A new approach to pipeline FFT processor
Summary (1 min read)
I. INTRODUCTION
- Pipeline FFT processor is a specified class of processors for DFT computation utilizing fast algorithms.
- The architecture design for pipeline FFT processor had been the subject of intensive research as early as in 70's when real-time processing was demanded in such application as radar signal processing [SI, well before the VLSI technology had advanced to the level of system integration.
- Here different approaches will be put into functional blocks with unified terminology, where the additive butterfly has been separated from multiplier to show the hardware requirernent distinctively, as in Fig. 1 .
- The input sequence has been broken into two parallel data stream flowing forwatrd, with correct "distance" between data elements entering the butterfly scheduled by proper delays.
- By the observations made in last section the most desirable hardware oriented algorithm will be that it has the same number of non-trivial multiplications at the same positions in the SFG as of radix-4 algorithms, but has the same butterfly structure as that of radix-2 algorithms.
I v . R2'SDF ARCHITIECTURE
- Mapping radix-2' DIF FFT algorithm derived in last section to the R2SDF architecture discussed in section 11, a new architecture of Radix-2' Single-path Delay Feedback (R2'SDF) approach is obtained.
- Fig. 4 outlines an implementation of the R2'SDF architecture for N = 256, note the similarity of the datapath to R2SDF and the reduced number of multipliers.
- A BF I1 (log, N)-bit binary counter serves two purposes: synchronization controller and address counter for twiddle factor reading in each stages.
- The input data from left is directed to the shift registers until they are filled.
- On next N/2 cycles, the multiplexors turn to position "1" the butterfly computes a 2-point DFT with incoming data and the data stored in the shift registers.
V. CONCLUSION
- A hardware-oriented radix-2' algorithm is derived which has the radix-4 multiplicative complexity but retains radix-2 butterfly structure in the SFG.
- Based on this algorithm, a new, efficient pipeline FFT architecture, the R2'SDF architecture, is put forward.
- The hardware requirement of proposed architecture as compared with various approaches is shown in Table 1 , where not only the number of complex multipliers, adders and memory size but also the control complexity are listed for comparison.
- For easy reading, base-4 logarithm is used whenever applicable.
- It shows R2'SDF has reached the minimum requirement for both multiplier and the storage, and only second to R4SDC for adder.
Did you find this useful? Give us your feedback
Citations
163 citations
Cites methods from "A new approach to pipeline FFT proc..."
...Algorithms including radix-4 [2], split-radix [3], radix- [4] have been developed based on the basic radix-2 FFT approach....
[...]
...In this paper, we propose several novel parallel-pipelined architectures for the computation of RFFT based on radix- [4] and radix- algorithms [28]....
[...]
...The other approaches [4] are not specific for the RFFT and can be used to calculate the CFFT....
[...]
...We can simply use the Radix-2 single-path delay feedback approach presented in [4], with just modifying the first complex butterfly stage into real....
[...]
120 citations
Cites methods from "A new approach to pipeline FFT proc..."
...[2005], and Johnson et al. [1990] use this type of representation to specify and generate software implementations of the FFT....
[...]
...Similarly, Kee et al. [2008] target the FFT on an FPGA, expressing a radix 2 FFT algorithm as a double loop using National Instruments LabVIEW, where each loop can then be unrolled by the tool....
[...]
99 citations
Cites background from "A new approach to pipeline FFT proc..."
...Such memory requirement may be forbidden if Ns is large, because the area of memory does not shrink as much as that of logic gates when fabrication technology advances, due to the use of sense amplify circuitry....
[...]
83 citations
83 citations
Cites background from "A new approach to pipeline FFT proc..."
...The authors are with the CEIT and TECNUN, University of Navarra, 20018 San Sebastián, Spain (e-mail: acortes@ceit.es; ivelez@ceit.es; jfsevillano@ceit. es)....
[...]
...Multiple-path architectures, such as [1]–[5], where data is input using several parallel paths, are used when the throughput needs to be increased for a given clock frequency of the FFT processor....
[...]
References
3,249 citations
623 citations
327 citations
"A new approach to pipeline FFT proc..." refers methods in this paper
...By the observations made in last section the most desirable hardware oriented algorithm will be that it has the same number of non-trivial multiplications at the same positions in the SFG as of radix-4 algorithms, but has the same butterfly structure as that of radix-2 algorithms....
[...]
304 citations
204 citations
"A new approach to pipeline FFT proc..." refers methods in this paper
...R4SDC: Radix-4 Single-path Delay Commutator [10] uses a modified radix-4 algorithm with programable 1=4 radix-4 butterflies to achieve higher, 75% utilization of multipliers....
[...]