The document summarizes the steps to implement a 128-point FFT algorithm in C code. It discusses: 1) How to calculate the reverse order of array subscripts following a carry rule, 2) Looping from 1 to N-2 to fully cover all situations when N=128, 3) Calculating twiddle factors at each level of the butterfly operations, 4) Performing M levels of operations by looping M times and calculating butterflies, 5) Combining the long array into complex numbers and deriving real and imaginary parts after FFT, 6) Setting a Hamming window function. The overall code defines necessary functions and arrays.
The document summarizes the steps to implement a 128-point FFT algorithm in C code. It discusses: 1) How to calculate the reverse order of array subscripts following a carry rule, 2) Looping from 1 to N-2 to fully cover all situations when N=128, 3) Calculating twiddle factors at each level of the butterfly operations, 4) Performing M levels of operations by looping M times and calculating butterflies, 5) Combining the long array into complex numbers and deriving real and imaginary parts after FFT, 6) Setting a Hamming window function. The overall code defines necessary functions and arrays.
The document summarizes the steps to implement a 128-point FFT algorithm in C code. It discusses: 1) How to calculate the reverse order of array subscripts following a carry rule, 2) Looping from 1 to N-2 to fully cover all situations when N=128, 3) Calculating twiddle factors at each level of the butterfly operations, 4) Performing M levels of operations by looping M times and calculating butterflies, 5) Combining the long array into complex numbers and deriving real and imaginary parts after FFT, 6) Setting a Hamming window function. The overall code defines necessary functions and arrays.
The document summarizes the steps to implement a 128-point FFT algorithm in C code. It discusses: 1) How to calculate the reverse order of array subscripts following a carry rule, 2) Looping from 1 to N-2 to fully cover all situations when N=128, 3) Calculating twiddle factors at each level of the butterfly operations, 4) Performing M levels of operations by looping M times and calculating butterflies, 5) Combining the long array into complex numbers and deriving real and imaginary parts after FFT, 6) Setting a Hamming window function. The overall code defines necessary functions and arrays.
-The original array subscript I is arranged in order, then the next array must be obtained by adding 1 to the previous array subscript, that is, it satisfied right-to-left carry rule. For example: the next number of 1011001 must be 1011010. Then after the code position is reversed, it will satisfy the the carry rule from left to right, for example, the code point of 1011001 is in reverse order of 1001101, the next number is 1011010, and the code point in reverse order is 0101101. It can be seen that 0101101 is obtained by adding 1 to 1001101 from left to right.
-When N=128, the original subscript is 1101111, which means that the decimal system is 111, and its reverse order. It is 1111011, which means 123, so half of the traversal does not fully cover all situations. So we should go from 1 to N-2 (the inverse of 0 and N-1). The framework is as follows:
Next, we consider how to get the reverse order J. The rules of carry from left to right can be understood in this way. After adding 1 from left to right, if the highest bit is 1, then the bit becomes 0, and the search continues, if it is still 1, the bit also becomes 0, until the next bit is 0, change 0 to 1, stop searching; if the highest bit is 0, it will directly change to 1 and stop searching. Since we start from 1, the reverse order of 1 must be N/2, Therefore, the initial value of J is set to FFTN/2, and the procedure is as follows: -First of structure programing, we should complete M-level operations, so we first need to have M cycles, and secondly, for each level of butterfly operations, we know that a total of calculated times, but not all calculations use the same twiddle factor, so you need to calculate each twiddle factor first. two arrays are needed and corresponding to the real and imaginary parts respectively, so the overall procedure of the FFT operation part is as follows:
-Long array, combined into a complex array, bring it into the FFT program to calculate the FFT, and then derive the two arrays from the calculation results to perform separately.FFT results, and finally the FFT results of the two arrays are subjected to a butterfly transform, and the final real FFT result is obtained. -Set the hamming window function:
The above is the overall code of the optimization program, in which functions and arrays such as cos_fft are defined as follows: It is a summary of my original exploration of FFT Implement C++ code for 128-points FFT(radix 4) alogorithm: -Set FFT.h: -Main c++ code:
