter signal will be detected in each channel from successive launches. Any acoustically induced phase modulation 1803 will however vary the phase bias position of the output signal and will lead to a variation in output intensity in the channel which can be detected. The example shown in FIG. 18a shows a phase bias position that leads to a intensity modulation that would be readily detectable. However, for the example shown in FIG. 18b the channel corresponds to a phase bias position that leads to a low intensity signal. FIG. 18 also illustrates the noise floor of the sensor. The skilled person will appreciate that there will be a certain noise floor, defined by various parameters, which sets the lower signal limit which can reliably be detected. It can be seen that the output signal generated from channel samples acquired at sample time T.sub.2 will mainly be masked by the noise floor.

(114) Over time thermal variations will also alter the phase bias position of different channels. This thermal variation is slow on the time scale of the acoustic disturbances being monitored and thus acoustic disturbances can clearly be detected. However thermal variations may result in a change of SNR of an individual channel over time.

(115) Referring back to FIG. 16 it can therefore be seen that, by using a pulse pair configuration with gap between the pulses of the order of the pulse duration or longer, the time between samples can be set so that each successive sample relates to a substantially independent set of scattering sites but the section of fibre defined by the scattering sites in each case largely overlaps. This overlap means that any acoustic signals causing a path length change in the relevant section of optical fibre can be detected from either sample. Thus if diversity samples acquired at one sample time, T.sub.1, after launch of each pulse pair (i.e. corresponding to a first channel) happen to have a low SNR due to the phase bias position of the resultant interferometer, the diversity samples from another sample time, T.sub.2, (a second channel) may have a better SNR and may be used to determine the acoustic signals affecting that section of fibre

(116) As illustrated in FIG. 16 a pulse configuration with relatively narrow pulses compared to the gap between pulses may allow a number of different diversity samples to be acquired, each corresponding to substantially independent scattering sites and all of which define portions of fibre which partly overlap. Thus a plurality of diversity samples may be acquired which correspond to a plurality of overlapping spatial sensing portions of optical fibre.

(117) The sample rate of the detector may therefore be set at a rate such that a plurality of samples are acquired in the time taken for the pulses to propagate in the optical fibre by a distance equal to the gauge length. This can reduce the problem of fading as compared to acquiring a single sample from a section of fibre equal to the gauge length as clearly the likelihood that all the sample positions have faded is much lower than the probability that an individual sample position has faded. Thus by taking multiple diversity samples within the gauge length the SNR of the sensor is improved.

(118) It will be noted however that such temporal oversampling does not increase the achievable maximum spatial resolution of the distributed acoustic sensor, which as noted above is defined by the pulse pair widths and separation.

(119) The plurality of successive diversity samples may be analysed in various analysis bins, each analysis bins corresponding to a certain defined sensing portion of the optical fibre to provide a single measurement for each analysis bin. In other words consider that the pulse pair configuration is such that the gauge length is, for example, 10 m. This corresponds to a pulse separation of the order of 100 ns. The analysis bins may therefore correspond to contiguous 10 m sections of optical fibre. The sample rate may be such that, for example, 10 or so diversity samples are acquired in the time take for the backscattered radiation reaching the detector to correspond to a different 10 m section of fibre, i.e. twice the time taken for the pulses to move 10 m within the optical fibre (to allow for time for the pulses to move 10 m into the fibre and the light to return that extra 10 m distance). In other words, taking the refractive index of the fibre to be about 1.5 the samples may be acquired at a rate of about 100 Mhz.

(120) In this example each analysis bin may therefore comprise ten channels, each successive channel receiving successive diversity samples acquired following launch of a pulse pair. Each channel therefore corresponds to a 10 m section of fibre with the relevant 10 m section being displaced by 1 m between each channel and each channel will be updated at the launch rate of the pulse pairs. The section of fibre to which the relevant analysis bin pertains may therefore be defined as the 10 m of fibre which is in the middle of length of fibre corresponding to all the channels of the analysis bins. In other words if the first channel of the analysis bin defines a 10 m section of fibre from a position x to x+10 m into the fibre, and the second channel defines a section from x+1 m to x+11 m and so on till the tenth channel defines a position from x+9 m to x+19 m then the analysis bin may be defined to relate to the section of fibre from x+4.5 m to x+14.5 m. It will be apparent that grouping the channels into bins in this way does mean that the any modulation affecting a section of fibre corresponding to one analysis bin will also have an effect in an adjacent analysis bin. This does have an impact on spatial resolution but the advantages of the diversity processing scheme compensate for such reduction in spatial resolution.

