Despite all technocrate’s dreams, aerial combat between peer opponents was always visual-ranged. Reasons for that vary; main reasons are inadvisability of using active sensors, low probability of kill for BVR missiles, and IFF problems. All of these problems are far greater against numerically and technologically comparable (or simply numerically superior) opponent than against numerically and technologically inferior opponent. Thus, WVR combat is likely to remain standard for aerial warfare, along with its large accent placed on OODA loop.
OODA (Observation-Orientation-Decision-Action) loop is fundimental principle of air combat. Fighter pilot first observes situation; after that, he orients based on previously-avaliable and acquired information (nationality of opponent, cultural considerations likely to affect opponent’s actions in current situation, etc.), then decides on further course of action and acts based on that decision. In the next loop, he observes opponent’s reaction to his own action so far as well as new situation, with rest of loop proceeding as in first one, though “orientation” part takes far less importance unless new information comes into play. In any case, breaking opponent’s OODA loop or going through it faster than opponent is prerequisite for victory. Opponent’s OODA loop can be broken by denying him vital information (done through usage of passive sensors, small visual and IR signature of one’s own aircraft, employment of various forms of jamming and environment-based interference), as well as by going through the loop faster than him – be it through faster observation/orientation/decision or executing action faster than opponent, which requires maneuverable aircraft capable of quick transients from one maneuver to another.
OODA loop of UAV operator is always imperfect, and worse than that of fighter pilot. Major problem is a delay from two to five seconds between UAV recording image and image being seen by UAV operator. Total delay between drone’s sesors recording opponent’s action and drone finally reacting to it – delay between „observe“ and „act“ part of the loop – can therefore reach ten seconds. Due to this delay, unmanned vehicles will be completely incapable of being inside human-piloted fighter’s OODA loop, which is a prerequisite for victory in a dogfight. But there are even more shortcomings than that.
In particular, each part of OODA loop is in itself imperfect. Observation made solely with information from mechanical sensors is never perfect as we have yet to design sensor as good as human eye. Imperfect observation means that imperfection continues to snowball through latter three parts, ending in action with some measure of disconnection from reality – and that can continue through multiple loops.
While drones are much smaller and cheaper than manned fighters, it is only result of their mission. If modern drones are faced with SAMs, MANPADS or enemy fighters, engagement is a foregone conclusion – and one not in drone’s favor. Drone operators cannot detect threats to their aircraft, and if drone was to be designed to be as capable and survivable as manned jet fighter, it would be just as large and costly, if not more, due to the need for advanced computers and communication systems. Even current, relatively simple, drones have much higher operating costs than manned aircraft, and are as much as ten times as prone to crashing – and both shortcomings can only worsen with increased size and complexity required for aerial combat.
Gigantic data transfers required to operate drones can easily lead to communication systems being overburdened – single Global Hawk drone uses as much bandwidth as did all US forces in the invasion of Afghanistan. Bandwidth is also a hidden cost of UCAV – while UCAV itself may be cheap, it requires very expensive (on order of hundreds of millions USD) equipment for data transfers, and even with modern UCAVs performing relatively simple tasks, data transfers can take up lion’s share of 250-million-USD satellite’s bandwidth. As such, entire package (UCAV and equipment required to operate it, which is actually part of UCAV despite not being in the airframe) can rival or exceed cost of manned fighter, with latter being a certainity in any UCAV capable of air-to-air combat.
Further, increased bandwidth automatically means increased vulnerability to jamming and other forms of electronic countermeasures. Main way datalinks defend against jamming is by reducing data transfer speed in exchange for increased reliability; that, however, may not be an option for data-hungry UCAV. As such, UCAV’s will be incapable of executing missions in heavily jammed environment, unlike manned aircraft, especially since it is far easier to build very powerful spread-spectrum jammer than to create jam-resistant uplinks.
Drones are also vulnerable to computer viruses, which could take control of a drone and order it to do anything by simulating incoming traffic from its operator.
It is also important to realize that UCAV capable of matching or exceeding the aerodynamic performance, load carrying capability and combat radius of manned fighter would be exactly as large and heavy as fighter in question. This would mean similar production cost to manned fighter (not counting control and data transfer systems), but at far higher maintenance costs, as much as several times higher, which would make it impractical to replace manned fighters with UCAVs on one-for-one basis. Further, having UCAV brings no operating cost savings, since it actually requires more operating and maintenance personnell than manned fighter due to the greater complexity.
As a result of everything above, replacing a manned fighter would require a fully-functional AI with almost identical cogniscive capabilities to a human brain – a feat that is, at this point, completely beyond both our knowledge of human brain, as well as beyond our hardware capabilities, and will remain so for some time – interestingly, contrary to MIC technology advocates, computer science experts have a complete disagreement on wether true AI can be achieved by the 2040; in any case, past trends do not give any reasons for optimism in that regard. Even when that is achieved, such complex programs would present serious reliability, maintainability and implementation challenges, possibly to the point of making an AI UCAV basically unflyable.
Drones will, however, remain useful in intelligence gathering, a role they have been used in since Vietnam, as drone being shot down does not carry the risk of operator being captured for questioning, and is much more politically acceptable. While these advantages also exist in regards to manned combat aircraft, disadvantages are simply too large.