6-4] CORRELATION AND STORAGE RADAR TECHNIQUES 305 



Monopulse techniques are particularly useful for applications where 

 pulse-to-pulse amplitude fluctuations due to target variations or interfer- 

 ence signals can degrade conical or sequential scanning tracking techniques. 



6-4 CORRELATION AND STORAGE RADAR TECHNIQUES 



Signal storage has played a most significant role in the success of radar. 

 From the earliest use of the cathode ray tube in echo ranging with A-scope 

 presentation to modern sophisticated and complex magnetic storage devices 

 for predetection integration, the use of storage has become increasingly 

 important. Today the lack of high-capacity memory, high-speed operation, 

 and wide dynamic range storage are perhaps major contributing factors 

 impeding the development of more effective long-range radar. The in- 

 creased emphasis on integration by storage has been brought about in part 

 by the growing popularity of correlation and information theory methods 

 for signal enhancement. The idea of correlation in itself is not new to radar 

 — the World War II SCR-584 used a limited form of cross-correlation 

 detection to separate the bearing and elevation errors. Here the correlation 

 was not of the statistical nature currently in favor for signal enhancement. 

 For this latter purpose, the cross-correlation device requires some form of 

 storage and integration in order to fulfill its mission. Since storage can be 

 considered a part of the correlation process, we will discuss the more general 

 subject of correlation first. 



Correlation Processes. Two correlation techniques have appeared in 

 radar during the last two decades: (1) autocorrelation, defined mathe- 

 matically as 



^n(r) = lim -^jj^W^i^ - r)dt (6-17) 



-:; 2T 



where t is a time displacement (delay in the case of a radar echo), and 

 (2) cross correlation, defined as 



<P,,{t) = lim ;r^ / /i(/)/2(/ - r)dt. (6-18) 



Both of these techniques are defined in the time domain and exist theoret- 

 ically only in the limit as the total observation time becomes infinite. In 

 practice, of course, infinite time is not available, and it becomes necessary 

 to reinterpret the functions using finite limits. Let us define an incomplete 

 autocorrelation function as 



^„(r,T) = ^/^/iW/i(^ - r)dt (6-19) 