(121) The analysis to produce a single measurement value for an analysis bins could be performed in many different ways.

(122) For instance the processing may involve averaging the samples from all the channels forming the analysis bin and using the average sample value to determine a value of phase for the analysis bin.

(123) However in a preferred embodiment the diversity samples in the individual channels may be assessed using a quality metric and the result of the assessment used in combining the samples from the various channels.

(124) One possible basic metric is the average intensity of the diversity samples in each channel. This average intensity may be determined over time and used as an indication of the SNR for that channel. Referring back to FIG. 18 it will be clear that the average intensity depends on the relevant phase bias position that applies for that channel. The samples from channel with the highest SNR may then be used as the samples for the whole analysis bin, with the other samples being discarded. Alternatively samples with an SNR below a certain threshold, i.e. with an average intensity below a certain threshold may be discarded and any remaining samples average together.

(125) In one embodiment however the diversity samples in each channel are processed to determine a phase value for that channel, the resulting phase values for each channel are assessed using a quality metric and then a combined value for the analysis bin is determined.

(126) In this embodiment therefore each channel may be separately processed to determine a phase value for that channel. The phase value may be determined for each channel using standard demodulation techniques. For instance when using pulse pairs according to the embodiment described above in relation to FIG. 5 a demodulation method may be applied to each channel as described with respect to FIG. 6. When using pulse pairs according to the embodiment described with reference to FIG. 9 or 12 the detected signals may, for instance, inherently comprise I and Q components which can be low pass filtered and converted to a phase value through rectangular to polar conversion or otherwise can be converted to I and Q components.

(127) The result of such processing is a phase value, and optionally, an amplitude value, for each channel.

(128) The AC to DC ratio for each channel may be used as an indication of signal to noise ratio for the relevant channel. Thus the phase values for the channels may be combined taking this ratio for the channels into account. A relatively high ratio may indicate a relatively high SNR whereas a low value may represent a low SNR. The derived phase values from channels with an AC to DC ratio value below a threshold may therefore be discarded. Additionally or alternatively the phase values may be given a weighting in forming the combined value with the weighting being, at least partly, based on the AC to DC ratio.

(129) In one embodiment the phase values derived for the channels may be analysed according to a quality metric based on the degree of self similarity of the signals from each channel. For most applications the assumption can be made that, where diversity samples are acquired from overlapping sensing portions of the optical fibre, any acoustic disturbance acting on the fibre at that general location will lead to substantially the same phase modulation in each channel. Thus all processed phase values for all of the channels can be compared to each other to determine how similar the results from each channel are to one another. Channels that exhibit a high degree of similarity can be assumed to be measuring the same stimulus, whereas a channel that is significantly different to the others may effectively be noise dominated.

(130) A quality metric may therefore be applied to the phase values determined for each channel in order to determine the degree of self similarity to the other channels. The phase values may then be combined with an appropriate weighting applied to each phase value based on this quality metric. In other words phase values which are very similar to one another may be given relatively high weightings in the combination whereas phase values which are less similar to one another may be given relatively low weightings in the combination. In this the way the combination gives more weight to good quality samples and less weight to poor quality samples and thus improves the signal to noise ratio compared to conventional combination techniques.

(131) The degree of self similarity of a sample with the other samples may be determined by comparing the phase values determined for a channel with the phase values determined for the other channels. Preferably a relatively simple and straight-forward correlation is used to minimise processing overhead.

(132) In one embodiment a first metric, M1, is used to determine the similarity of the variation of the signals in the channels being compared. The metric M1 applied to the signals A and B from channels A and B may be of the form:
M1(A,B)=(A<A>).Math.(B<B>)Eqn. (1)

(133) This metric can give a large result for a signal with a large DC component. Advantageously therefore a second metric, M2, may be used to determine a measurement of the magnitude of the difference between the two signals. The metric M2 may be of the form:
M2(A,B)=((A<A>)(B<B>)).sup.2Eqn. (2)

(134) These two metrics may be calculated for each combination of channels within the analysis bin and used to determine the channels which are most similar to one another. A single overall metric M.sub.Q may be calculated by:
M.sub.Q(A,B)=M1(A,B)M2(A,B)Eqn. (3)

(135) The value of the calculated quality metric M.sub.Q may then be used to determine those channels which are most self similar.

