GB2503692A

GB2503692A - Method of acquiring data with underwater nodes

Info

Publication number: GB2503692A
Application number: GB201211863A
Authority: GB
Inventors: Harry George Dennis Gosling; Arran James Holloway; Roman Lloyd Kingsland
Original assignee: Go Science Ltd
Current assignee: Go Science Ltd
Priority date: 2012-07-04
Filing date: 2012-07-04
Publication date: 2014-01-08
Anticipated expiration: 2032-07-04
Also published as: GB201211863D0; GB2503692B

Abstract

A method of acquiring field data with a plurality of underwater nodes, typically located on the seabed. Each node acquires a position estimate. Each node transmits, on its own assigned frequency, its position estimate, and preferably also a time estimate, to a plurality of other nodes. Transmission may be via acoustic waves. Each node receives position estimates from at least three other nodes. Each node determines a range between itself and the nodes from which it has received position estimates, perhaps based on signal attenuation level, or time of arrival. Each node improves its own position estimate based on the position estimates and ranges of the other nodes, for instance by trilateration, preferably taking into account signal-to-noise ratio. Each node may also receive time estimates transmitted by at least two other nodes and synchronise its clock based on those other time estimates. Field data, such as seismic, temperature, or pressure data, is acquired using each of the nodes.

Description

METHOD OF ACQUIRING DATA WITH UNDERWATER NODES

FIELD OF THE INVENTION

The present invention relates to a method and apparatus for acquiring underwater field data, typically although not exclusively from the seabed. It should be noted that the term "seabed" is used herein as a generic term not limited to the bed of a sea, but including the bed of any body of water such as a sea, lake or river.

BACKGROUND OF THE INVENTION

A method of guiding autonomous seismic sensors to target positions on the seabed using an acoustic navigation system is described in WO2006/l06085. Autonomous underwater seismic acquisition devices are dropped from a vessel and guided to target locations on the seabed by receiving signals from an acoustic navigation system comprising acoustic sources (beacons) on the seabed. The acquisition devices have a hydrodynamically shaped body and are steered towards their target positions using rudders or by transversely displacing the battery within the device to displace the centre of gmvity of the device and therefore alter its trajectory. The beacons transmit idtrasonic pulses at known times, using timepieces synchronised to a common time reference system. The navigation system may aJso use UPS acoustic relay buoys to communicate with the beacons on the seabed and equipment on the surface. An autonomous underwater node may be used to determine the exact positions of the beacons and the acquisition devices.

When the acquisition devices have all been located on the seabed, the beacons are released from their ballast, by remote control, and return to the surface. When the seismic acquisition is complete, the acquisition devices return to the surface.

Accurate estimates of time and position are required at each acquisition device on the seabed to enable an analysis of seismic data from the devices to yield accurate information on the geology of the seabed.

SUMMARY OF THE INVENTION

The invention provides a method of operating a plurality of underwater nodes, and an underwater node, as set out in the appended claims.

Each node not only transmits its own estimate of position and/or time to other underwater nodes, but also receives estimates of position andlor time from other underwater nodes in order to improve its own estimate. The field data acquired by each sensor can then be associated with the improved position estimate. Similarly a time of acquisition of the field data at each sensor can be determined in accordance with its respective clock and

associated with that field data.

Each node may be a passive device which is deployed to the seabed by dropping it from a surface vessel. However more preferably the node comprises a vehicle with a navigation system and the method further comprises operating each vehicle to deploy the node to a respective deployment location, acquire a series of estimates of its position as it does so, and use the series of estimates to navigate, preferably autonomously, to its respective deployment location. This enables the nodes to be located more accurately in a regular grid or other desired pattern.

The node may glide or sink to its deployment location, but more preferably the node comprises a vehicle with a propulsion system such as one or more propellers which can be operated to deploy the node.

Preferably the node comprises a vehicle with steeling means which is capable of steeling the vehicle to a deployment location, for instance by changing a direction of thrust of its propulsion system, operating a control surface such as a rudder, or transversely displacing a mass such as a battery within the vehicle.

