3. The Basic Design of Centimetre AI Systems
AI systems have to be designed to meet the problems of night interception, which have continually changed throughout the course of the war. Certain basic requirements have, however, remained unchanged; the most important are:-

1. Maximum and minimum range. Best possible.
2. Best possible coverage in azimuth and elevation.
3. As continuous, direct and simple a presentation as possible.
4. Freedom from interference, particularly "window".

To these should always be added the obvious necessity for serviceability, ease )f operation and maintenance, and adaptability to different types of aircraft, which are common requirements for all airborne radar systems.

The variables at the disposal of the designer have been severely limited by standardisation of components and units; briefly the wavelength must be S, X or K band, the transmitter power cannot yet exceed 200 kW peak and has usually to be some standard value less than this; the scanner dimensions are almost certain to be limited by problems of aircraft installation, Quantities such as pulse length, repetition rate and receiver band-width can be chosen within fairly wide limits. In order to see the inter-relation between all the disposable quantities, the following relations are noted.

Let
Transmitter Power                   = Pt
Maximum Range                     = Rmax.
Power gain of aerial system     = G₂ (proportional to γ²/λ² for circular paraboloid)
Receiver band width               = f.(MC/S)
Pulse Length                           = t (μsecs)
Pulse repetition rate                = p sec^-1
Picture repetition period          = T sec
Width of polar-diagram of
aerial system at half power      = θ₀ (∞ γ/λ)

Then considering a target at distance, R from an AI equipped fighter, the energy received per unit time per unit surface of the target ∞ P_TG/R² when the maximum of the beam is pointing at the target, and regarding the target as a uniform radiator, the reflected energy received back at the fighter per cm² per unit time ∞ P_TG/R⁴ and the energy concentrated into the receiver by the aerial system ∞ P_TG² λ²/R⁴ i.e. the effective receiving area of a mirror of gain G being proportional to Gλ². The minimum detectable signal is a definite quantity set by receiver sensitivity and indicator characteristics; when the energy returned from the target reaches this value R = R max. and we therefore have

P_Tγ⁴ / Rmax⁴λ² = Const                                                           (1)

or Rmax = Const P_T^¼γ / λ^¼

Maximum range can therefore be increased slowly by increasing the transmitter power and rapidly by increasing aerial gain; the latter however results in a narrow beam and consequent scanning difficulties. Thus supposing that the area to be scanned is bounded by angular ± α in elevation and ± β in azimuth, the scanning must be arranged so that there are no gaps throughout this volume, i.e. in each cycle of scanning every point must be scanned by a part of the beam whose power should not be less than 80% of that of the maximum of the beam (corresponding to 5% reduction of maximum range and occurring at θ = θ₁ from maximum). The number of scanning lines per complete picture required to give this coverage is proportional either to α/θ₁ or β/θ₁ according to the nature of the scanning and if ω is the angular velocity of the scanner the time taken for a complete picture must be at least

T =   αβ / θ₁ω   αβ / ω                                                   (2)

Now in most centimetre AI systems, the received signals are presented as intensity modulation on a cathode ray tube screen; and the signals from successive pulses are spread out spatially by an amount proportional to the scanning speed. In order to build up an easily observable echo there must be some overlap between successive pulses and the pulse repetition rate must therefore increase proportionately with the scanning speed; it must also increase inversely as the beam width in order that a given number of pulses may be received back from the target as the beam sweeps over it. We therefore have:-

p ∞ ω/θ₁∞ ωγ/λ

The limit to p.r.f. is set either by the permissible dissipation in the transmitter valve or by the possibility of getting echoes back on the second timebase (e.g. a p.r.f. of 2000 gives a maximum timebase of 50 miles and a 55 mile ground return would look like an echo at 5 miles). With present aircraft generator and TR boxes transmitter dissipation becomes the limiting factor with powers of the order of 200 kW.

The picture repetition time T is primarily dictated by the operational necessity for having as continuous a presentation information as possible. In the later stages of an interception very high rates of relative motion between bomber and fighter are possible, and it is found that T = 1-2 secs. is the maximum interval between pictures that is desirable. A faster rate still would be advantageous but would involve large pulse repetition rates and power dissipations.

