Description of the Prior Art Multichannel receivers are used to survey objects by sampling waves scattered or reflected by the objects in order to generate a"beam", where a beam contains information on the waves scattered towards the receiver from within the"beam pattern" of the receiver. From the beams, an output can be generated for analysis, for example an image of the object being surveyed may be generated from the beams for output on an appropriate display. A variety of different types of waves may be sampled by appropriate multichannel receivers such as acoustic, electromagnetic and seismic waves. For example, acoustic/sound waves may be sampled by a sonar multichannel receiver, as is typically used for underwater surveying of objects submerged within the water, surveying of the seabed, etc. Ultrasound is another example of an acoustic wave which finds application in medical and veterinary examination as well as in the non-destructive testing of materials. Alternatively, various types of electromagnetic radiation may be sampled by a radar multichannel receiver, for example to detect objects in the air, for example aeroplanes.

Multichannel receivers consisting of a plurality of separate receiver channels, or receiver elements, are often used instead of block receivers, because they provide the opportunity to steer the beam electronically, without having to turn the receiver physically. Further, a block receiver can only be constructed so that it can focus at one range, whereas a multichannel receiver can be focused electronically at any range.

Until recently, the quality of the beam pattern produced by a multichannel receiver has generally been considered adequate for most applications. However, the quality of the beam pattern is a limiting factor in high resolution imaging, and the need for higher and higher resolution imaging is now becoming more acute. Considering the example of sonar-based multichannel receivers, examples of applications where higher

resolution imaging is required are synthetic aperture sonar, and auto focusing techniques for sonar, auto focusing being part of the process that may be performed by the beam forming logic of a multichannel receiver system. Similarly, the quality of the beam pattern is important in ultrasound applications however, the application to ultrasound of the synthetic aperture approach has not yet met with acceptance in clinical environments.

Since a multichannel receiver consists of a plurality of separate receiver channels, and accordingly there are multiple receive paths within the system, then differences in characteristics of the electronic components within the various receive paths, along with the physical tolerances with which the array is assembled, can introduce errors into the beams ultimately formed from the data sampled by the various receive paths. One way of seeking to alleviate this problem is clearly to use higher quality components manufactured with low tolerances with the aim of ensuring consistency between the various receive paths. However, it will be appreciated that this will only partially alleviate the problem, and significantly increases the cost of such multichannel receivers.

Summary of the Invention Viewed from a first aspect, the present invention provides a method of calibrating a plurality of receiver channels in a multichannel receiver, comprising the step of : (i) sampling at each of said receiver channels a transmitted wave having a predetermined frequency range in order to generate calibration data for each of said receiver channels ; (ii) for each said receiver channel, transforming the corresponding calibration data into frequency domain calibration data; (iii) for each of a plurality of frequencies within the predetermined frequency range, calculating the phase difference between each receiver channel's frequency domain calibration data and the corresponding frequency domain calibration data of a reference channel in order to determine a phase profile for each said frequency; and (iv) for each said frequency, creating a phase error profile by removing from the corresponding phase profile any phase difference resulting from the timing differences of the wavefront of the transmitted wave reaching each of said receiver channels, so as to generate a set of phase error profiles which together form a correction matrix identifying for each frequency the phase error between receiver channels.

In accordance with the present invention, a calibration technique is performed in order to generate a correction matrix identifying the phase errors between various

receiver channels at particular frequencies. This then enables that correction matrix to be applied to real survey data subsequently obtained by the multichannel receiver in order to correct for the phase errors between the various receiver channels. The difference in operating characteristics of the various electronic components within the various receive paths of the multichannel receiver, along with the physical tolerances with which the array is assembled, are the predominant cause of the phase errors between channels, and accordingly by enabling these phase errors to be corrected, the quality of the beam pattern produced is significantly improved.

In accordance with the present invention, a transmitted wave is generated for reception by the multichannel receiver, this transmitted wave have a predetermined frequency range which preferably covers the entire frequency range in which the multichannel receiver is intended to operate. The sampling of this transmitted wave by the various receiver channels of the multichannel receiver results in the generation of calibration data for each receiver channel. In accordance with the present invention, this calibration data for each receiver channel is transformed into the frequency domain, and then for a plurality of frequencies within the frequency range, the phase difference between each receiver channel's frequency domain calibration data and the corresponding frequency domain calibration data of a reference channel is determined in order to determine a phase profile for each frequency. At this point, the phase profile will contain both a component which represents the phase errors introduced by the various receive paths of the multichannel receiver, and will also typically include a component resulting from the fact that the wavefront will not necessarily have been received by each receiver channel at exactly the same time. For example, if a point like source is used to generate the transmitted wave used in the calibration process, then that will result in the generation of a transmitted wave having a curved wavefront. In preferred embodiments, the multichannel receiver is planar, with the various receiver channels being distributed along that plane, and accordingly in such situations it is clear that the wavefront will not be received by each receiver channel at the same time.

