Research On Modem Techniques For BICM-ID Systems

Motivated by its low complexity, strong agility, high spectrum efficiency, and excellent performance, bit-interleaved coded modulation with iterative decoding (BICM-ID) has been a key technique of next generation of wireless communication, which includes 4G system and IEEE 802.11n. The research on the modulator of BICM-ID and bit-interleaved space-time coded modulation with iterative decoding (BI-STCM-ID), which combines BICM-ID with multi-antennas and space-time codes, mainly focuses on FEC, bit interleaver, and labeling maps etc., and the design of the corresponding demodulator includes joint iterative equalization and decoding, soft in soft out (SISO) demapper etc. There are still many problems to be resolved, such as optimized labeling mapper, low-complexity SISO demapper etc. Under these considerations, some advanced modem algorithms of BICM-ID systems are proposed in this paper.Firstly, the analysis methods of BICM-ID system are studied, which include the performance bound of BICM-ID and BI-STCM-ID, and performance bound with more precision is discussed further. Because EXtrinsic Information Transfer (EXIT) chart gives only the relationship of mutual information between SISO demapper and SISO decoder with a fixed SNR value, two new charts derived from EXIT chart are proposed, which are called as output mutual information chart with Defined Iterative Degree (DID Chart) and Near Optimal output Mutual Information chart (NOMI Chart) respectively. Different from EXIT chart, they can show the iterative performance on the whole signal-to-noise ratio range with one single curve, whereas computation complexity is greatly reduced compared with conventional BER performance curve. As an efficient analyzing tool, DID chart and NOMI chart are employed in other chapters of this paper. One application of the new charts is to implement iterative control according to a near-optimal iterative degree vector determined by NOMI chart. By this low-complexity control algorithm, low delay can be achieved.Secondly, Optimization of the BICM-ID's modulator is addressed. We analyze various modulating factors, such as FEC, interleaver, and labeling maps. Designing guidelines of them are presented also. For example, good iterative performance can be achieved if the labeling maps with lower error floor are combined with simple FECs. Optimization of labeling maps is addressed in this section, which includes conventional two-dimensional maps, multi-dimensional maps and symbol maps of multi-antenna systems. One side, by analysis of Error Free Bound (EF Bound) and Character parameters of various labeling maps, main advantages and disadvantages of different labelings are illustrated; on the other side, computer simulations and EXIT charts (or DID, NOMI charts) are used to prove our results further. We present the optimal labeling map is strongly related with the crucial designing targets of the system, such as SNR regions, complexity, BER performance etc. A Defined Region Optimal (DRO) symbol mapper is proposed by using uncertain apriori information based on the optimized PEP cost function. An adaptive labeling map method is also proposed to obtain a global near optimal performance for the whole SNR region. By using DID chart and NOMI chart, we have resolved two difficult problems efficiently. One is how to select a set, which have the local-optimal performance, from the universal set; the other is how to determine the exact switching SNR point.Besides the optimization of conventional two-dimensional labeling, Multi-Dimensional (MD) labelings of different modulations are exploited. Iterative performance can be improved greatly if the MD labeling is adopted, whereas the complexity of the corresponding MAX-LOG-MAP algorithm is increased. The optimal constructing methods of QPSK and 8-PSK are introduced and their performances are presented. Based on a modified segment BSA algorithm, MD labeling is firstly expended to 16QAM.An Improved low-complexity List Sphere Decoder (ILSD) algorithm was proposed by modifying the search radius of the LSD algorithm adaptively to be the largest radius associated with the candidates in the list. Unlike the conventional LSD with fixed initial radius, adaptive radius was adopted to accelerate the list construction. According to our analysis, computational savings of ILSD over LSD are more apparent with more transmit antennas or more large constellations. Due to its radius-insensitivity, the BER performance can also be improved by selecting a sufficient large initial radius. The modification offers initialization-insensitive radius selection and seems to provide better performance as well. The choice of the radius has become a secondary consideration since we can simply set the initial radius to a relatively large positive number. Its advantages make iterative joint detection and decoding more realizable in a multiple-antenna system, which is regarded as a promising technique for the next-generation wireless communication system.According to the MAX-LOG-MAP algorithm of conventional demappers, a soft in soft out (SISO) demapping algorithm called as MD-MAX-LOG-MAP is derived for BICM-ID with MD mapping. The computational complexity of the MD demapper is further decreased significantly by combining the classical tree-searching algorithm or list sphere decoding algorithm of multi-antenna systems for the first time. As two different methods to select the transmitted symbol vectors with high possibility, they are named as Multi-Dimensional Tree-Searching (MD-TS) algorithm and Multi-Dimensional List Sphere Decoding (MD-LSD) algorithm respectively. The reduced complexity is in proportion to the list length. Computer simulations show that the asymptotic performance is improved compared with conventional mappings and the simplified MD demapping algorithm brings only neglectable performance degradation with proper parameters.BI-STCM-ID system is not suitable for multi-path fading channel, which exhibits frequency-selective fading properties. To overcome the frequency selectivity, Orthogonal Frequency Division Multiplexing (OFDM) is an ideal technique. Joint iterative detection and decoding scheme is exploited for a bit-interleaved space-frequency coded modulation system (BI-SFCM) with high data rate, which is also called as BICM MIMO-OFDM. A simplified channel estimation utilizing channel frequency-domain correlation is proposed to decrease FFT order and provide channel state information (CSI) for the MAP detector. ILSD is utilized to implement SISO demapper/detector of the MIMO-OFDM system. The computation complexity of ILSD can be further reduced by the simplified channel estimation. As a promising method, the BICM MIMO-OFDM scheme combining the simplified channel estimation with ILSD algorithm is robust to frequency selective fading channels based on extensive simulations over various timing-vary multi-path channel models. Furthermore, the degraded performance can be compensated by increasing the iterative number.