1: \begin{abstract}

2: This paper presents a novel adaptive reduced-rank { multi-input

3: multi-output} (MIMO) equalization scheme and algorithms based on

4: alternating optimization design techniques for MIMO spatial

5: multiplexing systems. The proposed reduced-rank equalization

6: structure consists of a joint iterative optimization of two

7: equalization stages, namely, a transformation matrix that performs

8: dimensionality reduction and a reduced-rank estimator that retrieves

9: the desired transmitted symbol. The proposed reduced-rank

10: architecture is incorporated into an equalization structure that

11: allows both decision feedback and linear schemes for mitigating the

12: inter-antenna and inter-symbol interference. We develop alternating

13: least squares (LS) expressions for the design of the transformation

14: matrix and the reduced-rank estimator along with computationally

15: efficient alternating recursive least squares (RLS) adaptive

16: estimation algorithms. We then present an algorithm for

17: automatically adjusting the model order of the proposed scheme. An

18: analysis of the LS algorithms is carried out along with sufficient

19: conditions for convergence and a proof of convergence of the

20: proposed algorithms to the reduced-rank Wiener filter. Simulations

21: show that the proposed equalization algorithms outperform the

22: existing reduced-rank and full-rank algorithms, while requiring a

23: comparable computational cost. \vspace{0.5em}

24: 

25: \begin{keywords}

26: MIMO systems, equalization structures, parameter estimation,

27: reduced-rank schemes.

28: \end{keywords}

29: \end{abstract}