ARPA

 

Introduction

 

The IMO Performance Standard for an ARPA requires that it should ‘ . . . reduce the workload of observers by enabling them to automatically obtain information so that they can perform as well with multiple targets as they can by manually plotting a single target’.

It also states:

‘The display may be a separate or integral part of the ship’s radar.  However, the ARPA display should include all the data required to be provided by a radar display in accordance with the performance standards for navigational radar equipment’.

Integral ARPA’s

In the    modern integral ARPA a computer, usually referred to as the processor, is incorporated in the radar/ ARPA system so that the ARPA data etc. can be displayed on the same screen as the conventional radar data. 

When a ship required to be fitted with an ARPA is at sea and a radar watch is being kept on the ARPA, the installation shall be under the control of a person qualified in the operational use of ARPA, who may be assisted by unqualified personnel.

Rate aiding

When the target is first acquired, a large gate is necessary since there is uncertainty as to the direction in which the target will move. 

The radius of the gate is really a measure of confidence in the tracking.

The smaller this value becomes, the more precise the prediction will be.

The advantages of a reduced tracking gate are:

A lower likelihood of target swop

An improved ability to track targets through rain and sea clutter.

An ability to continue tracking, even when target response is intermittent.

One problem which can arise with reduced gate size is that if a target manoeuvres and, as a result, is not found by the computer in the predicted position, the computer may continue to track and look in the predicted direction and end up by losing the target altogether. 

To avoid this possibility, as soon as the target is missed, the gate size is increased. If the target is still detectable and subsequently found, the tracking will resume and a new track will gradually stabilise.

If, after six fruitless scans, the target is still not found then an alarm is activated and a flashing marker is displayed at the target’s last observed position.

The analysis of tracks and the display of data

In either case, if the target is acquired manually or automatically, the ARPA should present, in a period of not more than 1 minute, an indication of the target’s motion trend and display, within 3 minutes, the target’s predicted motion in accordance with the Performance Standard.

Display of target data as specified above are in two levels of accuracy:

A lower level relating to the target’s motion trend, which is an early indication of the target’s relative motion.

A higher level relating to the target’s predicted motion; this means the best possible estimate of the target’s relative and true motion data.


General tracker philosophy

Targets within the filtered area of the memory are selected for tracking when, either manually or automatically, a gate is placed over their responses.  As the aerial sweeps past a ship-target, it will register a number of strikes on successive timebase and it may be that such a target activates more than one successive radial range cell. 

In the case of picture storage these digitised responses will aggregate in the memory to generate on the display an echo having the outline of the distinctive echo paint. Clearly it is neither necessary nor desirable for the computer to track each individual element present in the resolution cell. 

If the target has been acquired, and is being successfully tracked, a tracking window will be centred on that particular memory location within the hit matrix, which corresponds with the target’s range and bearing.  The co-ordinates of the window can be extracted and stored in a further area of the tracker memory.  This area is sometimes referred to as the track file and there will have to be a separate track file for each tracked target.  Thus, rotation by rotation, as the gate moves in steps following the target’s position through the hit matrix, sequential positions of each tracked target can be stored in the appropriate track file.

The processor (which is that part of the computer, which manipulates the data and carries out the mathematical operations) must operate on the recorded positions to calculate the most probable track of the target.  It is difficult to carry out calculations based on positions which are expressed in terms of range and bearing because the rates at which the bearing and range change are not constant for a target on a straight track.  Further, the spatial resolution varies with range (i.e. it is geometrical).  For these reasons it is usual to convert the target positions into Cartesian co-ordinates of North and East.

The effect of inherent errors is that, even for a target on a steady track, the plotted positions do not form a perfectly straight line but are scattered about the correct track; the observer has to attempt to draw the line that is the best fit.  Exactly the same effect occurs with automatic plotting and it is further exacerbated by quantizing errors introduced by the digital storage.

Since the data must eventually be displayed as a stable straight-line vector, the processor must calculate a length and direction, which represents the best fit to the scattered observations. 

When a target is first acquired, the computer will commence storing positions, obtaining updated co-ordinates each time the aerial sweeps across the target.