Accordingly, in accordance with the present invention, the calibration method further comprises the step, for each frequency, of creating a phase error profile by removing from the corresponding phase profile any phase difference resulting from the

timing differences of the wavefront of the transmitted wave reaching each of the receiver channels. This results in the generation of a set of phase error profiles which together form a correction matrix identifying for each frequency the phase error between receiver channels. As discussed above, this correction matrix can then be used to correct subsequently obtained survey data in order to produce a more accurate beam pattern from the multichannel receiver array.

It will be appreciated that the receiver channels of the multichannel receiver may be arranged in a variety of configurations. However, in preferred embodiments, the receiver channels are displaced along an axis with respect to each other in order to form a planar multichannel receiver. Further, the transmitted wave used in the calibration process can take a variety of forms. For example, it is possible that the transmitted wave may have a planar wavefront. Considering the example of a planar multichannel receiver, if the transmitter is arranged such that the wavefront is parallel to the planar axis of the multichannel receiver, this could result in the virtual elimination of the timing differences of the wavefront reaching each individual receiver channel. Nevertheless, it would be appreciated by those skilled in the art that it would be extremely difficult to ensure that the wavefront was truly planar, and truly parallel to the multichannel receiver such that no timing differences existed, and accordingly in order to obtain the most accurate set of phase error profiles to be used for the correction matrix, it would still be appropriate to seek to determine and remove any phase differences resulting from the small timing differences of the wavefront of the transmitted wave reaching each of the receiver channels.

In preferred embodiments, instead of trying to generate a planar transmitted wave, the transmitted wave is instead arranged to have a curved wavefront, such as may simply be generated by a point like transmitter, and the step (iv) of the calibration method comprises the step of, for each said frequency, generating the phase error profile by removing a parabolic component from the phase profile, the parabolic component being dependent on the curvature of the wavefront. It can be seen that this approach in effect removes the range dependent component of the phase response indicated by the phase profiles generated in step (iii) of the calibration method.

It will be appreciated by those skilled in the art that, considering the example of a point like transmitter, the parabolic component could be derived simply from a knowledge of the distance of the point like transmitter from the multichannel receiver.

This parabolic component could then be removed from each phase profile to produce the corresponding phase error profile. However, in preferred embodiments, it is found that there is some dependence of the parabolic component on the frequency, and accordingly in preferred embodiments the parabolic components are estimated for each frequency, this further improving the accuracy of the phase error profiles incorporated within the correction matrix resulting from the calibration method. In preferred embodiments, the parabolic component is estimated using a least mean squares fit technique.

It will be appreciated that the calibration data can be stored in a variety of formats. However, in preferred embodiments, the calibration data generated by the receiver channels of the multichannel receiver upon receipt of a transmitted wave is stored in complex vector format, and the method further comprises the step of converting the correction matrix into a unit complex vector correction matrix. In such embodiments, any sample survey data subsequently obtained by the multichannel receiver is also preferably stored in complex vector format, which facilitates a simpler application of the unit complex vector correction matrix to the survey matrix formed from the survey data. The data is preferably stored in complex vector format because it typically enables the data to be stored in less memory.

As will be appreciated by those skilled in the art, a simple comparison of the frequency domain calibration data of particular channels with that of a reference channel will produce a phase profile which identifies relative phase changes. In such a form, there will typically be discontinuities in the phase profile resulting from transition of the phase across the 2B boundary. In preferred embodiments, it is desirable to remove such discontinuities, and accordingly in preferred embodiments, step (iii) of the calibration method further comprises the step of : for each frequency, unwrapping the phase differences across the receiver channels in order to produce absolute phase differences for inclusion in the corresponding phase profile.

In addition to discontinuities in the relative phase values resulting from a transition across the 2B boundary, it is also possible that discontinuities may appear if

any of the receiver channels are dysfunctional, for example because they are not operational, or are producing spurious data due to malfunctioning components within the corresponding receive path. Accordingly, in preferred embodiments, when performing such"unwrapping", the method further comprises the step of : identifying any dysfunctional receiver channels and omitting the corresponding phase differences for any dysfunctional receiver channels from the unwrapping step. Furthermore, in preferred embodiments, such dysfunctional receiver channels are also omitted from the estimation of the parabolic component used in step (iv) of the calibration method.

In certain implementations, the multichannel receiver system is arranged to "baseband"received survey data, this in effect shifting the received survey data in the frequency domain. This is typically done to shift the survey data to a lower frequency, as this tends to reduce the memory space taken up by the data without losing any information, and reduces the required processing power. In such embodiments, the calibration method preferably further comprises the step of, prior to step (ii) of the calibration method, frequency shifting to a lower frequency the calibration data for each receiver channel. This ensures that the correction matrix ultimately produced by the calibration method can be more quickly applied to the basebanded survey data subsequently obtained by the multichannel receiver.