The direct effect of pulse repetition rate on maximum range, providing that there are sufficient pulses to give a clear picture on the particular type of indicator used, is small and is complicated by. the fact that signals are never steady but often fade out entirely for several seconds. There is also a rapid fading with a period of the order of the pulse rate (1000 per sec.) which is important in some types of automatic computing systems, but in general the precise pulse repetition rate, provided that it is adequate, does not affect AI performance.

Effect of Ground Returns
Ground returns are produced when the scanning beam of radiation strikes the ground sufficiently near the fighter aircraft. The reflecting area of the ground is considerably larger than that of a target aircraft and the response from the ground is equal to that of an aircraft in the maximum of the beam for quite small amounts of radiated power. In other words the effective beam width for ground returns is θ₂ where θ₂ is greater than θ₁ the effective beam width for target detection within 5% of maximum range. If now the fighter flies at height h above the ground, and the downward angle of scanning is φ, ground returns will commence at a range given by γ = h cosec (θ₂ + φ) and this will be the maximum range for targets in the direction φ. Now θ₂∞θ₀ ∞γ/λ so that good low altitude performance entails a high gain mirror system.

Effect of Window
Window interference is due to reflections from a spatial distribution of metallic objects. The confusing effect of Window arises from the presentation of a large number of responses on the cathode ray tube in addition to that of the target, and the resolution of the two is therefore very dependent on the method of indication. There are, however, factors in the design of the system which can reduce the ratio of Window to aircraft response. Thus consider a pulse of energy incident on an aircraft in a distribution of Window. If the transmitted energy is P_T and the pulse length is t, the aircraft reflects a constant amount of this energy so long as ct is greater than the length of the aeroplane. The energy reflected by the Window however, comes from a volume which is proportional to effective beam cross section times pulse length, i.e. to θ_2²t, and the ratio of aircraft response to Window can be improved by making this small, i.e. by use of narrow beams and short pulse length. This also helps to reduce the interference due to ground returns. The use of short pulses however, has an effect on maximum range for two reasons (a) Build up on indicator which makes Rmax ∞f (t) and (b) Receiver band width, which to give proper reproduction of pulse shape has to be f∞ 1/t. The receiver noise level is proportional to f^_-½ so that signal to noise ∞t^₂ and Rmax ∞t^¼. Combining these two effects
                       Rmax ∞t^¼f(t)                                        (4)
and the product θ₂t cannot therefore be indefinitely reduced by shortening the pulse if maximum range is to be preserved.

We now have the following relations:-
Rmax = const P_T^¼γ / λ^½
T = const γαβ/λω
p = const ωγ/λ
Rmax = const t^¼ f(t) for given P, γ, λ,

and the choice of the various quantities for any given. AI set is determined by the following conditions:-

(a) Pt pt to be less than maximum allowed transmitter dissipation
(b) γmax to be conditioned by aircraft installation
(c) ω to be less than maximum angular velocity at which good mechanical balance is possible
(d) λ^2t / γ²to be small.

The following examples will show how the problem of designing an AI equipment has been solved in some actual equipments.

4. Present day AI Systems
(1) AI Mark VIII (obsolescent)
This equipment was the first cm AI to be introduced into the RAF (Beaufighter and Mosquito aircraft) on a large scale. It is the production version of AI Mark VII and is an S-band spiral-scanning AI with Radial timebase presentation as described for AI Mark VII. The approximate constants of the system are:-

Rmax (Mosquito Target) = 4½ miles
G       (28" mirror) about 200 (in central position)
f         = 4 Mc/s
t         = 1 μsec.
p        = 2500 sec^-1
T        = 1.2 secs.
θ₀       = 7° approx.
α         = β = 45°
P_T       = 30 kW.

