All my military science career - it was a run on the spot
The system was formed by the mid-1980s. She was named the tracking system of the foreign policy situation and the threat of a nuclear attack, "Splav." By 1990, she was able to process enormous streams of information that flowed along all the lines of reconnaissance work — satellite, radio, external intelligence. It even took into account which windows are burning in the Pentagon late in the evening. One of the checks "Splav" was the prediction of the start of US operations against Iraqi troops in Kuwait in early 1991.
“Then we could not decide on the date. And suddenly it dawned on one of the employees: Lord, the wife of President Bush, like his associates, is obsessed with astrology! Just like the Nazi elite. Therefore, they will choose the time of the strike, in accordance with astrological predictions, - Vladimir Kravchenko recalls. - We immediately laid in the system also horoscopes - and got time for hour X. We were mistaken for only two hours ... " (authors of the project Alloy)
From the height of past years, I see that my research on the identification of airplanes, as reflected in my doctoral dissertation, was a dead end.
The only plus is that I learned how to independently set and solve applied scientific problems in the field of signal processing.
In the future, it allowed to become a good teacher in this or related fields.
But I consciously changed the direction of my activity. Therefore, the resources spent did not give the proper return. Such was the Soviet system - inefficient.
')
Recognizing the types of aircraft from the echoed signal now and then was not an urgent task and had little effect on the effectiveness of the air defense.
It has lost its relevance after the United States abandoned the creation of false targets - imitators of bombers.
This decision was made by US Congress in December 1975 - one year before my candidate defense.
Photo by the author (to the left of Gagarin)
The futility of recognizing ballistic missile warheads from the reflected radar signal became clear by about 1970, but the colleagues in my research laboratory continued to successfully defend student dissertations on this topic and even defended two doctoral ones. These were exercises in complex mathematical calculations that had no practical value.
The only positive result is that the theses authors have learned something themselves and were able to become teachers afterwards.
My doctoral work was inspired by the project " Splav ", initiated by Andropov himself in 1984.
At his order, applied research was launched to search for methods to recognize the intentions of the United States and NATO and to prevent them from preparing for a sudden nuclear attack.
Knowing nothing about where their legs grow from, I, too, harnessed to this research initiative in 1985.
It was only after the collapse of the USSR that I first heard about Splav, which was clearly re-secret.
If I knew the task setting at the highest level from the very beginning, I could have done my part better.
Nevertheless, by 1988 I had built my own theory of recognition of situations based on the analysis of facts that have different references to time and space, and together with my students I developed a computer-based demonstration program.
It contained 25 thousand lines of PL / 1 code, of which 2/3 occupied the user interface - something that can now be obtained almost free of charge in any programming environment. And then it took a lot of effort.
I have prepared several typical demonstration tasks for analyzing intelligence.
The system received small isolated facts at the entrance, analyzed them and gave an opinion on the compliance with this or that scenario - explaining the logical inference steps and indicating the time and place.
Including I made a demo with recognition of the nature of the attack of the USSR by enemy ballistic missiles.
This demonstration made a strong impression on the head of my academy, Colonel-General V.K. Strelnikov.
The fact is that at one time he was the first commander of the first division warning of a rocket attack and in his practice there were several cases of false alarms, because of which the world could be put on the brink of world nuclear war.
Therefore, Strelnikov began to respect me, and in case of chance meetings with me in the corridors of the academy, I was always warmly welcomed.
Most of the officers of the academy he personally did not know and usually passed by with a frown, not paying attention to the people they met.
For which he received the nickname "Messerschmitt", as he with his inhospitable air resembled an enemy fighter plane, from which one should not expect anything good.
And I also made a script commissioned by the KGB of the USSR and then thanks to this I received an invitation for a job interview in Moscow - right to Lubyanka. But I will write about it separately - if anyone is curious.
That is, my method of analyzing intelligence was fairly universal.
My work was not easy, the deadlines were broken several times, for which I received several stitches from my superiors until the system started working.
But when she earned in May 1988, each VIP who attended our academy was offered to look at “artificial intelligence”.
I remember there were two visitors even in the rank of marshal of the armed forces and countless military officials smaller (I do not remember the details).
They all experienced the joy of joining the secrets of artificial intelligence.
But most of all I was pleased with the visit of Major General, retired Academician G.S. Pospelov.
He is rightly considered the founder of the national school of artificial intelligence methods.
During the visit he was already 75 years old.