It will be appreciated that a suitable correction matrix may be obtained by performing the calibration method on a single transmitted wave, and that in certain implementations this may provide sufficient accuracy. However, in preferred embodiments, at said step (i) a plurality of transmitted waves are sampled at each of said receiver channels in order to generate a plurality of sets of calibration data, each set comprising calibration data for each of said receiver channels for a corresponding transmitted wave, and said steps (ii) and (iii) are repeated for each set of calibration data.

This results in the generation of a plurality of phase profiles for each frequency which can then be used in the generation of the correction matrix.

More particularly, in preferred embodiment, prior to said step (iv), the method further comprises the step of : for each frequency, generating an averaged phase profile from the corresponding phase profiles determined for each set of calibration data.

Accordingly, in such embodiments, the phase error profiles generated in step (iv) of the process are formed from the averaged phase profile for each frequency.

However, it will be appreciated that it is not necessary to average the phase profiles prior to step (iv) of the process, and alternatively it would be possible to generate a plurality of sets of phase error profiles, and hence a plurality of correction matrices, and these could then be averaged to produce a final correction matrix. Hence, in such embodiments, said step (iv) is repeated for each set of calibration data in order to generate a plurality of sets of phase error profiles, and the method further comprises the step of : for each frequency, generating an averaged phase error profile from the corresponding phase error profiles in each set, so as to produce a final set of phase error profiles which together form the correction matrix.

As mentioned previously, the multichannel receiver may be used in a variety of implementations. In preferred embodiments, the multichannel receiver is a multichannel sonar receiver, and the transmitted wave is an acoustic wave. It will be recognised that the term acoustic wave is also intended to encompass ultrasound such as is used in medical, veterinary and non-destructive testing applications. However, alternatively, the multichannel receiver may be a multichannel radar receiver, and the transmitted wave in such instance would be an electromagnetic wave.

Further, it will be appreciated that the receiver channels of the multichannel receiver may take a variety of forms. However, in preferred embodiments, the receiver channels are piezoelectric receiving elements. In such embodiments, the step (i) of the calibration method includes the step of converting the output of each piezoelectric receiving element into a digital signal used as the calibration data. Typically, the piezoelectric receiving element will produce an analogue voltage signal which is then passed through a digitiser to produce a digital signal used as the calibration data.

It will be appreciated that the reference channel whose frequency domain calibration data is used in step (iii) of the calibration method could be an entirely separate receiver channel to those used in the multichannel receiver. However, in preferred embodiment, the reference channel is one of the receiver channels of the multichannel receiver.

Viewed from a second aspect, the present invention provides a method of correcting survey data produced by a plurality of receiver channels in a multichannel receiver from sampling of a wave incident on the multichannel receiver, comprising the step of storing a correction matrix obtained in accordance with the method according to the first aspect of the present invention; determining the number of samples in the survey data; modifying the correction matrix in the frequency domain to produce a modified correction matrix having the same number of samples as the survey data; for each said receiver channel, transforming the corresponding survey data into frequency domain survey data to form a survey matrix of frequency domain survey data; and applying the correction matrix to the survey matrix to correct for the phase errors between the receiver channels.

By this approach, it will be appreciated that the basic correction matrix can be used to correct survey data of varying lengths, and with a varying number of samples.

In preferred embodiments, the survey data generated by the receiver channels of the multichannel receiver is stored in complex vector format, the correction matrix is a unit complex vector correction matrix, and the step of applying the correction matrix to the survey matrix comprises multiplying the survey matrix by the correction matrix.

Viewed from a third aspect, the present invention provides a computer program product operable to configure a computer to perform a method of calibrating a plurality of receiver channels in a multichannel receiver in accordance with the first aspect of the present invention.

Viewed from a fourth aspect, the present invention provides a computer program product operable to configure a computer to perform a method of correcting survey data produced by a plurality of receiver channels in a multichannel receiver in accordance with the second aspect of the present invention.

Viewed from a fifth aspect, the present invention provides a calibration unit for calibrating a plurality of receiver channels in a multichannel receiver, the receiver channels being arranged to sample a transmitted wave having a predetermined frequency range in order to generate calibration data for each of said receiver channels, and the calibration unit comprising: transforming logic arranged, for each said receiver channel, to transform the corresponding calibration data into frequency domain calibration data;

phase difference determination logic arranged, for each of a plurality of frequencies within the predetermined frequency range, to calculate the phase difference between each receiver channel's frequency domain calibration data and the corresponding frequency domain calibration data of a reference channel in order to determine a phase profile for each said frequency; and a correction matrix generator arranged, for each said frequency, to remove from the corresponding phase profile any phase difference resulting from the timing differences of the wavefront of the transmitted wave reaching each of said receiver channels, so as to generate a set of phase error profiles which together form a correction matrix identifying for each frequency the phase error between receiver channels.