The scanner rotates at about 1000 r.p.m. and causes the beam to describe a spiral of maximum semi-angle 45° comprising 10 lines outwards interlaced with 10 lines inwards. The average spacing between the lines is about 2° so that with the given beam-width there is no loss of range between lines, and there are about 4 pulses per beam-width. No afterglow is used on the C.R.T., and the presentation factor is not good, partly owing to the fundamental nature of the indication which has to be spread out over a large area of the tube face. This latter factor is also the reason for the high repetition rate.

The equipment is engineered on conventional lines and in its basic form consists of scanner, transmitter, modulator, power pack, receiver and indicator, weighing, without auxiliaries, about 300 lbs. The electrical units are of simple design and are built round the CV64 magnetron operated at about 100 kW peak input; a hard valve modulator has to be used to handle the high pulse repetition rate.

(2) AI Mark X (SCR 720)
This equipment is now the standard AI of the RAF, and has replaced Al Mark VIII in all Commands. It is an S-band helical scanning system, based on the original American system started at the time of AI Mark VII, and discontinued in this country in order that effort might be concentrated on AI Mark VIII. The presentation used employs two tubes, known as range azimuth (B scope) and azimuth elevation (C scope) respectively. On these tubes the x-coordinate is in each case azimuth; on the former range is measured in the y-direction, and on the latter elevation. The timebase is moved synchronously with the scanner, and echoes are presented as intensity modulation. A spot on the "B" scope therefore gives directly the range and azimuth of the target; when a strobe line is placed on this spot, the "C" scope is brightened up and elevation readings are than available.

The constants of the system are:-
Rmax (Mosquito Target) = 5½ miles
G       about 200
t         = 1 μsec.
p        = 1500 sec^-1
θ₀       = 7° approx.
T (full scan) = 2 secs.
α        = -20° to +40°
β        = -90° to +90°
P_T      = 100 kW.

The most complicated unit is the scanner, whose mirror rotates at 360 r.p.m. The limits of elevation can be varied. The transmitter and modulator are pressurised and the indicator has a high intensity magnetic tube. A rotary spark gap modulator is used. The units are not designed for installation in any particular aircraft, but have been successfully and conveniently laid out in the Mosquito; the total weight without auxiliaries is about 400 lbs.

(3) AN/APS6A (AIAI)
This equipment is of interest since it is the only pilot operated AI in Service. It is an X-band spiral scanning system, using a special variety of indication known as"H" presentation. In the "H" system signals are presented on a normal range-azimuth tube of the "B" type, but alternate timebases are displaced up or down by a small amount according to whether the scanning mirror is pointing up or down. All signals therefore appear double and the position of the second "dot" with respect to the first gives a rough indication of the elevation of the target. This system is of course much less accurate than the radial timebase or the "B" and "C" scopes used in AI Mark X, but fortunately it is unnecessary in the early stages of an interception to have very accurate elevational information.

The constants of the AN/APS6A system are as follows:-
Rmax (Mosquito Target) = 3 ½ miles
G       about 200
t         = 1 μsec.
p        = 2000 sec^-1
T        = 4 secs.
α        = β = 60°
P_T      = about 30 kW.max

The units are simple; the scanner is electrically driven and waveguide fed, and the transmitter and modulator are pressurised. Typical American construction, with small components is used throughout, and the weight of the complete basic system is under 200 lbs. The equipment is installed in fighter aircraft of the American Navy.

5. Extra facilities required of AI Systems
The discussion in previous paragraphs has related entirely to the interception function of airborne radar, since it is in this role that the introduction of centimetre wavelengths has resulted in the development of new techniques. It should not however, be overlooked that the following facilities are also required in conjunction with all AI systems:-

1. A method of identification
2. A method of homing to ground beacons
3. A method of beam approach.

Of these the first is of paramount importance, since the concentration of friendly aircraft in regions of enemy activity is great. The homing beacon is used as a reference point for controlling patrols and the beam approach system is needed as a landing aid.

