Detection and decoding for two-dimensional interference channels
Date of Issue2013
School of Electrical and Electronic Engineering
Data Storage Institute, Data Storage Systems Center
In this thesis, we study the problem of communication over two-dimensional (2-D) interference channels which represent an important class of data storage and communication systems. The maximum-likelihood page detection (MLPD) for 2-D interference channels is firstly studied and the lower and upper performance bounds are derived. Although MLPD is computationally prohibitive and unachievable in practice, the performance bound of MLPD provides a benchmark of the optimal detection of 2-D interference channels. We also propose an approximated bounding method which can accurately predict the performance bound while greatly reducing the computational complexity. A joint iterative detection/decoding scheme (JIDDS) is proposed for 2-D interference system with low-density parity-check (LDPC) codes. The channel detector is composed of two constituent detectors that both adopt trellis-based maximum a posteriori probability (MAP) algorithms and they work in the down-track direction and cross-track direction, respectively. To achieve better performance, iterations are introduced between the two constituent detectors and among the constituent detectors with LDPC decoder. Simulation results show that the proposed scheme outperforms some existing algorithms. The better performance of JIDDS is achieved at the cost of higher computational complexity. A reduced-state channel detector is also proposed to save considerable complexity of the system but only degrade system performance marginally. In JIDDS, the channel detector and LDPC decoder work individually, which may result in some delays in detection/decoding. An appealing practical aspect of the decoding of LDPC codes is that it consists of many small, independent decoding functions (i.e., the variable nodes and check nodes) which can work in parallel. In order to make use of the low-delay advantage of LDPC codes, a joint message-passing decoding algorithm (JMPDA) of LDPC codes and 2-D interference channels is proposed. The channel detector consists of many state-processing nodes which can be implemented in a completely parallel manner in conjunction with the decoding of LDPC codes. The performance of the JMPDA is almost the same as that of the JIDDS. An advantage of the JMPDA is its parallel implementation, leading to a high-speed decoder. Lastly, the extrinsic information transfer (EXIT) chart and density evolution (DE) are adopted in the analysis of the decoding of LDPC coded 2-D interference systems. The EXIT chart visualizes the evolution of mutual information between channel detection and LDPC decoding during the iterations. The effects of parameters to the decoding system are investigated by analyzing the EXIT curves. It is shown that given a good combination of parameters, the proposed algorithm will converge to the correct result. A graph representation of the JMPDA is explained. It is proved that the message-flow neighborhood of the JMPDA will be tree-like for a sufficiently long code length. It is further proved that the average performance of the JMPDA will concentrate around the performance in which message-flow neighborhood is tree-like. Based on the tree-like message-flow neighborhood, a modified DE algorithm is employed to track the message densities during the iterations. A threshold is then calculated using the DE algorithm which represents the performance limit of the system.
DRNTU::Engineering::Electrical and electronic engineering::Electronic systems::Signal processing