Viewed from a sixth aspect, the present invention provides a multichannel receiver system, comprising: a multichannel receiver having a plurality of receiver channels; means for storing a correction matrix generated by a calibration unit in accordance with the fifth aspect of the present invention; data correction logic for correcting survey data produced by said plurality of receiver channels from sampling of a wave incident on the multichannel receiver, the data correction logic being arranged to perform the steps of : (a) determining the number of samples in the survey data; (b) modifying the correction matrix in the frequency domain to produce a modified correction matrix having the same number of samples as the survey data; (c) for each said receiver channel, transforming the corresponding survey data into frequency domain survey data to form a survey matrix of frequency domain survey data; and (d) applying the correction matrix to the survey matrix to correct for the phase errors between the receiver channels.

In preferred embodiments, the multichannel receiver system further comprises: beam forming logic for receiving the survey matrix as corrected by the data correction logic and for forming a chosen beam. As will be appreciated by those skilled in the art, by taking temporally displaced samples from the various receiver channels, it is possible to steer the beams, such that the beams generated provide data about objects viewed from a particular desired angle with respect to the multichannel receiver.

Brief Description of the Drawings

The present invention will be described further, by way of example only, with reference to preferred embodiments thereof as illustrated in the accompanying drawings, in which: Figure 1 is a block diagram of a multichannel receiver system in accordance with an embodiment of the present invention; Figures 2A and 2B are flow diagrams illustrating the calibration technique performed by the calibration logic of preferred embodiments; Figures 3A to 3C illustrates how the phase response is unwrapped, and the parabolic component removed, across the various channels in order to produce a phase error profile for a particular frequency; Figure 4 is a flow diagram illustrating the correction process applied by the data correction unit of preferred embodiments of the present invention; Figure 5 is a diagram illustrating a chirp signal used for calibration of the multichannel receiver; and Figure 6 is a diagram illustrating parameters used during the calculation of the parabolic component by the calibration logic of preferred embodiments of the present invention.

Description of Preferred Embodiments For the purposes of describing a preferred embodiments of the present invention, a multichannel sonar receiver will be described, which is arranged to sample acoustic waves. Whilst a multichannel sonar receiver is described, those skilled in the art will recognise that the description is applicable to a multichannel receiver intended to operate with radar or ultrasound, subject of course to the difference in operating frequencies. In an example deployment, the multichannel receiver would be placed underwater, and associated with a transmitter which would be arranged to periodically emit a transmit signal, or"chirp", of varying frequency. This transmit wave would scatter off objects within the water, and the scattered waves would be sampled by the multichannel receiver. It will be appreciated that in addition to using the multichannel receiver system to detect objects in the water, this system could also be used to survey surfaces under the water, for example the seabed.

In preferred embodiments, in order to increase the quality of the beam pattern that is produced from the array of receiver channels forming the multichannel receiver, the multichannel receiver is first calibrated to form a correction matrix identifying for each frequency the phase error between receiver channels.

Figure 1 is a block diagram illustrating a multichannel receiver system 100 in accordance with preferred embodiments. The multichannel receiver system 100 comprises a multichannel receiver 10 consisting of a plurality of receiver elements or channels 20, which in preferred embodiments are spaced apart from each other along an axis to form a planar array of receiver elements. Although in Figure 1 the receiver elements 20 are shown abutting each other, it will be appreciated by those skilled in the art that they may actually in practice be spaced apart from each other. In practice, the spacing between the receiver elements 20 is dictated by the degree of beam steering required by the multichannel receiver system and the frequency range of the acoustic waves to be detected by the multichannel receiver system. Further, it will be appreciated that the size of the multichannel receiver, and the number of receiver elements within the multichannel receiver can be varied, depending on the application. Generally speaking, the longer the length of the multichannel receiver 10, the higher the resolution of the beam pattern that can be obtained. As will be appreciated by those skilled in the art, such multichannel sonar receivers can be used to detect objects over a wide range of distances, from several metres to several kilometres.

In preferred embodiments, each receiver element 20 is formed from a piezoelectric device, which is arranged to output an analogue voltage relative to the acoustic wave acting on that receiver element. In preferred embodiments, each such piezoelectric receiving element 20 has a digitising element 50 associated therewith to convert the output from the piezoelectric element 20 into a digital signal for storage within the storage unit 55. In preferred embodiments, each digitiser 50 consists of a pre- amplifier circuit and an analogue to digital (A/D) converter. The sampling rate of the digitiser can be chosen as appropriate, but in preferred embodiment is at least twice that of the highest frequency in the chirp signal.