The node's initia' underwater position estimate may be determined in a number of different ways, including but not limited to: acoustic navigation using acoustic sources (beacons) on the seabed as in W02006/106085 and/or acoustic sources (GPS beacons) on the surface of the sea; or inertial navigation using motion sensors and rotation sensors to continuously calculate via dead reckoning the position, orientation, and velocity of the

I

node. The node's initial position estimate is typically determined without reference to the position estimates of other ones of the underwater nodes.

Typically the node comprises a vehicle and a sensor. The sensor may be physically connected to the vehicle both dunng deployment and during field data acquisition. For instance after the node has been deployed to the seabed the sensor may be placed on the seabed next to vehicle to which it is connected by an umbilical data line. Alternatively, after the node has been deployed to the seabed the sensor may be placed on the seabed next to the vehicle and then acquire the field data without being physically connected to the vehicle. In such a case the vehicle may remain next to the sensor during field data acquisition, or it may move away and retrieve the sensor later. Where the sensor and vehicle are physically separated in this way after deployment, then the transmitter and receiver may be part of the sensor or part of the vehicle.

The method may be repeated one or more times, each repeat being based on the improved position or time estimate generated in a previous iteration.

Each node's position estimate may be improved in accordance with a least mean squares algorithm. A recursive nonlinear weighted least mean squares method based on an Extended Kalman filter (EKF) or Unscented Kalman Filter (UKF) could also be used.

Each node may determine signal-to-noise ratios of the signals carrying the three or more other position estimates, and take into account these signal-to-noise ratios when improving its own position estimate.

The clocks may be re-synchronised by taking an average time estimate, or by some other algorithm such as Marzullo's algorithm. Preferably the clocks are re-synchronised based on six or more other time estimates (for instance six neighbours in a hexagonal grid) or eight or more other time estimates (for instance eight neighbours in a square grid).

Typically each node's position or time estimate is improved based on those of only a subset of the other nodes -for instance only its nearest neighbours. In this case the nearest neighbours will also improve their position and/or time estimate on the basis of their nearest neighbours, so improved estimates can be obtained through the entire collection of nodes.

PreferaHy each node's position estimate is improved in accordance with a trilateration algorithm.

Preferably each node's position estimate is improved based on six or niore ranges and other position estimates (for instance six neighbours in a hexagonal grid) or eight or more ranges and other position estimates (for instance eight neighbours in a square grid).

The nodes may transmit their position estimates in response to a request from the node to which they are transmitting. In such a two-way communication process the two way round-trip travel time between the nodes can be used to calculate the range. Alternatively a one-way communication process may be used. For example the nodes may all transmit their position estimates at precisely timed and regular intervals, in which case the one-way trip travel time can be deduced from the time of arrival and used to calculate the range. Alternatively the nodes may transmit their position estimates at different times but with a precisely known transmit signal level which enables the range to be deduced from a level of attenuation of the received signal leveL Communication between the nodes typically uses an encoding technique such as Direct Sequence Spread Spectrum DSSS, Quadrature Amplitude Modulation QAM or Frequency Shift Keying FSK.

Communication between the nodes may use an acoustic communications channel, an Electric Field (EF) communications channel (optionally propagating via the seabed) or any other suitaNe channel.

Preferaffly communication between the nodes is wireless, for instance propagating via water or the seabed.

The field data acquired by the sensor may be for example seismic data, seabed current data, pressure data (for instance to indicate the depth of the seabed), temperature data, salinity data, or dectromagnetic data for electromagnetic classification of hydrocarbon deposits beneath the seabed.

The sensor may be on the seabed during the data acquisition step, or it may be in the water above the seabed during the data acquisition step. k the latter case then the nodes and sensors may be stationary during the data acquisition step, or they move during the data acquisition step (preferably retaining a substantially regular grid formation as they move).

The method can be performed at a minimum by three or four underwater nodes, but more typically it will be performed by more than ten or more than one hundred underwater nodes.

