Non Cooperative Target Recognition Information Technology Essay

Non Cooperative Target Recognition Information Technology Essay One of the most important aspects of any combat situation is target identification or recognition. In aerial combat, the problem is compounded because of large ranges and similar looking fighter aircraft. Unless a threat is ascertained, counter offensive action cannot be initiated. The problem of target identification in the present air combat scenario has been further complicated due to the large range of the new generation Beyond Visual Range (BVR) missiles  [1]  . While in the yesteryears, procedural control and systems like Identify Friend or Foe (IFF) were in place, they did not guarantee positive identification, since they were not stand alone in nature  [2]  . Therefore there was need to develop a stand alone system capable of identifying airborne threats. 2. Thereafter Non Cooperative Target Recognition (NCTR) techniques were developed, which did not depend on any direct participation from the target for identification purposes. With the emergence of this technology, a big hope was created that this technique would be resolve all the identification issues. The NCTR technology started gaining momentum in the mid eighties and is yet to prove its capability, almost 25 years on  [3]  . Despite technological advances in computers, NCTR is yet to see the light of the day. Though many modern day Airborne Interception (AI) radars feature NCTR, it is still not being used as the primary means of target identification. NCTR at present is being used in conjunction with IFF and other procedural methods primarily due to its reduced reliability. 3. Incorporation of NCTR technology in the older AI radars may involve change of the radar itself. Considering this and the fact that NCTR technology has not fully matured, is NCTR really the most economically viable solution for airborne target identification at present? Or is there a better system available for this purpose? Methodology 4. Statement of Problem. Modern day air combat environment is dense with both friendly as well as hostile aircraft. Identification is a serious problem and this has been further compounded by significantly large ranges of new generation air to air missiles. Erstwhile methods of target identification like IFF have failed in many situations, leading to fratricides. Would employment Non Cooperative Target Recognition techniques alone resolve this issue? 5. Hypothesis. NCTR as a technology is very demanding on the computer hardware as well as software. So at present NCTR may not be the best solution for problems related to aircraft identification in a dense air combat environment. It is possible to get inputs from all available sensors like AI radars, AWACS, aerostats etc and integrate them to form an intelligent composite picture and identify friends from foes. However NCTR as a concept is still very lucrative. 6. Justification of Study. The requirement and importance of target identification is fairly obvious and history will bear testimony to the fact that most of the fratricides have been due to incorrect identification. Earlier systems have failed primarily because the onus of identification was with the target. Inability to prove himself for whatever reason, resulted in a friendly target being declared hostile. While NCTR as a concept seemed to be a good solution, the technological implementation of the concept has still not emerged fail proof. Also, integration of various other existing sensors in a networked environment to form an intelligent composite picture seems to be a more viable solution. 7. Scope. The scope of this paper is to examine as to whether NCTR is really the path ahead, as far as target identification goes. The paper will look into the complexities associated with database required for airborne target identification (3 dimensional space, multiple aircraft configurations etc.) and also the technical limitations associated with implementation of NCTR. The paper would also examine the feasibility of implementing other techniques of identification. The scope would be limited to logical analysis based on available information on the subject and would not involve any scientific experiment or mathematical computations. 8. Methods of data collection. NCTR as a subject has gained momentum only in the past few years and hence has very few publications giving basic insight. Most of the data has been collected on the internet including soft copies of the books and validated thesis by others. The sources are mentioned in the bibliography at the end of the paper. 9. Organisation of the Dissertation. It is proposed to study the subject in the following manner:- (a) Chapter I Introduction and Methodology. In this chapter the subject is introduced. This chapter also defines the statement of problem, the justification for the research and the methodology involved. (b) Chapter II Need for Airborne Target Identification systems. This chapter brings out the requirement and complexities involved in airborne target identification. (c) Chapter III History of NCTR. This chapter gives a brief insight into the history of NCTR (d) Chapter IV How exactly NCTR Works. To appreciate the complexity involved in the airborne identification process, it is important to understand how exactly NCTR functions. This chapter brings out the complexities involved in the working of NCTR. (e) Chapter V Various Types of NCTR Techniques. This chapter brings out the types of NCTR techniques available for use. Though the chapter is technical in nature, it is important to assimilate the various techniques in order to understand the complexity and