Whilst the bank of digitisers 50 can be packaged separately to the multichannel receiver 10, in preferred embodiments the bank of digitisers 50 is packaged with the

piezoelectric receiver elements 20 of the multichannel receiver 10, such that a single housing exists that is arranged to output a sequence of digital signals to the storage unit 55 indicative of the samples detected by the various receiver elements 20. Each receiver element 20 and its corresponding digitiser 50 can be considered to comprise a receive path, and as will be discussed in more detail later, the preferred embodiments of the present invention provide calibration logic 60 for calculating phase errors introduced due to the differing characteristics of electronic components making up the various receive. paths, and/or due to the physical construction tolerances.

The storage unit 55, which may for example be formed by the hard disk of a computer, is arranged to store for subsequent processing the digital data output from the bank of digitisers 50. When the multichannel receiver system 100 is deployed, the multichannel receiver 10 will be arranged to sample reflected acoustic waves from objects being surveyed resulting from the reflection by those objects of an acoustic chirp signal issued by a corresponding transmitter. This will result in the storage unit 55 receiving a series of digital data from each receive path, which collectively is stored as a set of survey data. It will hence be appreciated that through repeated transmissions of the chirp signal, a plurality of sets of survey data will be generated for storage within the storage unit 55.

However, prior to the generation of such survey data, one or more sets of calibration data will be obtained in a similar fashion, but rather than reflecting waves off objects, the calibration data is preferably obtained by placing a transmitter 40, for example a point like source, at a specified distance from the multichannel receiver 10, and then using a waveform generator 45 to generate a predetermined chirp signal for transmission by the transmitter 40 towards the multichannel receiver 10. In preferred embodiments, the multichannel receiver 10 is a wideband multichannel receiver responsive to acoustic waves within a predefined frequency range, for example 100 KHz to 200 KHz. An example of a suitable calibration chirp signal that may be generated by the waveform generator 45 is provided in Figure 5. As can be seen from Figure 5, in preferred embodiments the chirp signal consists of a signal of continuingly varying frequency within the operational range of frequencies of the multichannel receiver 10, the

chirp signal having a duration of time"t". Typically, the duration t will be of the order of a few milliseconds.

In preferred embodiments, this calibration chirp signal is preferably similar to that that will ultimately be issued by the transmitter of the deployed system for reflecting off objects in the water.

By issuing the chirp signal multiple times from the transmitter 40 it will be appreciated that a plurality of sets of calibration data will be received by the storage unit 55 and stored therein. In preferred embodiments, the storage unit 55 is arranged to convert all received data, whether it be calibration data or survey data, into complex vector format.

In the calibration phase, calibration logic 60 is connected into the multichannel receiver system 100 for retrieving the sets of calibration data from the storage unit 55, and performing predetermined calibration routines as will be discussed later with reference to the flow diagram of Figures 2A and 2B. This results in the generation of a unit complex vector correction matrix which is returned from the calibration logic 60 to the storage unit 55 for storage. The calibration logic 60 may take a variety of forms. For example, the calibration logic may be provided by appropriate software running on a computer system, or alternatively may be provided by a suitable Field Programmable Gate Array (FPGA). After the calibration process is complete, the calibration logic 60 can be removed from the multichannel receiver system 100, or alternatively left in place to facilitate recalibration at later dates.

Once the calibration process has been complete, the multichannel receiver system 100 can be deployed so as to be used to obtain sets of survey data for subsequent analysis. In preferred embodiments, the survey data is analysed by creating from the survey data beams which are then converted into a visual display for output to a user to provide an image of the objects scanned by the multichannel receiver 10.

When it is desired to analyse sets of survey data, these sets can be retrieved one at a time by the data correction unit 65, which is arranged to process each set of survey data so as to apply the unit complex vector correction matrix to that survey data to correct for the phase errors between the various receive paths of the receiver channels 20. The manner in which this is accomplished in accordance with preferred embodiments of the

present invention will be described later with reference to Figure 4. Once a set of survey data has been corrected, it is output to a beam forming unit 70 which is arranged to generate a set of beams from the survey data. The operation of such beam forming units will be well understood by those skilled in the art, and accordingly such operation will not be discussed in detail herein. However, in general terms, the beams can be steered electronically so as to provide images of any particular objects of interest, whether or not those objects are directly in front of the multichannel receiver 10, or off at an angle. To generate a set of beams which can be used to provide an image of an object that is off at an angle with respect to the multichannel receiver 10, temporally displaced samples from different receiver elements 20 can be taken by the beam forming unit in order to generate a beam which would be equivalent to that which would have been produced had the multichannel receiver actually been physically moved so as to point towards the object of interest.