After listening to my report and seeing my system at work, Germogen Sergeevich asked: “And did you ALL do it yourself?” - I replied in the affirmative. Do not lie. I did everything myself as I could - let others do better!
A huge amount of resources was spent on this work. There were about 90 people working on programming for more than 20 man-years, 75 of them were academy students and interns from the faculty of the Moscow Medical University, each of whom developed one software module (an average of 100 lines of code). I wrote specifications for the requirements for the modules myself (50%) and four of my adjuncts (the remaining 50%).
However, the practical value of my work was close to zero. Thanks to Gorbachev's restructuring and the discharge of voltage, the project “Splav”, in the part concerning me, was curtailed in 1987. I did not know about its coagulation as well as about the launch of this project. Therefore, when I rolled out my doctoral thesis in 1991, it had already lost its relevance. Plus, one of my key algorithms deals with the NP-complete problem and can be used very carefully, pre-filtering intelligence information using all available methods.
It's about recognizing situations.
Suppose there are N facts of different types (total K types). To make a decision about the situation we are interested in, it suffices to find M facts (M <K <N), which also have conditions for other parameters (for example, in space and / or time).
That is, from N you need to choose the appropriate M facts.
This is a type of maximum click search task . She is NP-full.
My algorithm works satisfactorily when the number of extracted facts is small, i.e. N is not much larger than M. Then it is easy to add a puzzle (however, this is not a trivial task when time and space are fuzzy). When there are many similar, but not relevant facts of this scenario, i.e. when N >> K, the solution of the problem requires too much time. Therefore, the facts must first be selected using heuristics.
I have described some of these methods in my dissertation, but much depends on the specifics of the subject area.
Therefore, further purely applied research is needed to ensure the efficiency of the system.
It was very interesting to work.
Because in the military academy, the top of the little we imposed on those studies and gave some freedom of choice. Not complete - as in American universities, but quite a lot. It was a pleasure.
It was bad that at times it was difficult to understand where the legs of this or that research initiative were coming from.
This is due to excessive (IMHO) secrecy.
In military academies, the level of secrecy was reduced due to the fact that teachers could uncontrollably blab secrets to students of the academy, guided by the most noble promptings - to teach listeners to the most advanced.
And then students from the academy graduated from all parts of the USSR and secrets were turned into leaks.
Therefore, in my military academy, me and my colleagues were not allowed to particularly serious secrets.
Because of this, it became difficult to navigate and understand what was planned above.
And the relevance of my own research depended on it. And she (urgency) suffered.
In the meantime, the USSR collapsed, and my secret dissertation remained in the storage room to collect dust on the shelf, and I was engaged in completely different things.
In particular, the question of survival in the conditions of Ukraine, where nobody needed my knowledge and skills.
In general, most of my career in the Soviet Army, I quickly fled. But - in place.
Article posted with the kind permission of the author.
“Then we could not decide on the date. And suddenly it dawned on one of the employees: Lord, the wife of President Bush, like his associates, is obsessed with astrology! Just like the Nazi elite. Therefore, they will choose the time of the strike, in accordance with astrological predictions, - Vladimir Kravchenko recalls. - We immediately laid in the system also horoscopes - and got time for hour X. We were mistaken for only two hours ... " (authors of the project Alloy)
From the height of past years, I see that my research on the identification of airplanes, as reflected in my doctoral dissertation, was a dead end.
The only plus is that I learned how to independently set and solve applied scientific problems in the field of signal processing.
In the future, it allowed to become a good teacher in this or related fields.
But I consciously changed the direction of my activity. Therefore, the resources spent did not give the proper return. Such was the Soviet system - inefficient.
')
Recognizing the types of aircraft from the echoed signal now and then was not an urgent task and had little effect on the effectiveness of the air defense.
It has lost its relevance after the United States abandoned the creation of false targets - imitators of bombers.
This decision was made by US Congress in December 1975 - one year before my candidate defense.
Candidate
Ufimtsev’s formulas, published in his 1962 book “The method of edge waves in the physical theory of diffraction,” allow us to calculate the reflected wave at a fixed frequency of electromagnetic radiation for several simple bodies: a flat plate, a sphere, a cylinder and a cone. It should be understood that the deviations of the shape of these bodies from the ideal by less than a quarter of the wavelength do not matter. For example, the tail keel of an aircraft can be considered a flat plate with a wavelength of more than 15-20 cm, at least in the direction close to perpendicular to the keel. Of course, Ufimtsev didn’t write about it in his book - I had reached it myself, working alone, without consulting anyone. However, from the publications of Ufimtsev, it did not follow anywhere, where and for what purpose his formulas can be applied. Before that, I also had to reach my mind. If Ufimtsev had written about this, his book would not have been allowed to publish by the Soviet censor.