(136) FIG. 19 shows one embodiment of how the relevant phase values may be determined for the various channels in an analysis bin.

(137) The I and Q components for each of the n channels in an analysis bin may be received and low pass filtered 1901 to derive I and Q values. These I and Q values may then by converter to a phase value and amplitude value A by rectangular-to-polar conversion 1902. The phase value and amplitude value A from each channel may be used in subsequent processing 1903 to determine self similarity. The phase value for each channel is also passed to a filter 1904 to perform time averaging and produce an average phase value <>. The phase value and average phase value <> for each channel may then be processed to determine the metrics discussed above. Once the channels which are self similar have been identified the phase values from the relevant channels can be combined. However as the metric processing takes some time the phase value for each channel may also be passed to a buffer 1905 as a delay. The delayed phase value .sub.d can be used as the value to be combined. In one embodiment the combination involves combining the difference between the current and average phase values and hence the delayed phase value .sub.d may be also filtered 1904 to produce an average which may also be used in the combination. Note that the same filter block 1904 is shown as acting on the phase value and delayed phase value to provide the relevant averages. It will be appreciated however that in practice separate filters may be applied.

(138) In one embodiment the subsequent processing determines a metric for each combination of channels and selects a pre-determined number of channels which are the most self-similar for combination. In other words the method may comprise the step of selecting the m most self-similar channels from the n channels available within the analysis bin. For example if there are 10 channels within the analysis bin, i.e. n=10, then the five most self similar results from the channels (m=5) may be chosen to be combined.

(139) The number of channels which are selected to form the combined phase result may be the same for each analysis bin and may be constant over time. This may ease some of the later processing as each combined phase value is formed from the same number of individual samples. It also will ensure a constant noise floor for the sensor. In one embodiment however the number of channels which are used in the combination is configurable by a user of the interrogator unit.

(140) Figure illustrates one implementation of the metric based processing to select the M most self-similar channels. The phase value and average phase value <> are received for each channel. For each channel the metric, M.sub.Q, is determined for the combination with each other channel to form a matrix 2001 of the metric score. As only the magnitude of the metric is important the metric for each combination of channels need only be calculated once, i.e. the magnitude of metric M.sub.Q(A, B) would be the same as that for M.sub.Q(B, A).

(141) This result, for each channel, in effectively a series of scores for each of the other channels. At this stage a predetermined number, X, of the highest metric scores for a channel may be identified. The corresponding channels for those highest metric scores may be identified and counted, for instance in a histogram type arrangement. Thus for channel 1 say, if the highest X metrics included the metrics from combinations with channels 2, 5 and 6 then the histogram count for channels 2, 5 and 6 would be increased by one. This process can be repeated for all channels. The histogram will then indicate which channels are the most similar to the other channels and the M channels with the highest histogram counts may be selected for combination.

(142) For each selected channel the delayed phase value .sub.d is used in the combination so that the data being combined is the same data from which the metrics were calculated. In one embodiment the combination is the sum, for each of the selected channel:
W.sub.i(.sub.d<.sub.d>)Eqn. (4)
where W.sub.i is an optional weighting for ith channel. For instance the weighting W.sub.i could be based on the histogram count.

(143) FIGS. 21 to 23 show some processed data from a plurality of diversity samples channels acquired from a distributed acoustic sensor interrogated with pulse pairs of the form shown in FIG. 2. The detector was sampled at 80 MHz to provide diversity channels of 1.25 m spacing and the optical fibre was stimulated with a pulsed modulation. Eight successive diversity channels were grouped together in an analysis bin.

(144) The top plot of FIG. 21 shows the demodulated phase data from each of the channels over time. It can be seen from the raw phase data that several channels show a pulsed phase modulated of about 3 radians amplitude but that other channels show a much less clear signal with a lower amplitude. The lower plot of FIG. 21 shows, for one channel, the calculated similarity metric M.sub.Q, for that channel with each of the other channels. It can be seen that for the selected channel there is a relatively high metric score for several of the channels but that the metric for the other channels (which in this case are the ones which are the noisy channels) is low.

(145) FIG. 22 again shows phase data derived for each of the eight diversity channels. FIG. 23 shows the output after processing according to the method described above, i.e. the resulting output phase waveform. It can be seen that the output waveform is a clean waveform that matches the waveform of the majority of channels. The result is therefore a combined phase value for the analysis bin with good signal to noise properties and a much reduced change of fading.