In British radar development emphasis has as far as possible been laid on achieving the least possible complexity. An attempt was therefore made, in AI Mark VIII, to make the basic AI equipment provide as many of these extra facilities as possible, with the minimum of extra equipment; this was achieved as follows. Interrogation of IFF Mark III was provided by operation of a separate 183 Mc/s transmitter/ receiver head feeding into the main AI receiver, and the IFF signals were displayed on the main AI tube. Beam approach facilities were obtained by using the interrogator head, with appropriate frequency changes. It was proposed to use a centimetre beacon for homing, and the main AI transmitter/ receiver channels were used for this. Appropriate timebase switching was arranged for these different facilities, and the whole system was found to work quite satisfactorily, although the introduction of the extra facilities resulted in an appreciable increase of complication in design.

In the comparable American radars such as AI Mark X the IFF and navigational facilities have from the beginning been entirely divorced from the search radar, and necessitate the installation of extra self-contained interrogator units with a separate indicator although IFF signals can be presented on the search radar scope by suitable synchronisation. Centimetre beacon facilities are available using the search radar channels. This trend of development complicated installation problems, but results in a clean design for the electrical units, and greatly simplifies switching; in particular it permits the use of high performance magnetic tubes in the search indicator which is not called upon to handle the deflection signals from ground beacons.

It seems highly probable that future development of AI systems will be along American lines, and will separate the centimetre equipment entirely from the navigational radar. The situation will, however, clearly be largely determined by future IFF development, in particular the fitting of IFF Mark V since it is a prime requirement to provide early and accurate identification of any radar contact directly on the search set.

6. Operational Results with Centimetre AI
(Note:- It is very much regretted that precise figures cannot be presented here. They will, however, be found in the O.R.S.(F.C.) reports 1942-l944 ).
The analysis of night fighter combat results is a highly specialised procedure and must take many aspects of the night fighting problem into account. It must sort out from the reports of pilots and navigators and ground controllers, the number of interceptions which can be classed as "close GCI control", "loose control" or "free lance". The whole course of the pursuit has to be followed through from GCI vector to AI contact, from contact to visual, and from visual to combat, and finally, if the contact is not brought to combat, the reasons for its loss have to be estimated. In a problem of this complexity it is obviously not easy to draw hard and fast lines, and the estimate of the precise benefit derived by the night fighter crew from the use of AI often tends to be obscured by the presence of fortuitous searchlights, or a bright moon or just luck. It has, however, been found that the interceptions made under close GCI control are the most useful in estimating AI performance, since they are most easily corroborated and generally turn out to be the most consistent; discussion will be confined to this type of interception.

The method of analysing interceptions is to record the following figures for a night's operations:-

No. of attempts to intercept ...  a
No. of AI contacts .... ... ... ...  b
No. of visuals .... ... ... ... ... ... c
No of combats .... ... ... ... ... . d
No. of E/A destroyed ... ... ...  e

The overall efficiency of the interception system is given by the ratio e/a, assuming that the GCI has always placed the hostile within the maximum range of the fighter AI. With the exception of help in the regaining of lost contacts, the GCI system has no effect on the ratios c/b, d/c, e/d which are entirely dependent on the AI system. The variation of these ratios over a period of time gives a picture of the development which has occurred in GCI systems, enemy countermeasures; aircraft performance and airborne radar.

The first successful AI systems was AI Mark IV, which was used during the early heavy night raids in 1941 and in the sporadic raids of early 1942. This system had a good coverage, but the direction-finding system was not really satisfactory except in the hands of experienced observers; it was therefore found that whereas contacts were fairly easy to handle, it was difficult to convert these into visuals in the fact of even mild evasive action and in addition low flying aircraft presented an almost insoluble problem. The Beaufighter night fighter aircraft used, although often very successful had not really enough performance to catch light enemy bombers of the JU88 type, and the report "bandit climbed away" began to occur very frequently in the log of night operations.