As I see it, I did two fundamentally new things with Ufimtsev's formulas.
(1) I guessed that by using several radiation frequencies at the same time in calculations, it is possible to obtain a model of a reflected radio signal for a signal with any complex frequency spectrum. Including with such a high latency time resolution (i.e., in the radial direction), which allows us to see separately the so-called “bright points” of the irradiated body. (For example, a cylinder is seen as two brilliant points at the ends of its generatrix closest to the observer - and nothing more.) And then, by the configuration of these points, you can determine the shape of the body and its orientation. Now it may seem trivial to someone because of the so-called “after-effect”. But I was the first to figure out how I did it.
(2) I figured out how to fit a plane with a combination of plates, cones and cylinders. For example, the front edge of the wing is a strongly elongated truncated cone, and the air intake of a jet engine is almost a disk, but with a reduced backscatter intensity of radio waves, since part of the probe radiation energy goes inside the compressor and is absorbed there. At the same time, I was guided by the results of experimental field measurements made in NII-2 in Kalinin by my classmate (before graduating from the academy, we both wrote diploma works from one research supervisor on the topic of recognition of radar targets). I found this information in secret scientific reports.
Thus, I was able to simulate radar signals with any frequency spectrum reflected from several types of aircraft on an antediluvian computer and simulate their coherent and incoherent processing. In principle, the main purpose of my research was to compare these two types of treatment.
In my military academy, two scientific schools struggled: coherents and noncoherents. The difference between them is that the second type of processing neglects the phase of the reflected signal and uses only its amplitude. The advantage of incoherent processing lies in the simplicity of implementation, since in the early 1970s the existing microwave technology did not yet allow coherent processing or the equipment would be too expensive. I belonged to an "incoherent" school; this was not my choice, and so I was appointed by order of the command.
With my research, I showed that in the presence of even natural noise (not to mention organized interference), incoherent signal processing loses much of coherent; on energy this loss is proportional to the width of the signal spectrum. That is, if you want to see the “radar portrait” of an airplane with the resolution of its individual brilliant points, then in the presence of noise, incoherent signal processing will most likely not allow this. With ballistic missile warheads requiring an increase in the width of the signal spectrum every 5-10 times, the situation was even worse. Thus, I planted mine under my own scientific school. But the truth was more precious to me.
This became the content of my PhD thesis, which I wrote without a supervisor. When it came to defense, I was still assigned a manager, but I listened to his advice on improving my dissertation and did the opposite.
During the defense, my report at the dissertation council caused a slight shock, since I stated bluntly that incoherent processing is a dead end in the development of radar. Head of the coherent school Ya.D. Shirman, understanding the political delicacy of the moment, in his speech helped round the corners and the scandal did not work out. However, the scandal to me was the least. I needed a PhD degree and promotion from captain to lieutenant colonel. I have achieved this goal.
Colonel Shirman in my academy was the most authoritative scientist (Dr. Sc., Professor, twice winner of the State Prize of the USSR). By age he was suitable for me as a father (he was 27 years older than me). A couple of years after my defense in casual meetings, he became the first to greet me and shook my hand. Later I understood the reason. Shirman quietly instructed one of his associates (that is, a graduate student) to analyze the theoretical calculations in my dissertation (it contains five pages in a row of mathematical formulas) and check their correctness. Then he instructed another adjunct to do the same. Both confirmed the correctness of my calculations, showing that incoherent processing is a dead-end path in radar. After that I became such a handshake.
Further, I made a sharp turn in my career, refusing to occupy the position at the department of missile defense, to which my scientific laboratory belonged, thus parting with radar, and asked to serve at the department of ACS, with little understanding in this area. At first it was difficult, but gradually I became, as it were, an IT staff.
As I see it, I did two fundamentally new things with Ufimtsev's formulas.
(1) I guessed that by using several radiation frequencies at the same time in calculations, it is possible to obtain a model of a reflected radio signal for a signal with any complex frequency spectrum. Including with such a high latency time resolution (i.e., in the radial direction), which allows us to see separately the so-called “bright points” of the irradiated body. (For example, a cylinder is seen as two brilliant points at the ends of its generatrix closest to the observer - and nothing more.) And then, by the configuration of these points, you can determine the shape of the body and its orientation. Now it may seem trivial to someone because of the so-called “after-effect”. But I was the first to figure out how I did it.
