Design and Implementation of a New Complex Sphere Decoder for MIMO Detection
Date Issued
2006
Date
2006
Author(s)
Lee, Tsung-Hsueh
DOI
zh-TW
Abstract
More and more services are provided in the next generation wireless communicaiton systems, including on-line videos, high definition TV (HDTV), interactive games and so on. The demand for throughput and QoS is getting higher. However, the bandwidth is limited. OFDM is widely adopted in many communication systems and has become a main modulation technique of many standards. MIMO will be the next widely accepted technique because of its provision for diversity and/or spectrum efficiency.
To apply the MIMO techniques, the number of transmitting and receiving antennas needs be more than one. Among spatial multiplexing MIMO detection, sphere decoding algorithm is the first that was proposed to do maximum-likelihood equalization’detection with acceptable complexity.
In the thesis, sphere decoder related literatures are surveyed and a new complex-plane sphere decoder is designed and implemented. Depth-first search with closest point first order is introduced directly on complex-valued signals. List-Enumeration algorithm is proposed to overcome the problem of enumeration in complex-plane sphere decoder. The list can be implemented by either an unified list and several individual lists. In addition, backward-Layer controlled algorithm and diminished dimensionality algorithm are proposed to further reduce the hardware complexity. Two architectures, parallel forward/backward search and single calculation unit, are proposed to implement the new algorithm. The tradeoffs of these two versions are between area, power, and throughput. The hardware can support different constellations and numbers of antennas. Simulation and implementation results indicate the proposed algorithm and architecture outperform other solutions and form a solid foundation for future wireless communication systems that adopt MIMO processing.
Subjects
多輸入多輸出
球面解碼
表列舉
最大相似
返回階層控制
MIMO
sphere decoding
list enumeration
maximum likelihood
backward-layer controlled
Type
thesis
File(s)![Thumbnail Image]()
Loading...
Name
ntu-95-R93943091-1.pdf
Size
23.31 KB
Format
Adobe PDF
Checksum
(MD5):1d79281aa0681fc58a32dae7d01344cf