壓縮感測還原演算法晶片設計

VLSI Implementation of a Compressed Sensing Recovery

林冠廷(Kuan-Ting Lin)

指導教授 ： 蔡尚澕

國立交通大學/電機學院/電機與控制工程系所/碩士(2012年)

https://doi.org/10.6842/NCTU.2012.01022

摘要

這篇論文中提出了兩個壓縮感測還原演算法改良自子空間追尋，這兩個演算法降低了源演算法的複雜度而且仍然保持著高還原率。這兩個提出的演算法和其他現有演算法的複雜度分析以及模擬結果都描述於此篇論文內，我們可以觀察到我們所提出的演算法擁有良好的效能。除此之外，此篇論文也將其中一種提出的還原演算法實做為電路，也特別設計一些硬體架構來完成此電路，我們所用的製程為台積電90奈米製程，此電路的時脈速度為100MHz，面積為11.69 mm2，而平均消耗功率為431 mW。

關鍵字

壓縮還原

並列摘要

This paper proposes two recovery algorithms modified from subspace pursuit(SP)
for compressed sensing problems. The two algorithms can reduce the complexity of
SP and maintain the high recovery rate. Complexity analysis and simulation results
are provided for the proposed algorithms and other conventional recovering
algorithms. We observe that the proposed algorithms can perform well. Besides, this
paper also proposes an architecture for one of the proposed recovering algorithm for
VLSI implementation. Several hardware units are dedicated designs for the
implementation. The proposed chip is implemented using a TSMC 90nm process and
can use a clock rate of 100MHz with total area is 11.69 mm2. The average power is
431mW.

並列關鍵字

Compressed Sensing

參考文獻

[1] D. Donoho, “Compressed sensing,” IEEE Trans.Inf. Theory, vol. 52, no. 4, pp. 1289-1306, Apr. 2006.

[2] E. Candes and T. Tao, “Decoding by linear programming”, IEEE Trans. Inf. Theory, vol51, no. 12, pp. 4203-4215, Dec. 2005

[3] E. Candes and T. Tao. “Near optimal signal recovery from random projections: Universal encoding strategies,” IEEE Transacation on Information Theory, vol. 52, pp.5406-5425, 2006

[4] E. Candes, R. Mark, T. Tao,“ Error correlation via linear programming.” in Proc IEEE Symp. Foundations of Computer Science(FOCS), Pittsburgh, PA, Oct. 2005, pp 295-308

[5] E. Candes, J. Romberg, and T. Tao. “Robust uncertainty principles: Exact signal reconstruction from highly incomplete frequency information,” IEEE inf. Theory, 52(2):489509, 2006.