BRIEF DESCRIPTION OF THE DRAWINGS

Embodiments of the invention will now be described with reference to the accompanying drawings, in which: Figure 1 shows an underwater navigation system; Figure 2 shows a method of determining the position of an underwater node from the ranges and positions of the buoys; Figure 3 is a block diagram of the main functional components of a node; Figure 4 shows a gnd of nodes on the seabed, viewed from above; FigureS is a timing diagram showing the relative timings of various signals; and Figure 6 shows a seismic field survey being carried out with a grid of seabed nodes.

DETAILED DESCRIPTION OF EMBODIMENT(S)

Figure i shows an underwater navigation system. Three transmitter buoys la-c are deployed on the surface of the water. Each buoy has a Globa' Positioning System (GPS) antenna 2, a processor 3 and an acoustic antenna 4. The GPS antenna 2 receives UPS data signals 10 from a GPS satellite 11 and from a Differential GPS (DGPS) reference station 12 on a surface vessel 13. The processor 3 of each buoy processes the GPS data signals 10 to determine the position of the buoy la-c in a known manner and transmits the position of the buoy to underwater nodes 40 in an encoded acoustic signal. The nodes 40 each have an acoustic antenna 44 for receiving the acoustic signals from the buoys I a-c and a processor 45. The processor 45 decodes the acoustic signals thereby determining the X, Y and Z co-ordinates of the buoys la-c.

Figure 2 shows how the data X, Y and Z co-ordinates of the buoys are used by each node to determine its position. In step 80 a raytracer algorithm determines a radial distance 81 in accordance with the ray travel time 77, a stored set of sound velocity profile data 82, and the node depth 83 measured by a pressure sensor onboard the node. This ray tracer a'gorithm 80 accounts for the fact that the sound waves will not travel in a straight line from the buoy to the node due to the increase in pressure with depth.

The node now has the radial distance (or range) 81 and position 79 of each one of the IS three buoys la-c. This data is than analyzed by a trilateration algorithm at step 84 to calculate the position 86 of the node. An input to the trilateration algorithm is the velocity 87 of the node (as measured by onboard algorithms which may interpret the data from devices such as accelerometers and/or as calculated based on previous position measurements). This takes into account the fact that the node may have moved between the beginning and end of the encoded acoustic signal, so the output 86 of the algorithm 84 is the position of the node at the end of the encoded signal.

Each node comprise a propulsion system for propelling it through the water comprising a pair of rotary propellers (not shown) each mounted on a thrust motor. Each motor is pivotally mounted so the propeller/motor unit can be independently rotated up and down to vary its angle of thrust relative to the hull of the node. Figure 3 is a block diagram showing the main functiona' elements of the node. An acoustic antenna 44 receives the acoustic signal pulses which are conditioned and analog-to-digital converted by a unit 106a and input to the processor 45 along with clock signa's from a clock 106d.

The processor 45 operates as described above to determine the position of the node. The processor 45 decodes the signals from the GPS buoys la-c to obtain the series of data sets encoded within them and determine the node position. The processor 45 also controls the angle and magnitude of the thrust of the propellers in order to navigate to a desired XY deployment location on the seabed, the desired deployment location having been previously stored in the node memory 46.

Although only two nodes 40 are shown in Figure 1 for purposes of simplicity, a large number of such nodes is typically provided (potentially one hundred or more) for the purpose of accurately distributing a grid of seismic sensors over a wide area of the seabed.

Figure 4 shows a square grid of twenty five such nodes 40 on the seabed, viewed from above. In this example the nodes 40 are in a square array but they may be distributed in other ways, for example in a hexagonal grid.

The nodes 40 are deployed to the seabed from the surface vessel 13, and as they are deployed each node acquires a series of estimates of its position using the surface buoys as described above and uses these position estimates to navigate to a predetermined position. Each node also acquires a seabed position estimate of its position on the seabed, either after it has reached the seabed or shortly beforehand. This position estimate is stored in its local memory 46.