These positions will have an inherent scatter and initially the mean line will be very sensitive to plots, which fall some distance from it. However, as the plotting duration increases and more plots are obtained, the mean line will stabilise and accuracy will improve. 

During the first minute of tracking the target will normally display only a symbol to indicate that it is being tracked. 

In most systems the vector will be suppressed until sufficient observations have been obtained to produce the indication of the target’s motion trend to the level of accuracy required b the Performance Standard. 

Some systems are designed to display vectors within a few seconds of acquisition.  This should not be seen as a sign of instant accuracy. 

Accuracy demands a number of successive observations and until the one-minute interval has elapsed there is no requirement to meet the Performance Standard accuracy. 

Any data derived directly or indirectly from these very early indications could be highly misleading.  In general, where such early display takes place, a study of the instability of the vector should convince the user that it is based on insufficient observations. 

After one minute the tracker will have smoothed about 12-20 observations and must then produce data to the lower of the two accuracy levels set out in the Performance Standard. The tracking period is allowed to build up to three minutes, at which stage the processor will be able to smooth some 36-60 observations and must then reach the higher accuracy level.

If a target response is not detected in the location forecast by the rate aiding, one possible explanation is that the target has manoeuvred.  The tracking gate will be opened out and if the target is detected, tracking will continue.  If the departure from the three-minute track is not significant, the processor will conclude that the departure was due to scatter and will continue to smooth the track over a period of three minutes.  On the other hand, if the departure is significant, the processor will treat the situation as a target manoeuvre and will reduce the smoothing period to one minute.  This reduction in smoothing period is analogous to the situation in which an observer decides that a target has manoeuvred and therefore discards a previous OA W triangle and starts a new plot.

If steady state conditions resume, low-level accuracy must be obtained within one minute and then the tracking period can again be allowed to build up to 3 minutes, allowing high level accuracy to be regained.

In general trackers will either:

smooth and store the relative track of a target to produce directly the output relative-motion data and hence calculate the true-motion data from the smoothed relative-track data and the instantaneous input course and speed data, which is normally un smoothed to avoid any loss of sensitivity to man oeuvres by the observing vessel; or

smooth and store the true track of a target to produce directly the smoothed true-motion data and reconstitute the relative-motion data from the smoothed true-track data and the (normally un smoothed) input course and speed data.

Note In order to smooth and store true tracks, the normally un smoothed course and speed data are applied to the raw relative-motion data.

In the steady state situation, i.e. where neither tracked target nor the observing vessel man oeuvres and no changes take place in any errors in the input data, both approaches will produce the same result.  If a change takes place, the two different approaches will produce differing results over the succeeding smoothing period.  To understand the differences it is necessary to consider in general terms how the calculations are performed.

If the input data error is constant for the full smoothing period, the smoothed true track will of course similarly be in error.  The computer will then use the wrong input data and the consistently wrong true track and as a result will arrive at the correct relative motion.

It is thus evident that, provided any error in the input course and speed data is consistent for the full smoothing period, it will not affect the accuracy of the CPA/TCPA data. 

However, if there is a fluctuating error, for example due to erratic log input, the relative vector will be inaccurate and unstable.

While recognising the advantage of this approach in ensuring relative data stability during manoeuvres by the observing vessel, many users are concerned about the ability of random input errors to influence the CPA.

Tracking history

The ARPA should be able to display, on request, at least four equally time-spaced past positions of any targets being tracked over a period of at least eight minutes.

This enables an observer to check whether a particular target has manoeuvred in the recent past, possibly while the observer was temporarily away from the display on other bridge duties.

Not only is this knowledge useful in showing the observer what has happened but it may well help him to form an opinion of what the target is likely to do in the future.

Relative history should be used with great caution.

Uneven tracks of targets or apparent instability of motion may be taken to indicate that tracking of that target is less precise than it might be and the displayed data should be treated with caution.

Because of the variations in the way this facility can operate, great care should be taken when observing history to ensure that one is certain of exactly what is being displayed.  In particular, one must establish whether true or relative history is being displayed and also which time spacing are in use.