Once the particular set of beams has been produced by the beam forming unit 70, it is output to a display generator 75, where it is converted in a known fashion into an image signal for outputting to a display device 80.

The calibration process performed by the calibration logic 60 in preferred embodiments will now be discussed in more detail. Prior to performing the actual calibration process, it is first necessary to collect one or more sets of calibration data, which as discussed earlier with reference to Figure 1 is preferably achieved by use of a point like source 40, which generates a circular transmit wave for output to the multichannel receiver 10. As illustrated by Figure 5, in preferred embodiments a linear frequency modulated chirp is emitted by the transmitter, and each receiver element 20 is then arranged to sample the transmitted wave, this resulting in the generation of a set of calibration data for each chirp signal emitted.

In order to obtain data which can be reliably used for calibration, there are some basic guidelines that should be followed. Firstly, the receive array of the system, namely the receiver elements 20 of the multichannel receiver 10 and the bank of digitisers 50, should be disassembled as little as possible between the calibration and the use of the system. In preferred embodiments, this requirement is achieved by housing both the multichannel receiver 10 and the bank of digitisers 50 within a single housing. It should

be noted that if any elements of the receive array are removed and refitted, recalibration of the system will probably be required.

Secondly, it is important that during the collection of data the transmitter 40 and the receive array should not move relative to each other, and accordingly this is best achieved in a controlled environment. Furthermore, the medium that the calibration is performed in (for example water) should be as still and as homogeneous as possible, which again is best achieved in a controlled environment. Furthermore, reverberation and noise should be minimised, and multi-path should not contaminate the signal received directly from the transmitter. Again, both of these requirements are best met by performing the calibration in a controlled environment. Finally, the transmitter should be placed such that it is within the beam width of all the receiver elements in the multichannel receiver 10.

With the above points in mind, it is preferable that the calibration takes place within a controlled environment, which will typically be at a location different to the intended deployed location of the multichannel receiver system 100. For example, the calibration may preferably take place within a laboratory-type environment, whereas the deployed location of the multichannel receiver system 100 would typically be on board a waterborne vessel, with the multichannel receiver 10 located under the water.

Once one or more sets of calibration data have been obtained and stored within the storage unit 55, then the calibration process to be performed by the calibration logic 60 can begin. In summary, the calibration is performed by calculating the phase error across the receiver channels of the multichannel receiver array as a function of frequency.

The phase error as a function of frequency for each channel is then converted into a unit complex vector matrix. The process performed in preferred embodiments of the present invention will now be described in more detail with reference to Figures 2A and 2B.

In preferred embodiments, each set of calibration data corresponds to a particular "ping", this ping resulting from the chirp issued by the transmitter 40. At step 200, a ping value Pi is set to zero. Then, at step 205, calibration data for each channel for ping Pi is loaded into the calibration logic 60. The process then proceeds to step 210, where it is determined whether basebanding of the calibration data is required. Some multichannel receiver systems 100 are arranged to frequency shift any measured survey

data to a lower frequency prior to using it in the later processing stages of the multichannel receiver system, and such frequency shifting is referred to as basebanding.

Such a basebanding process tends to reduce the memory space taken up by the data without losing any information, and reduces the required processing power. If the multichannel receiver system 100 is intended when deployed to use such a basebanding technique, then it is appropriate that the calibration process also performs such basebanding, and accordingly if basebanding is required, the process proceeds to step 215, where basebanding is applied to the calibration data. The process then proceeds to step 220, or alternatively proceeds directly to step 220 if basebanding is not required.

At step 220, the calibration data is transformed from the time domain to the frequency domain, this being an important step in enabling phase errors to subsequently be calculated as a function of frequency. The process then proceeds to step 230 where a reference channel CR is selected. In preferred embodiments, the reference channel CR is chosen to be one of the receiver channels of the multichannel receiver 10, and in particular preferred embodiments, a receiver channel from the middle of the multichannel receiver is selected as the reference channel CR.

The process then proceeds to step 235, where a frequency fi is set equal to a start frequency fs. In preferred embodiments, there will be a plurality of sample frequencies within the frequency range of the chirp signal which are used in the calibration process ranging from the start frequency fs to a maximum frequency FMAX The process then proceeds to step 240, where a receiver channel value Ci is set equal to zero, thereby identifying the first receiver channel in the multichannel receiver.

The process then proceeds to step 245, where the phase difference is calculated between channel Ci and the reference channel CR, this being achieved by comparing the frequency domain calibration data of channel Ci with the frequency domain calibration data of the reference channel CR.