Each node 40 is assigned a transmit acoustic frequency, and has prior knowledge of the frequencies of its eight nearest neighbours in the array. After they have been deployed and are stationary on the seabed, each node broadcasts via its acoustic antenna 44 an update request on its assigned transmit frequency and records the time [trequest)] of transmission in accordance with its respective clock. On receipt of the update. request, the eight nearest neighbours in the array each use their acoustic antenna 44 to broadcast their respective seabed position estimate [xn,yn] along with their local time [tn(response)] and clock confidence interval [ci.

Each node receives and stores seabed position estimates (jxl,ylj, x2,y2j Lxn,ynj) local times ([tl(iesponse)], [t2(response)] [tn(response)]) and clock confidence intervals ([ci], [c2] [cn]) from its neighbours in its local memory 46. The nodes at the centre of the grid have eight neighbours (so n=8) whereas the nodes at the edge of the grid have only three or five neighbours (so n=3 or 5). Each node records the time of receipt of the various responses ([tl(receipt)], [t2freceipt)] [tn(receipt)]) based on its own clock.

This process is then repeated and averaged to obtain a best estimate of the round trip duration Tn=tn(receipt) -t(request) for each neighbour.

Based on a suitable acoustic propagation model (most simply a speed of sound V in water), apriori knowledge of all the processing latencies L within the nodes, and the round trip duration Tn, each node calculates and stores the ranges ([rI], [r2] [rn]) of the neighbours as ni= 0.5(Tn-L)V. where Tn-L is the two-way time of flight (TOF) between the nodes.

Each node then improves its own seabed position estimate (Ix, y I) based on the ranges ([rl j, [r2] [rn]) and position estimates ([xl, y 1], [x2, y2] [xn, yn]) of its n neighbours. This is formulated as a trilateration problem as follows: ri = (x-xl)2 -(y -vi)2 r2= (x-x2)2 -(y-v2)2 iii = (x-ni)2 -(y -yn)2 and solved for x and y using a weighted least mean squares method. The weight matrix W can be based on the signal to noise ratio (SNR) of the received signals where high SNR and therefore more reliable signals have a greater influence in the position estimation: Ap = /) (A'wA)p = A'Wb p = inv(A'W.4)ATb where: x p= y x +y -2x1 -2yl 1 -2x2 -2v2 1 -2xn -2yn 1 wI 0 0 o w2 0 W= o o *..

r12 -x12 -y12 r22 -x22 -v22 2 2 in -xii -yn A recursive non-linear weighted least mean squares method based on an Extended Kalman filter (EKF) or tJnscented Kalman Filter (UKF) could also be used.

Aggregating position estimates from a large number of other nodes in this way has the effect of reducing random error in each node's estimate of its own position. After each node in the fleet has updated its own seabed position estimate, the process is repeated one or more times, each repeat being based on the improved seabed position estimates generated in a previous instance of the process. The process is repeated either a predetermined number of times, or until the seabed position estimates cease to change significantly between iterations.

The algorithm described above operates in two dimension (x,y) only. Optionally the nodes may transmit three-dimensional seabed position estimates ([xl,yl,zl], [x2,y2.z2] [xn,yn,zn]) including their z-coordinate based on a measurement of pressure, and the trilateration algorithm generalized accordingly to give an improvement of the z-coordinate as well as the x and y coordinates.

The clocks 106d of the nodes are synchronised before they are deployed. However the temperature gradient experienced during descent can cause the clocks to drift. As well as improving its own seabed position estimate (x, y) by the process described above, each node also re-synchronizes its own clock with its eight nearest neighbours on the seabed by the following process.

Each node has a two-way round trip duration Tn=tn(receipt) -trequest); a response time tn(response); and a clock confidence interval c for each neighbour. The round trip time Tn=Ll-l-L2+TOF. where: * LI is the outbound latency between t(request) and t(response); * L2 is the inbound latency between t(response) and t( eceipt); and * TOP is the two-way acoustic time of flight between the nodes Figure 5 shows these parameters, apart from c.

The clock difference between the node and its neighbour is: Adockn = tnreceipt)-tnresponse) -(0.5(Tn -LI -L2) + L2) If the clocks are perfectly synchronized then Aclockn=0.

