International Journal of Engineering and Advanced Technology (IJEAT) ISSN: 2249 – 8958, Volume-9 Issue-3, February, 2020 438 Retrieval Number: C4675029320/2020©BEIESP DOI: 10.35940/ijeat.C4675.029320 Published By: Blue Eyes Intelligence Engineering & Sciences Publication Abstract: In this paper is presented a novel area efficient Fast Fourier transform (FFT) for real-time compressive sensing (CS) reconstruction. Among various methodologies used for CS reconstruction algorithms, Greedy-based orthogonal matching pursuit (OMP) approach provides better solution in terms of accurate implementation with complex computations overhead. Several computationally intensive arithmetic operations like complex matrix multiplication are required to formulate correlative vectors making this algorithm highly complex and power consuming hardware implementation. Computational complexity becomes very important especially in complex FFT models to meet different operational standards and system requirements. In general, for real time applications, FFT transforms are required for high speed computations as well as with least possible complexity overhead in order to support wide range of applications. This paper presents an hardware efficient FFT computation technique with twiddle factor normalization for correlation optimization in orthogonal matching pursuit (OMP). Experimental results are provided to validate the performance metrics of the proposed normalization techniques against complexity and energy related issues. The proposed method is verified by FPGA synthesizer, and validated with appropriate currently available comparative analyzes. Keywords: Compressive sensing, Fast Fourier transform, Orthogonal matching pursuit, FPGA, etc. I. INTRODUCTION In recent years high speed and improved end quality results are emerged as important aspects of many digital systems like Clinical imaging [1], wireless communication [2] and IoT Applications [3] which give rise to both bandwidth and frequency. In general, almost in all fields of signal and image processing, data acquisition using sampling theory has been primarily used to narrow down the penalty gap that exists between desired qualities over rate of signal acquisition [4]. However, sampling the information based on Nyquist is unsuitable for many applications, often causes storage capacity burden or hardware complexity overhead. Additionally, owing to the inclusion of wideband and high throughput signal processing, this conventional method requires a high sampling rate, which tends to energy consumption problems. On the other side, from the basis of the sparse representation, signals parts can be discarded to attain some level of compression without compromising any sort of end Revised Manuscript Received on February 05, 2020. * Correspondence Author Alahari Radhika*, Research Scholar, Department of ECE, JNT University, Kakinada, AP, India. Email: radhialahari@gmail.com . K. Satya Prasad, Department of ECE, Rector, Vignan`s Deemed to be university, Guntur, AP, India. Email: prasad_kodati@yahoo.co.in . K. Kishan Rao, Department of ECE, Director-FD, Srinidhi institute of science and technology, Hyderabad, Telangana, India. Email: Kishanrao6@gmail.com . results quality which is referred to as Compressive Sampling (CS). In this method signals are acquired with some measured value and then utilize some unique algorithm at the receiver side to restore the input signal from the down rated measured values. It has advantage of least possible samples requirement for accurate reconstruction of input signal which is far less as compared to its counterpart sampling theory. In CS method, signals are represented sparsely based on prior statistics and characteristics of the signal to be compressed based on some orthogonal basis. Some of the prominent basis widely preferred is discrete cosine transform (DST), Fast Fourier Transform (FFT), wavelet, Gabor, etc. In general signal reconstruction from the measured values normally comes with several problems that need to be solved and the solution provided to mitigate these problems should be evolved with some optimization. Numerous methods investigated the influence of compressive sampling over effective signal representation. Some works also focused on the use of greedy methods for signal reconstruction [5]. This work aims to propose highly optimized FFT core for active correlation optimization in CS signal reconstruction which is a crucial step in CS analyses[6, 7]. The major contributions of this paper towards CS reconstruction are as follows: (i) the twiddle factor normalization using radix-2 k framework, which has low complexity and prominent impact since CS analysis always requires large number of computational resources due to its correlation computations. (ii) Conventional hardware optimization models in FFT computations come with performance trade off measures. In contrast, in this work, the hardware complexity and energy consumption problems are addressed without using any arithmetic computational techniques. Another metric of this FFT core is that it can perform FFT computation at high speed whereas conventional FFT methods always require pipelining or parallel process to accomplish this task. (iii) Due to the use of non trivial twiddle factors, the FFT computation process is robust to any sort of arithmetic error owing to its fixed width word constrain and can provide improved signal recovery as compared to other FFT methods. The organization of remaining part of the paper is as follows: In Section 2 is described various FFT hardware optimization models; Section 3 explores the potential metrics of FFT model along with the CS reconstruction algorithm framework. Radix factorization technique is elaborated in Section 4 based on index mapping framework for low complexity and energy efficiency. Experimental results and comparative analyses are addressed to demonstrate the area and power efficiency of this FFT twiddle factor normalization scheme, finally with a summary in Section 5. Low Complexity FFT Factorization for CS Reconstruction Alahari Radhika, K. Satya Prasad, K. Kishan Rao