| Summary: | 博士 === 國立清華大學 === 通訊工程研究所 === 96 === To satisfy the demand for ubiquitously accessing multimedia data services, researchers have been continuing to develop various techniques which aim at pushing the overall system throughput to its limit. From the literature, research challenges towards this end include the developments of 1) an efficient space-time encoding and modulation scheme to raise the transmission rate, 2) a practical iterative receiver to recover the corrupted transmit-signals, and 3) a brand-new network architecture with an efficient scheduling algorithm to exploit the possible reuse of spatial resources without mutual interference. Each of the three challenges is not an easy-to-deal-with obstacle, for example, designing an efficient scheduling algorithm involves a complex multi-dimensional optimization which in nature is an NP-complete problem. However, thanks to the recent celebrated break-through in the EXtrinsic Information Transfer (EXIT) chart and the Graph theory, this thesis presents the progress we have made toward solving the three challenges with the help of the EXIT chart and Graph techniques.
First, a close-to-Shannon-limit space-time encoding and modulation scheme for multiple-input multiple-output (MIMO) system is developed with the help of the EXIT chart by researchers at Bell Lab in 2004, which maps the coded bits of an irregular low-density parity-check (LDPC) code directly onto a modulation signal set. Instead of further searching for more capacity-approaching LDPC codes for all the practical MIMO system configuration, we develop several universal-good LDPC codes by investigating the relationship between the LDPC codes and the underlying MIMO system configuration through the EXIT chart and the large-system performance analysis technique. We demonstrate that the ratio of the numbers of the transmit and the receive antennas plays a crucial role in such LDPC-Coded MIMO systems, and the rewards for redesigning LDPC codes are not so much when the antenna ratio is unchanged.
Second, since the channel estimation techniques through Wiener filtering are believed to be efficient methods for receivers to recover the corrupted transmit-signals, it is especially desired to know whether the promising idea of iterative processing can be applied to the receivers consisting of soft-information-aided Wiener-filter-based channel estimators and error-correction decoders. Therefore, we wish to investigate 1) how different variants of the two functional blocks interact, and 2) whether such iterative interactions can improve or degrade the system performance through a unified EXIT chart analysis. It is shown that the iterative processing between Wiener-filter-based channel estimator and error-correction decoder does not always improve the receiver performance, depending on 1) which family the error-correction code is from, and 2) how we utilize the soft information of the coded bits at the Wiener-filter-based channel estimator.
Third, the cellular relay network capable of multi-hop data transmission seems to be a promising network architecture to boost the overall system throughput. However, since simultaneous data transmissions are allowed in such network, it needs an efficient scheduling algorithm to avoid the frequent data collisions problem. Otherwise, significantly higher data packet throughput in a relay network can never be achieved. A novel distributed algorithm tackling with this NP-complete scheduling problem is developed through the recent modeling and computational methodology of factor graphs. Numerical experiments show that the proposed distributed scheduling algorithm not only obtains average packet throughput comparable or even better than some exhaustive-search-based algorithms, but also takes care of the throughput fairness issue among all the MSs.
|
|---|
