DocumentCode
627259
Title
A low-power time-interleaved 128-point FFT for IEEE 802.15.3c standard
Author
Ahmed, Toufik
Author_Institution
Grad. Sch. of Inf. Sci., Nara Inst. of Sci. & Technol., Nara, Japan
fYear
2013
fDate
17-18 May 2013
Firstpage
1
Lastpage
5
Abstract
This paper proposed a 128-point FFT for low data rate Audio Visual PHY (AVPHY) mode of IEEE 802.15.3c standard. The architecture process a continuous flow of 2 samples in parallel, leading to a throughput of 317.25 MSamples/s with a symbol duration 492 ns. The FFT computed in seven stages that uses radix-2 butterflies and complex rotations. The radix-2 butterflies and complex rotations are time-interleaved by factor of 2. This technique increases the area slightly for the radix-2 butterflies and complex rotators than the conventional radix-2 butterfly and complex rotator. However, this technique reduces the operating frequency of the radix-2 butterflies and complex rotators, which results the reduction of the power consumption of the whole FFT. Besides, the proposed architecture meets requirements of the throughput and latency. Experimental results reveal that besides its requirements, the design is efficient in area and power consumption, improving the results of previous approaches. Specifically, the proposed design consumes 2.87 mW and its area is 0.57 mm2, while the wordlength is 16 bits.
Keywords
fast Fourier transforms; personal area networks; IEEE 802.15.3c standard; complex rotations; fast Fourier transform; low data rate audio visual PHY mode; low-power time-interleaved 128-point FFT; radix-2 butterflies; Clocks; Computer architecture; Power demand; Program processors; Radiation detectors; Read only memory; Wireless personal area networks; 128-point FFT; feedforward; radix-2; time-interleaved;
fLanguage
English
Publisher
ieee
Conference_Titel
Informatics, Electronics & Vision (ICIEV), 2013 International Conference on
Conference_Location
Dhaka
Print_ISBN
978-1-4799-0397-9
Type
conf
DOI
10.1109/ICIEV.2013.6572612
Filename
6572612
Link To Document