Abstract
Shalvi and Weinstein's single-input single-output (SISO) super-exponential algorithm (SEA) for blind econvolution was extended to multichannel SEA by Yeung and Yau, followed by Feng and Chi's lattice super-exponential algorithm (LSEA) for SISO systems. In this thesis, an LSEA for multi-input multi-output (MIMO) systems is proposed, that is an extension of Feng and Chi's LSEA for SISO systems, both sharing the merits of modularity, low sensitivity to parameter quantization effects as well as lower computational load. Furthermore, a 2-step LSEA for MIMO systems, also an extension of Feng and Chi's 2-step LSEA for SISO systems, is proposed that converges faster than the Yeung and Yau's multichannel SEA in addition to the preceding merits of the proposed LSEA for MIMO systems. In this thesis, we further propose a 2S-LSEA for MIMO systems that also shares all the preceding merits of the 2S-LSEA for SISO systems, and outperforms Yeung and Yau's SEA for MIMO systems with much faster convergence speed, much lower computational complexity, and more reliable performance. Some simulation results are presented to support the efficacy of the proposed 2S-LSEA for MIMO systems. The thesis is organized as follows. Section 2 presents a lattice super-exponential algorithm (LSEA) for MIMO systems. In Section 3, we propose a 2S-LSEA algorithm. Section 4 presents some simulation results. Finally, we draw some conclusions.