An extensible and real-time compressive sensing reconstruction hardware for WBANs using OMP

Shi, Weijing; Yi, Li; Han, Jun; Cheng, Xu; Zeng, Xiaoyang

doi:10.1109/asicon.2013.6811911

Cited by 2 publications

(4 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The number of iterations is 16 for a 16-sparse signal. Hence, the number of samples processed per second is 256/(32 × 3.2 × 16 × 10 −9 )=156.25 × 10 6 whereas the existing design of [8] can process 256/(474 × 4.15 × 16 × 10 −9 ) = 8.133 × 10 6 samples per second. As the input word length is set to 16, the proposed design can support a processing speed of up to 156.25 × 10 6 × 16=2.5 × 10 9 bits per second (bps), which is faster than the speed required for the reconstruction of a 1080p, 30fps, 4:2:2 YCbCr HD video stream.…”

Section: Synthesis Results and Comparisonmentioning

confidence: 99%

“…In Table II, we show the synthesis results of the proposed designs and existing designs [8,9] in terms of the maximum frequency, clock period, number of cycles (NOC) per frame, reconstruction time, area, and area-delay-product (ADP). The second and third column of Table II show the synthesis results with 90nm CMOS library of the proposed and existing design [8] when the input word length is 16 bits, M = 64, N = 256, and K = 16. The proposed design operates at a maximum frequency of 312MHz and completes the processing of every 256-sample frame in 32 3.2-nanosecond cycles.…”

Section: Synthesis Results and Comparisonmentioning

confidence: 99%

“…In the existing hardware implementation of the OMP [8,9], a signal frame is generally processed as shown in Fig. 1, using three circuits: the column selector, least-square problem (LSP) solver, and residue calculator.…”

Section: Hardware Architecturementioning

confidence: 99%

“…(3) can be solved by greedy iterative methods. For hardware implementation, greedy methods such as matching pursuit (MP) and orthogonal matching pursuit (OMP) are more preferable for their simplicity, high speed of reconstruction and low implementation cost [7,8,9]. …”

Section: Compressive Sensingmentioning

confidence: 99%

See 3 more Smart Citations

High-performance ASIC realization of orthogonal matching pursuit algorithm

Nguyen

Park

2018

IEICE Electron. Express

View full text Add to dashboard Cite

Compressive sensing (CS) is perceived as a breakthrough in sampling theory, as it proves that sparse signals can be reconstructed from fewer samples than required by the Shannon-Nyquist theorem. However, CS can hardly be applied to real-time applications because reconstruction algorithms are computationally demanding. To tackle this problem, in this paper, we propose a new high-speed architecture for implementing the orthogonal matching pursuit (OMP), which is one of the most popular algorithms for CS reconstruction. Specifically, a novel pipelined systolic architecture and an optimized scheduling strategy are proposed. From the synthesis results, we find that the proposed design takes 1.638 µs to reconstruct 16-sparse signal, which is 19.2 times faster than the existing VLSI implementation of the OMP.

show abstract

Section: Synthesis Results and Comparisonmentioning

confidence: 99%