short comings of the system. (f) Chapter VI Database Generation. The efficiency of entire process involved in NCTR depends on an accurate and exhaustive database. This chapter brings out the methods available and the limitations in generating adequate database. Again this chapter like the previous chapter is slightly technical in nature despite best efforts of keeping it as simple as possible. (g) Chapter VII Technical Limitations of the System. While so far the paper dealt with the software aspects, this chapter would bring out the technical limitations of the hardware required by the radar for functioning of NCTR. (h) Chapter VIII Alternatives to NCTR. This chapter looks at the alternatives to NCTR for target identification. (j) Chapter IX Conclusion. This chapter concludes the topic. CHAPTER II NEED FOR AIRBORNE TARGET IDENTIFICATION SYSTEMS If you know the enemy and know yourself, you need not fear the results of a hundred battles. SUN TZU Introduction 1. In any hostile situation it is important to identify the adversary. Without knowing who the enemy is, how can one fight? Same is true for aerial combat, only with added degree of difficulty. While in the earlier years, aircraft identification was mostly restricted to visual identification, with improvement in technology this was no longer possible. The increasing weapon ranges post World War II made it difficult to visually identify aircraft in air during combat. 2. And also, any given airspace may contain friendly, neutral as well as hostile aircraft. How can one fire at an aircraft without ascertaining its nature / intent. Therefore aircraft identification forms a vital part prior to weapon launch. Tragic Incidents. 3. History is full of tragic incident of misidentification in air. There are numerous examples of neutrals having been shot from the sky. There also many cases of fratricides even in the recent conflicts. Some of them have been enumerated below  [4]  :- (a) On 5 October 2001, a Russian Sibir Tu-154 Airliner flying from Israel to Russia was thought to have been shot down by a missile fired by Ukrainian troops in a military exercise. (b) In April 1994, 02 USAF F-15 aircraft under positive control, accidently shot down their own Army Black Hawk Helicopters in Northern Iraq  [5]  . Visual Identification of Aircraft 4. World War Era. In both the first and second world war, aircraft recognition did not pose any major problem. In this era, the kill solution was attained by use of guns and hence the weapon ranges were much closer as compared to the distances at which the targets were discernable. The aircraft were slow moving and it was possible to spot and identify the type of aircraft before commencing offensive action. The combat was largely within the so called Visual Bubble of the pilot/s. As can be seen from Fig 2.1, the aircraft can be visually recognized by identifying the distinct shape, colour and characteristic visual features of the aircraft. Due training was imparted for this. In fact, in United Kingdom, The Royal Observer Corps (ROC) was formed, which had civilian personnel specially trained for identifying aircraft  [6]  . Fig 2.1 : Aircraft of World War Ihttp://www.military-aircraft.org.uk/ww1-fighter-planes/nieuport-17.jpgRoyal Aircraft Factory SE5A World War One Aircraft 5. Post World War II. However post World War II, there was a rapid improvement in the fighters and weapon technology. With the advent of close combat air to air missiles and increase in weapon ranges, the hostile aircraft had to be identified by at least 2-3 km to initiate evasive action. Also the newer jets were similar looking in terms of shape and size. In the Fig 2.2, photos of F-15 and MiG-29 aircraft are placed side by side for comparison. While F-15 is an American fighter, MiG-29 is from erstwhile Russia. One can easily appreciate the similarity of features which may result in one being mistaken for the other. If this is the case at close ranges (20 30m), it would be reasonable to assume that visual identification would be almost impossible at ranges of 2 km and beyond. Fig 2.2 : Visual Comparison of F-15 and MiG-29 Fightershttp://www.atfx.org/photos/f15a.jpghttp://www.aerospaceweb.org/aircraft/fighter/mig29/mig29_12.jpg 6. Post World War II, there also has been a significant increase in the speed of fighters engaged in aerial combat. This fact together with earlier mentioned aspects makes visual identification a passà ©. The use of radars did give some respite to the aerial fighters. However the weapon solution was still based on visual identification. With the advent of BVRs, the need for alternate means of identification gained importance. Without this, the full capability of BVR cannot be exploited  [7]  . 7. To tackle this problem of target identification, some techniques like the IFF (Identify Friend or Foe) have been developed so far. IFF systems are based on challenge response system. The system consists of a transponder fitted on the aircraft which replies in the form of suitable codes to any other IFF system. If the codes match, the aircraft is declared friendly. However should the code be incorrect or in case of no response, the aircraft is declared hostile. It should be kept in mind that response would not be possible in case of battle damage, loss of encryption keys, wrong encryption keys, or equipment failure. Therefore the major drawback of this system was that it was imperative on the part of the friendly aircraft to prove that he is friendly. In the event of faulty transponder, there are chances of fratricide. 