Use this URL to cite or link to this record in EThOS:
Title: High performance adaptive MIMO detection : from algorithm to implementation
Author: Zheng , C.
ISNI:       0000 0004 2720 6332
Awarding Body: Queen's University Belfast
Current Institution: Queen's University Belfast
Date of Award: 2012
Availability of Full Text:
Full text unavailable from EThOS.
Please contact the current institution’s library for further details.
Adaptive-Multiple Input Multiple Output (A-MIMO) techniques are regarded as one of the core techniques in 3G/4G wireless communication systems and beyond. One of the key challenges for A- MIMO systems is to retrieve the spatially mixed transmitted symbols at the receiver. However, only a few high performance detection algorithms have been successfully implemented to achieve real-time performance. This thesis concentrates on developing high performance adaptive modulation MIMO detection algorithms and Model Based Design (MBD) techniques that bridge the gap between detection algorithms and efficient embedded implementations. From the algorithm perspective, this work proposes a novel near-optimal low complexity detection algorithm, Real-valued Fixed-complexity Sphere Decoder (RFSD). The RFSD is derived to achieve quasi-ML decoding performance as FSD, which is the most promising low complexity high performance parallel detection algorithm in existence, but with over 70% complexity reduction. In addition, an adaptive detection algorithm is proposed. This detection algorithm alleviates the BER degradation current high performance detection algorithms experience and provides up to 46% BER improvement for small constellation dominated hybrid modulated MIMO systems. It also balances detection performance and complexity for MIMO configurations under different environments. From the implementation perspective, a Regular Choice Petri Net (RCPN) is proposed to accurately model and rapid implement the adaptive detection algorithms. The Texas Instruments l"MS320C64+ DSP-based realisations from the RCPN model demonstrate 90% reduction in run-time overhead and 10% reduction in code memory as compared to languages in existence. Furthermore, an MBD design approach is developed to convert RCPN models into embedded implementations, by creating an automated allocation method and introducing the kernel concept from streaming applications into the scheduling process. The resulting FPGA based multi-SIMD implementation achieves real time performance with at least 52.6% less hardware resource or over 65% reduction in mapping complexity as compared to conventional schemes.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available