Under these circumstances, the appearance of AI Mark VII was greeted with enthusiasm. Ranges as good as those customary with AI Mark IV were obtained at all heights above about 3000 ft and the excellent D.F. properties were most striking and immediately revealed any tendency to evasion on the part of the enemy. It was therefore found in 1942, when AI Mark VII came into service, that the detections/attempts ratio-remained about the same but that the efficiency of converting detections into combats increased considerably. The interception technique moreover suffered a significant change as a result of the introduction of AI Mark VII; owing to the much smaller coverage of the system and to the indirect nature of the presentation it was inevitable that a stern chase should develop, and the cutting off vectors which were possible with AI Mark IV in the hands of a skilled observer now had to be given up. This change was on the whole a step not to be desired, since the speed difference between the Beaufighter and the JU88 was not sufficiently great to justify stern chasing; this disadvantage was however, greatly outweighed by the other manifest advantages of AI Mark VII, and with faster aircraft such as the Mosquito in view it felt that the greatly improved overall AI performance justified a large scale programme with. AI Mark VIII.

Al Mark VII was the only centimetre in service with the RAF until December 1942, when the first squadron of AI Mark VIII was equipped. During, this time enemy raids on this country were on a small scale, but in this period, and in its subsequent history in the Mediterranean theatre AI Mark VII was instrumental in the destruction of upwards of 100 enemy aircraft, more than one for every set of equipment made. During 1942 and early 1943, however, the growing appreciation of two important features of night interception operations caused a re-orientation of policy with regard to centimetre AI, These features were:-

(i) The almost certain use by the enemy of Window (not actually used in raids over this country till late 1943)
(ii) The certainty that enemy aircraft were likely not only to become faster and smaller, (FW 90 and ME410 types), but also to take more and more violent evasive action, possibly as a result of being fitted with some type of Tail Warning equipment.

Our own bomber policy was loading to the inevitable use of Window as a confusing agent against the enemy ground and airborne radar, and tests of the effect of Window on AI Mark VII had shown that the radial timebase type of presentation was especially susceptible to this type of interference. It
was shown that a bomber dropping-the necessary (and possible) quantities of Window would have-no difficulty in avoiding interception by an AI Mark VII (or VIII) equipped fighter and that it would also be possible for a few enemy bombers so to infect a "lane" with Window as to provide a path for a main force in which fighter opposition guided-by ground radar, would be impossible. In addition to-this, the GCI system was rendered almost useless, so that controlled interception would be doubly impossible. The trouble with AI Mark VII and VIII arises from the fact that the display does not give a direct presentation of target position, and there is no easy method of distinguishing "Window" from the aircraft response.

The introduction of faster aircraft by the enemy again drew attention to the time likely to be wasted in the carrying out of the stern chase necessary with AI Mark VIII, and the likelihood of a tail warning device brought forward the desirability of being able to get initial contact on the enemy from the side, which is difficult with AI Mark VIII owing to the relatively limited azimuthal coverage.

For these reasons, even while AI Mark VIII was being introduced into the first squadrons of Beaufighters and Mosquitoes, a search was being made for an AI which would better meet the new conditions which were expected. This search finally resulted in late 1943, in the introduction of AI Mark X. For the whole of 1943 and well into 1944, AI Mark VIII remained the main AI equipment of the RAF. Enemy raids in 1943 were on a very small scale and the sum total hardly permits statistical analysis, but there was no doubt that the overall efficiency of the interception system was continuing to improve even in the fact of faster and better equipped enemy bombers. All the ratios for AI Mark VIII were better than those with corresponding aircraft equipped with AI Mark IV and V, the improvement in the figure for combats to detection being especially notable. It was also proved at the same time that centimetre AI was satisfactory in maintenance; serviceabilities of about 30 hours per fault were fairly quickly reached in all squadrons with normal training and in many were exceeded.

When at last the enemy used Window, the density employed was never sufficiently great seriously to embarrass the AI Mark VIII equipped squadrons, although the number of contacts lost rose slightly. Moreover the appearance of centimetre GCI at this time did much to counteract the new phenomenon, and on the whole, after a slight preliminary setback, the interception system during the spring raids of 1944 was not seriously impaired. Despite this, the decision taken early in 1943 to replace Al Mark VIII by AI Mark X, which was known to have a better performance in Window, was confirmed, and AI Mark X squadrons began slowly to form towards the end of 1943.