8. Therefore there is a need to have a stand alone system capable of identifying and recognizing aircraft to take timely responsive action and also to prevent fratricide / shooting down of neutrals. CHAPTER III HISTORY OF NCTR Know thy self, know thy enemy. A thousand battles, a thousand victories. SUN TZU 1. Need for a Stand Alone System. As can be seen from the last chapter, the IFF systems had a major drawback, that of requiring participation from the other aircraft. Inability to participate in the challenge response loop by other aircraft due to whatever reason was construed as Hostile by the system. This has resulted in many fratricides in the past. A number of such fratricides due to faulty IFF resulted in the need for a stand alone system. Some more cases of disasters due to faulty IFF are enumerated below:- (a) 1980. An Italian DC-9 passenger plane was shot down with a missile near Sicily. The missile was aimed by mistake because of incorrect IFF and discrimination of the type of plane. 81 passengers died.  [8]   (b) 1988. In the Persian Gulf, an Iranian Airbus was shot down by an SM-2 standard missile. The aircraft was erroneously recognized as an F-14 fighter aircraft. 298 people died.  [9]   (c) 2003. A glaring example of fratricide due to faulty IFF is that of two Royal Air Force Tornados returning after a mission in Iraq , which were shot down by US Patriot missiles on 22 Mar 03, in which both the crew were killed instantly. The investigating board attributed one of the causes as faulty IFF probably broken down due to power failure  [10]  . 2. These kind of mishaps led to the requirement of having a system which could identify aircraft without the active involvement of the unidentified aircraft. The concept of NCTR dates back from the mid seventies. A program started by the USAF code named Musketeer, introduced the idea of NCTR for the first time. In this program, the aircraft radar was used to count the fan blades in the engine  [11]  . This technology was later named as Jet Engine Modulation which would explained in detail in the subsequent chapter. The radar signature was unique to each engine. Since most aircraft had different engines, this technique was exploited for identifying the type of aircraft. 3. Though the concept was introduced in the mid seventies, the computational technology then was not adequate. The radar per se only gathers data and the extraction of useful information has to be done by the radar computers. Hence NCTR technology had to wait till mid eighties for the computer technology to catch up.  [12]  Thereafter the USAF had incorporated NCTR modes in the An/APG 63 radar of F-15 C aircraft. During the Gulf war in 1990- 91, use of this technique was made in conjunction with Airborne Warning And Control System (AWACS) for identifying hostile aircraft  [13]  . 4. Post Gulf war some progress was made in Automatic Target Recognition (ATR) systems. These systems did not restrict themselves to aircraft radars and were mostly ground based systems using millimetric wave radars  [14]  . The ranges of these radars were significantly lesser than airborne interception radars. While these radars were completely different in terms of technology and operating ranges, the basic concepts were similar. However the computations involved in aerial target recognition was much higher than ground targets. Though many experiments were conducted on ATR systems, the progress in NCTR was not at the same pace. 5. Post 1985, NCTR was tried and fitted on the USAF F-14, F-16 and F-18 aircraft. In addition, NCTR was also fitted on selected foreign F-15s (Israeli), British Tornados and French Mirage 2000-5s  [15]  . Today, in 2010, the technology has still not fully matured and still cannot be used in isolation for aerial target identification. CHAPTER IV PRINCIPLE OF OPERATION AND COMPONENTS OF NCTR Principle of Operation 1. Now before proceeding further, it would be prudent to understand how NCTR principles work. NCTR fundamentally relies on the returns from the target aircraft. Hence no active participation (as in IFF) is required from the target aircraft. Non Cooperative Target Recognition like any other identification process would in simple terms consist of first sensing the target. Then the relevant features of the target are extricated and compared with a database. Thus any NCTR system should essentially contain a Sensor, a Feature Extractor and a Classifier. 2. The principle of operation of NCTR is very similar to Optical Character Recognition (OCR) as in computers. The data from the target is collected by a suitable sensor and relevant data extracted. This is compared to the data existing in the library and then classified based on the results of the comparison. And for the success of this technique, a huge library of various aircraft profiles is required as database for comparison. The steps involved in identifying the target aircraft has been depicted in Fig 4.1. The flow chart of target recognition shows that the Recognition Algorithm takes two inputs namely measured target signature and signatures stored in the library and gives target identity as the output. Recognition Algorithms have been briefly explained subsequently. Fig 4.1 : Flowchart of Target Recognition SENSOR 3. Choice of Sensor. Choice of sensor is a very important part of this identification process. Without proper or adequate returns from the target, it would be nearly impossible to accurately classify the aircraft. For NCTR per se, a number of sensors are available Optical, Infra red, laser, acoustic and radar. However the optical, IR and laser sensors are prone to adverse weather conditions like moisture, dust etc. Also the detection ranges of these sensors are significantly lesser as compared to air to air weapon ranges especially in a BVR environment. Radar