Using the neighbour's clock difference QAdocknj) and dock confidence interval (jcnj) and the current time as read from the node clock t, the node can generate a time range sn = It + Aclockn -cn. t -i-Aclockn + cnl for each neighbour.

Since tn(receipt), tn(response), Tn, Li and L2 are all known, each node can calculate the time ranges for all of its neighbours (IsI], 1s21 [snj). These time ranges are then used as input to an implementation of Marzullo's algorithm and used to update the node's clock. Marzullo's algorithm is described iii K. A. Marzullo. Maintaining the Time in a Distributed System: An Example of a Loosely-Coupled Distributed Service. Ph.D.

dissertation. Stanford University, Depattrnent of Electrical Engineering, February 1984.

The best estimate is taken to be the smaflest interval consistent with the argest number of sources. So if for example the node has two time estimates [1200.OS,1200.l2] and [1200.11,1200.13] from neighboring nodes and its own estimate of time is 11200.10,1200.121, then the algorithm outputs an interval 11200.11,1200.121 which is consistent with all three time estimates and then takes the midpoint 1200.115 of that interval as its improved estimate of time.

After each node in the fleet has updated its own clock, the process is repeated one or more times, each repeat being based on the improved clock settings generated in a previous iteration of the process. The process is repeated either a predetermined number of times, or untfl the clock offset va'ues tend towards zero, or until a seismic survey starts.

After each node's seabed position estimate and clock have been adjusted as descnbed above, the nodes are then ready to acquire seismic data by the process shown in Figure 6.

The seismic survey is carried out by transmitting an acoustic pulse 121 from the surface vessel 13. Seismic waves 122 fmm the seabed 12 are then received and recorded by seismic sensor packs 51.

Each node 40 comprises a sensor pack 51 and a vehicle 52. The sensor pack 51 contains a hydrophone sensor and a set of three orthogonally arranged geophone sensors. The sensor pack 51 is transported by the vehicle 52 to the seabed and then placed or dropped onto the seabed next to the vehicle 52 connected via a data line 53. The seismic data from the sensor pack 51 is transmitted to the processor 45 via the data line for storage onboard the vehicle in the memory 46 and/or transmission to the surface node 13. For each of the nodes on the seabed, the seismic data is stored and associated with that node's seabed position estimate and loca' time of receipt based on the node's clock. The more accurate estimates of loca' time and position enable an analysis of seismic data from the fleet of nodes to yield more accurate information on the geology of the seabed.

In the example of Figure 6 the processor 45, memory 46, dock 406d, antenna 44 and all other elements of the node shown in Figure 3 (apart from the sensor pack 51) are part of the vehicle 52 which remains connected to the sensor pack 51 by the data line 53 during data acquisition. However in an alternative embodiment some or all of these elements may be integrated with the sensor pack 51 instead of the vehicle 52 so that the vehicle 52 can swim away after placing the sensor pack 51 and other elements on the seabed, and then retrieve them after seismic data acquisition is complete.

In the example of Figure 4 all nodes 40 in the fleet transmit both their position and time data to other nodes, and all nodes in the fleet also receive and store position and time data from other nodes. Although this is preferred, it should be noted that this level of functionality is not strictly essential for all nodes in the fleet. For instance some nodes in the fleet may receive but not transmit; and/or some nodes in the fleet may transmit but not receive; and/or some nodes in the fleet may neither transmit nor receive to/from other nodes.

In the example of Figure 4 each node in the fleet is assigned a unique transmit acoustic frequency. If insufficient bandwidth is available to assign a unique frequency to each node in the fleet, then frequencies may be shared between nodes as long as those nodes are sufficiently far apart when they are on the seabed to avoid interference. Also, transmissions on the same frequency can be separated in time.

Although the invention has been descnbed above with reference to one or more preferred embodiments, it will be appreciated that various changes or modifications may be made without departing from the scope of the invention as defined in the appended claims.