(146) As mentioned above selecting a predetermined number of channels to be combined means that the noise characteristics for each analysis bin may be generally the same and may remain constant over time. However other arrangements are possiblefor example a quality metric could be applied to the phase values of each channel to determine the degree of self similarity to other channels and all phase values which are sufficiently similar to one another may be combined. For example the channels may be grouped into sets wherein the metric for each channel combination in the set is above a certain threshold. The largest such set may then be selected for combination.

(147) Thus if, at any time, 8 out ten channels are similar to one another, e.g. score sufficiently highly, these 8 similar channels may be combined but, for another analysis bin (or for the same analysis bin at a different time where thermal fluctuations may have altered the channel properties), if only 4 channels are similar to one another then only those four channels may be combined.

(148) Changing the number of channels used in the combination will affect the noise properties of the system but this could be dealt with by suitable calibration. The processing circuitry may also provide an indication of the number of channels of an analysis bin that were used in performing the combination.

(149) It will also be appreciated that the phase centre of the analysis bin may vary depending on the channels actually selected for combination. For example, if five channels are selected for combination in an analysis bin of ten channels, then if the first five channels are the ones selected the phase centre of the resulting combined sample will be skewed to one side of the analysis bin. The processing circuitry may therefore also be arranged to provide an indication of the relevant channels or at least the spread of channels used in the combination and/or an estimate of phase centre.

(150) The metric discussed above looks for self-similarity of the demodulated phase values of the channels in the analysis bin. This relies on the assumption that each of the various channels is experiencing the same general stimulus. This is a reasonable assumption in most instances but it may not be sufficient for some precision applications or where incident waves with high wavenumber may be expected. In such cases it may be desirable to apply some time adjustment to the phase values of each channel prior to assessment to take into account a direction of arrival of the acoustic stimulus, i.e. to apply some beamforming type techniques.

(151) The embodiments described above use pulse pairs wherein different pulse pairs have different phase encoding in the order to recover a useable phase modulated signal, even within the baseband of the sensor. This avoids the need for very high carrier frequencies and the consequent need for very fast components.

(152) In another embodiment however the returns from a single pulse pair may be used to derive a phase value for a given section of optical fibre. A pulse pair is launched into the fibre wherein the pulses of the pulse pairs have a frequency difference between them (i.e. F1F2) so as to define a carrier frequency (F1-F2). In this embodiment the carrier frequency is related to the sample rate of the detector so that between successive samples of the backscatter signal received at the detector a signal at the carrier frequency will have evolved by a predetermined phase amount.

(153) The sample rate should be relatively high, so that subsequent samples correspond to overlapping sections of optical fibre such as described above, i.e. a plurality of samples are acquired in time taken for the pulses to move a distance equal to the gauge length in the optical fibre. Thus the sample rate may be of the order of 100 MHz or so. In this case the assumption can be made that the same general acoustic stimulus applies to the subsequent samples.

(154) Conveniently the carrier frequency is chosen in relation to the sampling rate such that a signal at the carrier frequency will have evolved by 90 between subsequent samples. Thus the carrier frequency may be arranged to be a quarter of the sample rate. Thus if the sample rate is 100 MHz the carrier frequency may be controlled to be 25 MHz. The carrier frequency may be controlled as described previously

(155) In this embodiment the configuration of the pulse pairs should be arranged relative to the sample rate such that successive samples are not substantially independent. In other words a reasonable number of scattering sites that contribute to a first sample should also contribute to the subsequent sample. Thus the sampling rate is preferably relatively high, of the order of 100 MHz or so, and the pulses are relatively wide, so that the scattering sites within the fibre from sample to sample do overlap to a reasonable degree. For example, if the pulse pair comprises two pulses that are 10 m long in the optical fibre with no separation between the pulses, i.e. a pulse duration of about 100 ns, then at each sample time the detector will received radiation from a first 5 m set of scattering sites illuminated by the first pulse and a second 5 m set of scattering sites illuminated by the second pulse. If the sample rate is 100 MHz then between samples the location of scattering sites will effectively have moved by 1 m and there will be 4 m of overlap in each of the first and second sets of scattering sites between the two samples.

(156) The subsequent samples can therefore be processed using a similar processing scheme such as described above in relation to FIG. 25.