(2) I figured out how to fit a plane with a combination of plates, cones and cylinders. For example, the front edge of the wing is a strongly elongated truncated cone, and the air intake of a jet engine is almost a disk, but with a reduced backscatter intensity of radio waves, since part of the probe radiation energy goes inside the compressor and is absorbed there. At the same time, I was guided by the results of experimental field measurements made in NII-2 in Kalinin by my classmate (before graduating from the academy, we both wrote diploma works from one research supervisor on the topic of recognition of radar targets). I found this information in secret scientific reports.
Thus, I was able to simulate radar signals with any frequency spectrum reflected from several types of aircraft on an antediluvian computer and simulate their coherent and incoherent processing. In principle, the main purpose of my research was to compare these two types of treatment.
In my military academy, two scientific schools struggled: coherents and noncoherents. The difference between them is that the second type of processing neglects the phase of the reflected signal and uses only its amplitude. The advantage of incoherent processing lies in the simplicity of implementation, since in the early 1970s the existing microwave technology did not yet allow coherent processing or the equipment would be too expensive. I belonged to an "incoherent" school; this was not my choice, and so I was appointed by order of the command.
With my research, I showed that in the presence of even natural noise (not to mention organized interference), incoherent signal processing loses much of coherent; on energy this loss is proportional to the width of the signal spectrum. That is, if you want to see the “radar portrait” of an airplane with the resolution of its individual brilliant points, then in the presence of noise, incoherent signal processing will most likely not allow this. With ballistic missile warheads requiring an increase in the width of the signal spectrum every 5-10 times, the situation was even worse. Thus, I planted mine under my own scientific school. But the truth was more precious to me.
This became the content of my PhD thesis, which I wrote without a supervisor. When it came to defense, I was still assigned a manager, but I listened to his advice on improving my dissertation and did the opposite.
During the defense, my report at the dissertation council caused a slight shock, since I stated bluntly that incoherent processing is a dead end in the development of radar. Head of the coherent school Ya.D. Shirman, understanding the political delicacy of the moment, in his speech helped round the corners and the scandal did not work out. However, the scandal to me was the least. I needed a PhD degree and promotion from captain to lieutenant colonel. I have achieved this goal.
Colonel Shirman in my academy was the most authoritative scientist (Dr. Sc., Professor, twice winner of the State Prize of the USSR). By age he was suitable for me as a father (he was 27 years older than me). A couple of years after my defense in casual meetings, he became the first to greet me and shook my hand. Later I understood the reason. Shirman quietly instructed one of his associates (that is, a graduate student) to analyze the theoretical calculations in my dissertation (it contains five pages in a row of mathematical formulas) and check their correctness. Then he instructed another adjunct to do the same. Both confirmed the correctness of my calculations, showing that incoherent processing is a dead-end path in radar. After that I became such a handshake.
Further, I made a sharp turn in my career, refusing to occupy the position at the department of missile defense, to which my scientific laboratory belonged, thus parting with radar, and asked to serve at the department of ACS, with little understanding in this area. At first it was difficult, but gradually I became, as it were, an IT staff.
Photo by the author (to the left of Gagarin)
The futility of recognizing ballistic missile warheads from the reflected radar signal became clear by about 1970, but the colleagues in my research laboratory continued to successfully defend student dissertations on this topic and even defended two doctoral ones. These were exercises in complex mathematical calculations that had no practical value.
The only positive result is that the theses authors have learned something themselves and were able to become teachers afterwards.
My doctoral work was inspired by the project " Splav ", initiated by Andropov himself in 1984.
At his order, applied research was launched to search for methods to recognize the intentions of the United States and NATO and to prevent them from preparing for a sudden nuclear attack.
Knowing nothing about where their legs grow from, I, too, harnessed to this research initiative in 1985.
It was only after the collapse of the USSR that I first heard about Splav, which was clearly re-secret.
If I knew the task setting at the highest level from the very beginning, I could have done my part better.
Nevertheless, by 1988 I had built my own theory of recognition of situations based on the analysis of facts that have different references to time and space, and together with my students I developed a computer-based demonstration program.
It contained 25 thousand lines of PL / 1 code, of which 2/3 occupied the user interface - something that can now be obtained almost free of charge in any programming environment. And then it took a lot of effort.
I have prepared several typical demonstration tasks for analyzing intelligence.
