Rf and if digitization in Radio Receivers: Theory, Concepts, and Examples

For most radio receiver applications, real-time operation is important. In many types of 

processing, such as computing Fast Fourier Transforms (FFTs), the data is partitioned into 

blocks of a finite length. Processing is performed on the entire block of data. In this block type 

processing, assuming a single processor, real-time operation essentially means that all processing 

on a given block of data (including any required data transfers) is completed before all of the 

next block of data to be processed is captured. This concept is illustrated in Figure 12 [26].  

Figure 12. Real-time processing for block data using a single processor. 

capture all of the next block of data (again assuming a single processor), data collection must be 

stopped until the processing is completed. At that time data collection may resume. Under these 

conditions some of the input data is missed and is not processed. This is an example of 

processing that does not take place in real time. Depending upon the application and the amount 

of data lost, this may not be acceptable.  

Data Block 1 

Data Block 2 

Data Block 3 

Processing 

Time for Data 

Block 1 

Time for Data 

Block 2 

t 

 

29 

This problem can be alleviated by using two or more processors operating cooperatively. This 

general technique is called multiprocessing and is used frequently. To illustrate how 

multiprocessing can be used to speed up overall data throughput, consider the previous example 

of data partitioned into data blocks but with two processors available instead of one. As shown in 

Figure 13, processor 1 operates on one block of data and processor 2 operates on the next block 

of data. The processors continue to operate on alternating data blocks. The processed data output 

is obtained by switching back and forth between the outputs of the two processors. This is 

sometimes called a “ping-pong” technique. Using this technique, the processing time of each 

processor can take longer than the time to capture the next block of data and still provide real-

time output. The processing time cannot exceed the time to capture the next two blocks of data, 

however. This technique can be extended to more than two processors to achieve even faster 

overall data throughput.  

 

Figure 13. Real-time processing for block data using two processors. 

tải về 0.99 Mb.

Chia sẻ với bạn bè của bạn: