
For FullText PDF, please login, if you are a member of IEICE,
or go to Pay Per View on menu list, if you are a nonmember of IEICE.

DCDBased Branch and Bound Detector with Reduced Complexity for MIMO Systems
Zhi QUAN Ting TIAN
Publication
IEICE TRANSACTIONS on Communications
Vol.E101B
No.10
pp.22302238 Publication Date: 2018/10/01
Online ISSN: 17451345
DOI: 10.1587/transcom.2017EBP3336
Type of Manuscript: PAPER Category: Wireless Communication Technologies Keyword: DCD, BB, low complexity,
Full Text: PDF(1.1MB) >>Buy this Article
Summary:
In many communications applications, maximumlikelihood decoding reduces to solving an integer leastsquares problem, which is NPhard in the worst case. It has recently been shown that over a wide range of dimensions and SNRs, the branch and bound (BB) algorithm can be used to find the exact solution with an expected complexity that is roughly cubic in the dimension of the problem. However, the computational complexity becomes prohibitive if the SNR is too low and/or the dimension of the problem is too large. The dichotomous coordinate descent (DCD) algorithm provides low complexity, but its detection performance is not as good as that of the BB detector. Two methods are developed to bound the optimal detector cost to reduce the complexity of BB in this paper. These methods are DCDbased detectors for MIMO and multiuser detection in the scenario of a large number of transmitting antennas/users. First, a combined detection technique based on the BB and DCD algorithms is proposed. The technique maintains the advantages of both algorithms and achieves a good tradeoff between performance and complexity compared to using only the BB or DCD algorithm. Second, since the first feasible solution obtained from the BB search is the solution of the decorrelating decision feedback (DF) method and because DCD results in better accuracy than the decorrelating DF solution, we propose that the first feasible solution of the BB algorithm be obtained by the boxconstrained DCD algorithm rather than the decorrelating DF detector. This method improves the precision of the initial solution and identifies more branches that can be eliminated from the search tree. The results show that the DCDbased BB detector provides optimal detection with reduced worstcase complexity compared to that of the decorrelating DFbased BB detector.

