Design and Application of Data Detection in the Presence of Two-Dimensional Interferences
Date Issued
2011
Date
2011
Author(s)
Chen, Chi-Yun
Abstract
Recently, two-dimensional (2-D) interferences have become inevitable in channels of many scenarios, such as inter-carrier interference (ICI) and inter-antenna interference (IAI) in mobile multiple-input multiple-output (MIMO)-orthogonal frequency division multiplexing (OFDM) systems or 2-D inter-pixel interference (IPI) in holographic data storage (HDS) systems. Conventional solutions treat 2-D interferences separately and often suffer from the notorious error propagation problem. On the other hand, dealing with 2-D interferences simultaneously is infeasible due to extremely high complexity.
This thesis proposes a Turbo Receiver with Interference-aware Dual-List (TRIDL) detection that enjoys the advantages of bit-interleaved coded modulation (BICM)-with iterative decoding (ID). In this architecture, an Interference-aware Dual-List (IDL) detection algorithm is designed to combat 2-D interferences and achieve near maximum a posteriori (MAP) performance by exploiting dual lists. The list-based concept further eliminates redundant calculation, leading to benefits of low complexity.
The strength in error rate performance of the IDL detector is shown in two applications: mobile MIMO-OFDM and HDS systems. This thesis not only derives an upper bound on the uncoded symbol error rate (SER) but also provides insightful discussion about the bit error rate (BER) performance through Monte Carlo simulations. Concerning feasibility, this thesis also presents several techniques to reduce complexity in arithmetic operations. By eliminating redundant calculation and sharing common sub-blocks with initial detection block, up to 99.63% of complex additions, 99.69% of complex multiplications and 67.25% of comparisons are saved in the mobile MIMO-OFDM system. As a result, TRIDL running for 10 iterations requires around 4.3-fold complex additions, 2.5-fold complex multiplications and 36-fold comparisons used by the conventional non-iterative algorithm. Similar strategies also eliminate about 57.40% of arithmetic operations in the binary HDS system while over 73.78% is saved in the 3-bit gray-scale HDS system, resulting in similar number of additions and 50% of the multiplications used by the conventional algorithm.
The thesis also discusses hardware design of the TRIDL algorithm. The entire TRIDL detection for a 4x4 MIMO-OFDM system with 16QAM and 64QAM as well as for a 3-bit gray-scale HDS system are designed and verified in FPGA. We exploit techniques such as resource sharing, pointer-based delay buffer, and folding transform to achieve low-power and low-complexity design. In consequence, experimental results indicate about 34.8% reduction of resources in TRIDL for the MIMO-OFDM system.
Subjects
Multiple-input multiple-output (MIMO)-orthogonal frequency division multiplexing (OFDM)
Holographic data storage (HDS)
Maximum a-posteriori (MAP) detection
Iterative receiver
Inter-carrier interference (ICI)
Hardware design
Type
thesis
File(s)![Thumbnail Image]()
Loading...
Name
ntu-100-F93943006-1.pdf
Size
23.32 KB
Format
Adobe PDF
Checksum
(MD5):7af6930fde6fd5dd4dc2c335ee0224a3