The main advantages of AI Mark X over Mark VIII are:-
(i) Better discrimination between aircraft and Window owing to direct nature of B scope display
(ii) Better coverage
(iii) Better maximum range.

The improved coverage and maximum range are both useful under conditions of Window and of violent evasive action, and permitted a return to the interception technique employing cutting off vectors. Despite the obvious advantages of AI Mark X over Mark VIII in this respect, the statistical analysis of the GCI controlled interceptions over this country, up till D-Day 1944 show very little difference in the operational efficiency of the two types. It is certain that the Window density has never been sufficiently serious to affect the interception to any great extent, and it is probable that GCI
control was always sufficiently good to put the fighter within maximum range of the target for both equipments. The final combat is in any case largely an argument between machines, and the intervening stages) from detection to visual should not be particularly affected by the differences between AI Mark VIII and Mark X. In both cases the overall efficiency of this operation from attempt to destruction is about 12%. A detailed analysis of the reasons why contacts have not been turned into combats had also been made, and the chief causes are as follows:-

(i) Enemy speed and evasive tactics
(ii) Faulty identification
(iii) Enemy entering gun defended area
(iv) Window.

Of this, identification always contributes the greatest proportion; the number of contacts lost owing to enemy speed and evasive action tends to increase slightly. The effect of Window has not been so large as anticipated and the improvement shown by AI Mark X squadrons over AI Mark VIII in this respect is not very great, although appreciable. Broadly speaking the efficiency of the interception system has not greatly improved over the two years during which centimetre AI has been in service, but it must be remembered that the technical advances made by the enemy in aircraft and countermeasures have made the task of interception progressively more difficult over this period. That the efficiency has maintained a fairly constant value in the face of only small scale enemy activity of the fighter bomber type is in itself a tribute to the improvements and developments in our own technique and training.

7. The Future of AI
At the end of 1944, with the war in Europe, moving slowly but steadily towards its close, centimetre AI is in the estimation of the Air Staff a tried and essential adjunct to the equipment of any night fighter expected to counter the type of enemy activity experienced over this country up till D-Day. The question that present-day designers must answer is: "Will development of AI systems along the lines which have proved adequate during this war suffice for the future". Fortunately the way to an answer is pointed by a number of recent developments in. the air war, which it will be of interest to consider in detail.

Since D-Day the role of AI in the RAF has shifted from the defensive to the offensive. The complete reliance of the enemy an V-weapons for all attacks on this country for six months following D-day has changed the character of night fighting; it is no longer necessary to have an accurate airborne radar, and it would in fact be almost impossible to achieve much efficiency in a V-1 chase owing to the extremely low altitude and-high speed of the operation. The emphasis has in fact suddenly shifted from controlled interception, involving careful analysis of position and estimation of vectors, to gunnery, involving accurate range and prediction only. The Mosquito squadrons of Air Defence of Great Britain derived no great advantage from their AI equipment in V-1 operations, and the main harvest while it lasted was reaped by the faster Tempests and by the AA guns. Concurrently, however, with the V-1 attacks, 85 Group Mosquitoes were carrying out defensive operations of a standard character over our beach-heads in Normandy, and have steadily supported our advance through France and the Low Countries ever since. In the early days of the invasion, enemy activity was frequent; later it became intermittent, and finally the only way of keeping the squadrons in operation was to allow them to make night sorties over enemy territory in search of aircraft taking off or landing, usually in connection with our bomber activity. In this way the role of the mobile night fighter squadrons accompanying the invasion forces became offensive, and the object of the night fighter squadrons remaining in Britain became to get overseas as soon as possible.