The system received small isolated facts at the entrance, analyzed them and gave an opinion on the compliance with this or that scenario - explaining the logical inference steps and indicating the time and place.
Including I made a demo with recognition of the nature of the attack of the USSR by enemy ballistic missiles.
This demonstration made a strong impression on the head of my academy, Colonel-General V.K. Strelnikov.
The fact is that at one time he was the first commander of the first division warning of a rocket attack and in his practice there were several cases of false alarms, because of which the world could be put on the brink of world nuclear war.
Therefore, Strelnikov began to respect me, and in case of chance meetings with me in the corridors of the academy, I was always warmly welcomed.
Most of the officers of the academy he personally did not know and usually passed by with a frown, not paying attention to the people they met.
For which he received the nickname "Messerschmitt", as he with his inhospitable air resembled an enemy fighter plane, from which one should not expect anything good.
And I also made a script commissioned by the KGB of the USSR and then thanks to this I received an invitation for a job interview in Moscow - right to Lubyanka. But I will write about it separately - if anyone is curious.
That is, my method of analyzing intelligence was fairly universal.
My work was not easy, the deadlines were broken several times, for which I received several stitches from my superiors until the system started working.
But when she earned in May 1988, each VIP who attended our academy was offered to look at “artificial intelligence”.
I remember there were two visitors even in the rank of marshal of the armed forces and countless military officials smaller (I do not remember the details).
They all experienced the joy of joining the secrets of artificial intelligence.
But most of all I was pleased with the visit of Major General, retired Academician G.S. Pospelov.
He is rightly considered the founder of the national school of artificial intelligence methods.
During the visit he was already 75 years old.
After listening to my report and seeing my system at work, Germogen Sergeevich asked: “And did you ALL do it yourself?” - I replied in the affirmative. Do not lie. I did everything myself as I could - let others do better!
A huge amount of resources was spent on this work. There were about 90 people working on programming for more than 20 man-years, 75 of them were academy students and interns from the faculty of the Moscow Medical University, each of whom developed one software module (an average of 100 lines of code). I wrote specifications for the requirements for the modules myself (50%) and four of my adjuncts (the remaining 50%).
However, the practical value of my work was close to zero. Thanks to Gorbachev's restructuring and the discharge of voltage, the project “Splav”, in the part concerning me, was curtailed in 1987. I did not know about its coagulation as well as about the launch of this project. Therefore, when I rolled out my doctoral thesis in 1991, it had already lost its relevance. Plus, one of my key algorithms deals with the NP-complete problem and can be used very carefully, pre-filtering intelligence information using all available methods.
It's about recognizing situations.
Suppose there are N facts of different types (total K types). To make a decision about the situation we are interested in, it suffices to find M facts (M <K <N), which also have conditions for other parameters (for example, in space and / or time).
That is, from N you need to choose the appropriate M facts.
This is a type of maximum click search task . She is NP-full.
My algorithm works satisfactorily when the number of extracted facts is small, i.e. N is not much larger than M. Then it is easy to add a puzzle (however, this is not a trivial task when time and space are fuzzy). When there are many similar, but not relevant facts of this scenario, i.e. when N >> K, the solution of the problem requires too much time. Therefore, the facts must first be selected using heuristics.
I have described some of these methods in my dissertation, but much depends on the specifics of the subject area.
Therefore, further purely applied research is needed to ensure the efficiency of the system.
It was very interesting to work.
Because in the military academy, the top of the little we imposed on those studies and gave some freedom of choice. Not complete - as in American universities, but quite a lot. It was a pleasure.
It was bad that at times it was difficult to understand where the legs of this or that research initiative were coming from.
This is due to excessive (IMHO) secrecy.
In military academies, the level of secrecy was reduced due to the fact that teachers could uncontrollably blab secrets to students of the academy, guided by the most noble promptings - to teach listeners to the most advanced.
And then students from the academy graduated from all parts of the USSR and secrets were turned into leaks.
Therefore, in my military academy, me and my colleagues were not allowed to particularly serious secrets.
Because of this, it became difficult to navigate and understand what was planned above.
And the relevance of my own research depended on it. And she (urgency) suffered.
In the meantime, the USSR collapsed, and my secret dissertation remained in the storage room to collect dust on the shelf, and I was engaged in completely different things.
In particular, the question of survival in the conditions of Ukraine, where nobody needed my knowledge and skills.
In general, most of my career in the Soviet Army, I quickly fled. But - in place.
Article posted with the kind permission of the author.
Source: https://habr.com/ru/post/371865/
All Articles