as a sensor has significant advantages as compared to the other sensors for the following reasons  [16]  :- (a) Aircraft are generally constructed by radar reflective material (even those with carbon composite have metal parts that reflect radar). (b) Radar can be used by day or night and is not affected by haze, fog, snow and rain. (c) Radar ranges are much higher than other sensors due to reduced atmospheric propagation attenuation.. (d) Radar has been one of the most extensively used sensors for aerial surveillance and the technology to generate, receive and process radar signals has been continuously refined for nearly 100 years. Hence radar is the optimum choice of sensor for NCTR on fighter aircraft. Moreover radars are already in place in all the modern generation fighters. Hence there would be no requirement to add a separate sub system on the aircraft. 4. Sensor Requirements. The radar capability required for the NCTR is specific. It is misnomer that any radar can be made NCTR capable. Suitable radar software is not the only thing that required for NCTR process. It is important that the radar design is appropriate to the type of measurement that has to be performed and to provide the type of target signature required. In order to obtain signatures of high integrity, the waveform must be carefully designed and the radar must support the transmission and reception of the signal without distortion  [17]  . The radar must be designed to minimize the effects of multiple reflections and clutter. The radar must also possess sufficient energy so that the returns from the target contain adequate data, required for the recognition process. Therefore it may not be possible to upgrade the existing radars with NCTR capability. Should it be possible, it may entail change of major components, which may not be the most economical solution for preventing fratricides. FEATURE EXTRACTOR 5. Function. The returns from the radar are simply electromagnetic pulses. It is of no use unless, relevant data is extracted from them. This is done by the Feature Extractor. Feature extraction is a process which happens inside the radar computer by means of suitable algorithms. These algorithms look for a specific attribute and isolate them for comparison. The algorithm for feature extraction would depend on the type of NCTR technique being employed for target recognition. Types of NCTR techniques are explained in detail in the next chapter. 6. Constraints. Feature extraction is a highly complex problem due to highly dynamic nature of fighter aircraft  [18]  . The radar returns from the aircraft are highly aspect dependent. More the aspect, more and stronger the radar returns, hence better feature extraction Aspect angle in simple terms is a measure of how much of the target aircraft is visible. When viewed from front / rear minimum area is visible and this is Zero aspect. When viewed from broad side, maximum area of the aircraft is visible. Now the aspect angles in 3D space are both in terms of elevation as well as azimuth. Due to highly aspect dependent nature of the scattered signature, the process of feature extraction becomes even more difficult. Also due to movement of the target, the radar signatures get distorted and this poses problems not only in feature extraction but also classification. These aspects are discussed in detail in Chapter 5. 7. After the radar has transmitted the pulses, detected and processed the return from the target of interest, the target signature is now ready for the recognition process. In order to perform correct target recognition, a lot of information needs to be known about the target. These information are contained in the form of a library in the radar computer. The details of database and database generation are discussed in detail in Chapter 6. CLASSIFIER 8. Classifier. The classifier in an NCTR system compares the extracted features of the target signature with the already existing library of aircraft signatures. The comparison is made with the help of recognition algorithms  [19]  . Recognition algorithms are mathematical techniques or formulae that compare the signatures which have been measured with the mathematical models of aircraft existing in the radar library. The algorithms are designed to analyse and assess how close a match a particular signature is to reference signature contained in the library. With the help of these algorithms, the classifier now assigns the most appropriate and closest identity to the target. Shown in the fig 4.2 is a target signature being compared with the three reference signatures. On close analysis though the target signature does not perfectly match any of the reference signatures, it is closest to Target C. Therefore the classifier would it assign the measured signature as Target C. Fig 4.2 : Target Signature Comparison with Reference Signatures 9. Target signatures have to be measured, modelled and analysed, to enable the attributes used for recognition to be identified. The techniques for reliably extracting these attributes from the target have to be determined. Libraries of target data have to be assembled, ordered, managed and updated as new targets emerge and more target signature data becomes available. The development of reliable recognition algorithms is clearly critical for developing a high-performance target recognition capability. 10. While a computer can easily identify various objects based on the size, shape, colour etcetera, it would still lack the ability to identify objects logically as the human brain does. Till the neural and genetic algorithms are not fully developed, it would be difficult to optimize NCTR for aerial identification.