Claims

CLAIMS1. A method of acquiring field data with a plurality of underwater nodes, each node comprising a sensor, the method comprising: a. for each node acquiring a position estimate which is indicative of its underwater position; b. transmitting from each node its position estimate to a plurality of other ones of the nodes; c. receiving at each node three or more other position estimates from three or more other ones of the nodes; d. determining a range between each node and the three or more other ones of the nodes from which it has received position estimates; e. improving each node's position estimate based on the three or more other position estimates and the three or more ranges; and f. after step e. acquiring field data with the sensor of each node.
2. The method of claim I wherein the method further comprises prior to step a.operating each node to deploy to a respective deployment location, acquire a series of estimates of its position as it does so, and use the series of estimates to navigate to its respective deployment location.
3. The method of any preceding claim wherein the range is determined between the receiving node and each one of the three or more other nodes based on a signal attenuation level and/or a time of arrival of its respective position estimate.
4. The method of any preceding claim further comprising repeating steps b-c one or more times, each repeat being based on the improved position estimates generated in a previous instance of steps b-c.
5. The method of any preceding claim wherein each node determines signal-to-noise ratios of the signals calTying the three or more other position estimates, and takes into account the signal-to-noise ratios when improving its position estimate.
6. The method of any preceding claim wherein each node's position estimate is improved based on the position estimates and ranges of only a subset of the other nodes.
7. The method of claim 6 wherein the subset includes the nearest neighbours of each node.
8. The method of any preceding claim further comprising: transmitting from each node a time estimate based on a clock of the node to a plurality of other ones of the nodes; receiving at each node two or more other time estimates which have been transmitted from two or more other ones of the nodes; and synchronising each node's clock based on the two or more other time estimates received at the node.
9. An underwater node comprising: a. a system for obtaining a position estimate of the node's underwater position; b. a memory for storing the position estimate; c. a transmitter for transmitting the node's position estimate to a plurality of other underwater nodes; d. a receiver for receiving three or more other position estimates from three or more other underwater nodes; e. a processor programmed to: i. determine a range between the node and the three or more other nodes from which it has received position estimates.ii. improve the node's position estimate based on the three or more other position estimates from the receiver and the determined ranges, and iii. update the memory with the improved position estimate; and I a sensor for acquiring fie'd data.
10. The node of claim 9 further comprising: a clock programmed to generate a time estimate, wherein the transmitter is arranged to transmit the clock's time estimate to a plurality of other nodes and the receiver is arranged to receive two or more other time estimates from two or more other ones of the nodes, and wherein the processor is programmed to synchronise the dock based on the two or more other time estimates received from the other nodes.
11. Four or more underwater nodes according to claim 9.
12. A method of acquiring field data with a plurality of underwater nodes, each node comprising a clock and a sensor, the method comprising: a. transmitting from each node a time estimate based on its own clock to a plurality of other ones of the nodes; b. receiving at each node two or more other time estimates which have been transmitted from two or more other ones of the nodes; c. synchronising each node's dock based on the two or more other time estimates received in step b; and d. after step c. acquiring field data with the sensor of each node.
13. The method of claim 12 further comprising repeating steps a-c one or more times, each repeat being based on the improved time estimates generated in a previous iteration of steps a-c.
14. The method of claim 12 or 13 wherein each node's dock is synchronised based on the time estimates of only a subset of the other nodes.
15. The method of claim 14 wherein the subset includes the nearest neighbours of each node.
16. The method of claim 12 wherein the method further comprises prior to step a.operating each node to deploy to a respective deployment location, acquire a series of estimates of its position as it does so, and use the series of estimates to navigate to its respective deployment location.
17. An underwater node comprising: a. a clock programmed to generate a time estimate; b. a transmitter for transmitting the clock's time estimate to a plurality of other underwater nodes; c. a receiver for receiving two or more other time estimates from two or more other underwater nodes; d. a processor programmed to synchronise the clock based on the two or more other time estimates received from the other underwater nodes; ande. a sensor for acquiring field data.
18. Three or more underwater nodes according to claim 17.