At about the same time, in Bomber Command, the use of centimetre AI equipped night fighters for Bomber support began to assume new importance. Two types of operation had for some time been found profitable; in one, night fighters accompanied our own bomber force and attempted interceptions on enemy night fighters coming up to attack; in the other enemy aerodromes involved in the night fighter defence of Germany were regularly patrolled. Both types of operation benefited by the introduction of centimetre AI for obvious reasons; the never very easy task of intercepting an enemy aircraft, either at high level (as with the bomber stream) or at low level (as in aerodrome patrol) was facilitated by the better accuracy and greater range of AI Mark X compared with AI Mark V. In particular, the operations of the Serrate-type in which first warning of an enemy aircraft is obtained by listening for AI transmissions from the enemy, often resulted in rearward contacts, which can only be successfully dealt with if good and clear AI indications, in both fore and aft directions are available. The provision of rearward cover for night fighters operating over enemy territory has also become of great importance, owing to the danger of the enemy attacking the intruding fighter, and it is clear that any future AI likely to operate under offensive conditions must have this faculty. There is also little doubt that it would improve the general efficiency of defensive fighters especially in the regaining of lost contacts. Rear cover is now felt to be the chief improvement required in AI Mark X to make it the ideal AI for the present phase of the war.

Centimetre AI is now in use in the Pacific theatre on an increasing scale and a number of useful indications on its future development can be derived from this type of operation. In the Pacific, enemy night activity other than reconnaissance is largely confined to low level torpedo attacks on ships; it has been countered by American Hellcat fighters equipped with the pilot operated AN/APS6. In these operations it has been found that the qualities of the Japanese as a bomber pilot are not good, and neither Window nor evasive action have been exploited to such an extent as to make the pilots task of interception impossible. A determined enemy with aircraft such as are now available to the Germans could however, make night interception in single seater machines flown from carriers a very difficult task, and the development of a direct and simple pilot's AI for-the small carrier-based fighter will be
an urgent requirement until aircraft of the Mosquito type can be flown from a deck.

It is therefore necessary that future AI systems should be provided with all-round looking facilities at all heights and that the presentation should be as clear, and the operation of the set as simple, as possible. Obviously too, the range should be increased to the limit set by ground returns, and all the technical advances in cathode ray tube screens and controls should be.. exploited. Such advances will however, only result in AI systems of the type which we now know and these will only succeed in deal with the kinds of attack which we now know. It may well be argued that the future lies with the high speed pilotless aircraft and that heavy radar equipped night fighters will be useless against this type of activity, just as our Mosquito squadrons derived little benefit from AI Mark VIII in attacking V-1 or as the German night fighters seem unable to combat our own fast Mosquito bomber raids. Or on the other hand it may well be pointed out that future bombers will be of the Super-Fortress type and will fly in such closely controlled formations, exploiting radar detection and fire control to such an extent that it would be useless for any normal present day fighter to approach them. In both these eventualities our present types of AI are likely to be of little service; pilotless aircraft have speed which always makes successful intercepting difficult, and large heavily armoured bombers are always likely to have the advantage in an argument with a pursuing fighter which ultimately depends on radar efficiency.

What then is our programme for the future? First let it be said that it is not likely that a jump to any radically different type of AI will be made at once; later Marks will grow out of the present at a rate determined primarily by operational necessity while the war lasts, and by scientific enthusiasm and departmental encouragement in the ensuing peace. It is therefore in the first place profitable to consider any extra facilities now required or span/likely to be required in AI which might be introduced within a year or two. Of these the most obvious, apart from improved identification, is blind firing.

Blind firing is no new requirement, nor is it a. now development; unfortunately perhaps for the future of AI, it has never been an urgent necessity in this war. However with little stretch of the imagination, enemy bombers equipped with AGLT can be visualised which would always have the advantage over a fighter in sheer gunnery efficiency owing to their ability to fire blind and to utilise the important facilities for predicted firing offered by the, combination of radar with the gyro gunsight. Obviously the fighter can be similarly equipped, and indeed such an attempt has already been made with Al Mark IX, which is a lock follow system offering blind predicted firing. Unfortunately such systems as at present known suffer seriously and fundamentally from a deterioration of search performance, and this, at a time when the fast search scanning systems such as AI Mark X are meeting with such success, has greatly retarded their development. It is possibly that at some expense of time and ingenuity the lock follow-cum fast search system can be successfully engineered, and exploratory work on these lines should be done; but an alternative, and to same people more attractive proposition lies in the separating out of the interception and gunnery functions of the AI radar. In these schemes the interception would be carried out exactly as now, by an operator (AI Mark X) or pilot (AN/APS6); at a certain stage however, when the target had been brought to within, say, 1 mile, an independent gunnery radar would be-brought into operation, which would automatically track the target and present the pilot or his gyro gunsight with direction and range. By this means predicted blind firing could be obtained without sacrificing search performance; by careful development it is possible that the whole double set could be made sufficiently simple and small for installation in single-seater aircraft. In multiseater machines the problem would of course be simpler, and trials of some sort of system along these lines should be made.