The process then proceeds to step 250, where the value of Ci is incremented by one, after which it is determined at step 255 whether Ci is equal to N, N being the total number of receiver channels in the multichannel receiver. If Ci does not equal N, then the process loops back to step 245, in order to calculate the phase difference between the new channel Ci and the reference channel. Since Ci was initially set to zero to represent

the first receiver channel, it will be appreciated that the last valid value for Ci will be N-l, and accordingly when Ci is determined to equal N at step 255, the phase difference will have been calculated for every channel at the particular frequency fi.

Accordingly, the process then proceeds to step 260, where the phase differences are unwrapped across the receiver channels to generate a phase profile for frequency fi.

This unwrapping process is illustrated schematically with reference to Figures 3A and 3B. As is apparent from Figure 3A, the phase difference calculated at step 245 is inherently a relative phase difference with respect to the reference channel, and accordingly there will be discontinuities 355 in the phase response 350 due to the transition across the 2B boundary.

As will be appreciated by those skilled in the art, unwrapping is a standard process employed to remove such discontinuities, so as to generate the phase profile 360.

It is this phase profile 360 that is produced by the process performed at step 260. The transition from the phase profile 360 to the phase error profile 370 is performed later by the process illustrated by step 300 in Figure 2B, and will be discussed later.

Since discontinuities in the phase response 350 can also occur as a result of dysfunctional channels, a process step 225 is provided to determine the dysfunctional channels from the calibration data loaded at step 205, and the output from that determination step is also input into step 260, so as to allow any such dysfunctional channels to be omitted from the unwrapping process performed at step 260.

Once the unwrapping process has taken place, the process proceeds to step 265, where the frequency fi is incremented by) f, after which it is determined at step 270 whether fi is greater than fMAx. If fs is not greater than flax, then the process returns to step 240, where the calculation of the phase difference across the various channels relative to the reference channel is reperformed, and the phase differences are then unwrapped to produce a phase profile for that frequency. However, once it is determined at step 270 that fi is indeed greater than flax, the process then proceeds to step 275 illustrated in Figure 2B, where the ping value Pi is then incremented by one. Thereafter, the process proceeds to step 280, where it is determined whether the ping value Pi is equal to the total number of pings PTOTAL, i. e. the total number of sets of calibration data

generated by the system. If Pi does not equal PTOTAL, then the process returns to step 205, to repeat the above described process for a new set of calibration data.

Once it is determined at step 280 that Pi does indeed equal PTOTAL, then the process proceeds to step 285, where the frequency fi is once again set to the start frequency fs.

The process then proceeds to step 290, where the phase profiles generated for frequency fi from the various pings is averaged. The process then proceeds to step 295, where an estimate of the parabolic component is performed.

As is clear from Figure 1, since the transmitter 40 generates a curved wave, the wavefront will not reach each receiver element 20 at the same time, and the phase difference resulting from this phenomenon should be removed from the phase profiles generated by the calibration process in order to produce phase error profiles which identify solely the phase errors occurring between the various receiver channels. The estimate of the parabolic component at step 295 is intended to provide an estimate of the phase difference that occurs as a result of the curvature of the wavefront, so that that component of the phase difference can then be removed at step 300.

Figure 6 is a diagram illustrating some of the parameters used in the calculation of the parabolic component for an estimated location 620 of the transmitter 40. In accordance with preferred embodiments, a time difference) T is calculated for each receiver element 20 of the multichannel receiver 10 using the following calculation: As is apparent from Figure 6, R is the distance 600 between the estimated location 620 of the transmitter 40 and the reference channel of the multichannel receiver 10, the parameter x is the distance between the reference channel and the midpoint of a particular receiver element 20 being considered, and the angle 2 is the angle about the reference channel that the transmitter 40 lies on (i. e. the angle between the normal 610 and the line 600). Further, "c"is the speed of sound.

During a first iteration of the process, a first estimate of the location 620 of the transmitter 40 is made (in preferred embodiments the transmitter 40 will be located in front of the reference channel but is unlikely to be precisely in front).

Once the parabolic component for that estimated location of the transmitter has been calculated using the above equation, the theoretical phase profile that would result is then derived from) T, and the frequency fi, using the following equation: Phase = 2B) Tfi This theoretical phase profile is then compared with the actual averaged phase profile determined at step 290. In preferred embodiments, this process takes account of any dysfunctional channels, and accordingly step 295 is arranged to receive from the process step 225 an indication of any dysfunctional channels. A least means squares fit approach is then used to find the best fit by iterating R and 2 (i. e. to represent different estimated locations 620 of the transmitter 40) and repeating the above process until a best fit is obtained. When the best fit is determined, the corresponding parabolic component is then considered as the estimated parabolic component for the averaged phase profile determined at step 290.

Once the parabolic component has been estimated, the process proceeds to step 300, where the parabolic component is subtracted from the averaged phase profile to produce a phase error profile 370 illustrated schematically in Figure 3C, this phase error profile becoming a component of a correction matrix for frequency fi. As can be seen from a comparison of Figures 3B and 3C, step 300 results in the removal of the parabolic component which otherwise forms a very significant part of the phase profile 360. The perturbations in the phase error profile 370 then represent phase errors resulting from differences in characteristics of the electronic components within the various receive paths of the corresponding receiver channels, and from the physical tolerances associated with construction.

The process then proceeds to step 305, where the frequency fi is incremented by ) f, after which it is determined at step 310 whether fi is greater than fMAX. If fi is not greater than flax, then the process returns to step 290, to repeat processes. 290,295 and 300 for the next frequency.

Once it is determined at step 310 that fi is indeed greater than flax, the process proceeds to step 315, where the correction matrix formed from the components produced for each frequency at step 300 is then converted into a unit complex vector correction matrix, after which the unit complex vector correction matrix is stored at step 320.

Hence, it will be seen that the preferred way of generating the unit complex vector correction matrix has now been described with reference to Figures 2A and 2B.

However, it will be appreciated by those skilled in the art that it is not necessary to perform the steps in exactly the way illustrated in Figures 2A and 2B in order to generate the unit complex vector correction matrix. For example, whilst with reference to steps 235 to 270, it was illustrated that the phase difference was calculated across all the channels frequency by frequency, it will also be appreciated that the same result could be achieved by calculating the phase difference across all frequencies channel by channel.

Similarly, whilst the phase profiles where averaged at step 290, prior to estimation and removal of the parabolic component at steps 295 and 300, it will also be appreciated that the averaging process could take place after estimation and removal of the parabolic components, for example by producing a plurality of correction matrices which are then averaged.

Having described the calibration process performed in preferred embodiments by the calibration logic 60 of Figure 1, the correction process performed in preferred embodiments by the data correction unit 65 of Figure 1 will now be described in detail with reference to Figure 4.

Firstly, the data correction unit 65 obtains the number of samples Si within a particular set of survey data, and also the number of samples Sc within the unit complex vector correction matrix.

Then, the process proceeds to step 405, where the unit complex vector correction matrix Me is loaded into the data correction unit 65. At step 410, the matrix Me is then interpolated along frequency by a factor S,/SC to give a modified unit complex vector correction matrix Mci having the same number of samples as in the set of survey data.

The process then proceeds to step 415, where the ping value Pi is set equal to zero. Then, at step 420 the survey data for each channel for ping Pi is loaded into the data correction unit 65, after which it is determined at step 425 whether basebanding is

required. If it is, then the basebanding is applied at step 430. After application of any required basebanding, the process proceeds to step 435, where the survey data is transformed from the time domain to the frequency domain in order to form a survey matrix MI. In preferred embodiments, the survey data is stored in a complex vector format, and accordingly the survey matrix is in a complex vector format. Hence, when the process proceeds to step 440 to apply the correction, the correction can be applied by multiplying the survey matrix MI by the modified unit complex vector correction matrix Mci.

The process then proceeds to step 445, where the ping value is incremented by one, after which it is determined at step 450 whether Pi is equal to the total number of pings PTOTAL. If not, the process returns to step 420, where the correction is repeated for the next set of survey data. Otherwise, once it is determined that Pi is equal to TOTAL, then the process proceeds to step 455, where the process terminates.

Accordingly, as has been described above, the technique of preferred embodiment allows a beam pattern to be corrected by enabling a calculation of the wideband phase error across the various receive paths of a multichannel receiver, and for those phase errors to then be removed to allow correction of the beam pattern, thereby significantly improving the quality of the beam pattern, and accordingly allowing higher resolution imaging.

As described above, survey data can be corrected by multiplying the survey matrix by the correction matrix in the frequency domain, and in preferred embodiments this can be achieved by adapting the matched filter to perform this operation, therefore requiring no extra processing stages to be introduced to correct the beam pattern of the array. Hence, the data correction unit 65 of figure 1 can in preferred embodiments be embodied within the beam forming unit 70 using pre-existing hardware, with appropriate software being provided to perform the correction process described in figure 4.

Whilst a preferred embodiment of the present invention has been described with reference to a multichannel receiver system arranged to sample reflected acoustic waves from objects resulting from the reflection by those objects of an acoustic chirp signal issued by a corresponding transmitter, it will be appreciated that the described calibration

technique could also be used for passive multichannel receiver systems, for example passive sonar systems which only listen to sound but do not incorporate a transmitter.

Although a particular embodiment of the invention has been described herein, it will be apparent that the invention is not limited thereto, and that many modifications and additions may be made within the scope of the invention. For example, various combinations of the features of the following dependent claims could be made with the features of the independent claims without departing from the scope of the present invention.