Such considerations lead to the conclusion that the ultimate night fighter of the future, if it contains a pilot at all, must be a machine carefully designed to accommodate a highly trained pilot and a highly complex radar equipment. The present-day policy of adapting existing aircraft to carry an ill-assorted array of equipment with considerable discomfort to the crew is unlikely to succeed when the exigencies of the problem call for a highly integrated effort on the part of pilot, machine and equipment. All round looking, for instance, will involve new types of nacelle; pilot presentation, always desirable and later possibly essential is most likely to involve windscreen projection of radar and instrument information. Assuming then that by the expenditure of a great mount of effort the necessary collaboration between the RAF, aircraft and radar authorities has been achieved, it might be of interest to draw up a futuristic picture of a night interception operation. In this, early warning of the approach of enemy aircraft and approximate height and position would still be necessary, owing to the probable high speed of the aircraft. The pilot of the (single seater) night fighter would in the first place use ground control for getting his approximate direction and when at about 25 miles from the incoming bombers would switch on his long range radar. This would be a fast helical scanning system with scanners disposed about the aircraft to give all round coverage, and a sharp clear range azimuth picture would appear on the wind-screen. On this picture all ground returns and Window would have been eliminated by electrical discrimination against any echo moving with the forward velocity of the aircraft; there would however, be an artificial horizon and direction marker. The problems of identification and jamming would have been solved, so that friendly and hostile aircraft would appear on the range azimuth screen as spots of different shape (or even colour). Elevation indications would not be at first of great importance but would be available at will on a special elevation azimuth timebase at the top of the range azimuth tube, which would plot the difference between fighter and bomber height for every echo on the range azimuth picture. The pilot would always have this picture, on an appropriate scale, available to him, and present experience makes it fairly certain that with sufficient clarity of presentation, he should. have no difficulty in getting to within about a mile of his target. At this stage enemy evasion might start, and the fighter pilot would switch on his gunnery sot. This would immediately present him on the windscreen with an azimuth elevation or television picture of the forward hemisphere with one to one correspondence in direction; the search picture could be blacked out or maintained at will and tail warning could if necessary be aurally presented. At ranges of about 2500 ft the television picture of the target would be sufficiently clear for the pilot to detect impending changes of course by wing tilt; this would enable immediate correction. As firing range approached, automatic computing would supply range data and could even fire the guns at the right time through a pilot's ever-ride button; otherwise the television picture would be observed through a gyro gunsight and firing carried out as in the daylight case.

The whole of this process presupposes a necessary margin of both speed and armament. If the margin in speed were small or negative, as in the case of a pilotless or fighter type bomber with highly destructive bombs, the only hope for the fighter would be to convert it into a pilotless ground or aircraft controlled missile. If the margin in armament were small, or negative, the speed of the fighter would have to be exploited to the uttermost and the bomber would usually have the advantage. If the margin in speed and armament were both large, which is unlikely in the future, it would doubtless be profitable to reduce the radar complication by bringing back an observer.

In conclusion it may be said that to design a new AI system Without an intelligent estimation of the type of opposition which it is likely to encounter might be a worthless gamble AI development has always lagged behind corresponding development in offensive weapons because it is primarily dictated by factors which are not entirely known until it is actually in operation against the enemy. The effort then required to produce an effective countermeasure might however be so enormous that it would be folly now to neglect preliminary exploration of even the remotest possibilities.

Page last updated on 13th February 2018 by Colin Hinson.

.