Kewu Peng

APSK Constellation with Gray Mapping

Amplitude phase shift keying (APSK) constellation is superior to its quadrature amplitude modulation (QAM) counterpart from the mutual information point of view. However, due to the lack of Gray mapping, it introduces high independent demapping loss. In this letter, a special kind of APSK constellations with Gray mapping (Gray-APSK) is proposed, which provides considerable shaping gain compared with the QAM constellations with Gray mapping (Gray-QAM) in both independent and iterative demapping scenarios, as verified by average mutual information analysis and bit error rate simulations.

Bit Division Multiplexing for Broadcasting

This paper aims to improve the throughput of a broadcasting system that supports the transmission of multiple services with differentiated minimum signal-to-noise ratios (SNRs) required for successful receptions simultaneously. We propose a novel multiplexing method called bit division multiplexing (BDM), which outperforms the conventional time division multiplexing (TDM) counterpart by extending the multiplexing from symbol level to bit level. Benefiting from multiple error protection levels of bits within each high-order constellation symbol, BDM can provide so-called nonlinear allocation of the channel resources. Both average mutual information (AMI) analysis and simulation results demonstrate that, compared with TDM, BDM can significantly improve the overall transmission rate of multiple services subject to the differentiated minimum SNRs required for successful receptions, or decrease the minimum SNRs required for successful receptions subject to the transmission rate requirements of multiple services.

Iterative Channel Estimation for Block Transmission with Known Symbol Padding - A New Look at TDS-OFDM

This study proposes an iterative channel impulse response (CIR) estimation scheme for multicarrier block transmission with known symbol padding, in the context of the time-domain synchronous OFDM (TDS-OFDM) system. It is shown that the TDS-OFDM still fulfils the circular convolution relationship even if pseudo-noise (PN) sequence is used as guard interval. While conventional methods perform the channel estimation based on the cyclic prefix reconstruction or residual intersymbol interference cancellation in the time domain, by processing the data part and the known symbol padding in one TDS-OFDM block separately, the new proposal uses the one- tap frequency-domain equalization to separate the data part and the known symbol padding and the CIR can be estimated in an iterative manner. In a typical example for 3780-point FFT based TDS-OFDM system with 16QAM constellations, the mean square error (MSE) of of the proposed channel estimator can be reduced to about 10-5/2 after two iterations when the channel signal-to-noise ratio (SNR) is larger than 8 dB, without the initial CIR estimation. The BER performance of a QPSK TDS-OFDM system shows that the signal-to-noise-ratio (SNR) degradation based on the proposed channel estimator is less than 0.3 dB compared with the lower bound for both Brazil A and B channels.

EXIT-Aided Bit Mapping Design for LDPC Coded Modulation with APSK Constellations

In this paper, the bit mapping technique is applied to the low-density parity-check (LDPC) coded modulation system with Gray-APSK constellations for the first time. A multi-grid searching algorithm (MGSA) based on extrinsic information transfer (EXIT) chart analysis is first proposed to optimize the bit mapping distribution, after which a low-complexity bit mapper consisting of a structural bit interleaver and a demultiplexer (DEMUX) is devised. Compared to the LDPC coded QAM system specified in DVB-T2, the proposed APSK system with the optimized DEMUX can achieve up to 0.83dB gain under additive white Gaussian noise (AWGN) channel.

Coded Modulation with Signal Space Diversity

Signal space diversity (SSD) is a well-known power- and bandwidth-efficient diversity technique with constellation rotation and coordinate interleaving. This paper investigates issues of combining SSD with coded modulation systems (CMSs), e.g., bit-interleaved coded modulation (BICM) and its iterative version, BICM-ID. It demonstrates that the average mutual information (AMI) between the signal after rotated constellation mapping and the signal before or after the soft demapper varies with the rotation angle. A new criterion for determining the optimal rotation angle by maximizing such AMI is therefore proposed. Specific considerations of combining SSD with BICM (BICM-SSD) and BICM-ID (BICM-ID-SSD) are addressed. The demappers extrinsic information transfer (EXIT) curve in traditional BICM-ID systems exhibits different slopes under different channels, which consequently prevents traditional BICM-ID systems from having simultaneously excellent performance under different channels. It is shown that such a different-slope problem can be simply mitigated by a proper use of SSD. Analysis and simulation show that the proposed BICM-ID-SSD systems hold a near-capacity performance under both additive white Gaussian noise (AWGN) and Rayleigh fading channels simultaneously.

High-Throughput QC-LDPC Decoders

High-throughput design approaches for quasi-cyclic (QC) low-density parity-check (LDPC) decoders are presented in this paper. Three novel schemes for the horizontal process in min-sum algorithm and its revisions are derived to reduce design and implementation complexity. The schemes can be directly applied for variant QC codes and easily pipelined to increase the operating frequency of the decoder. Some improvements of the semi-parallel architecture are proposed to enhance throughput performance and hardware efficiency. Employing the proposed approaches, QC-LDPC decoders for Chinese Digital Television Terrestrial Broadcasting (DTTB) standard are implemented using field programmable gate array (FPGA). As shown in the results, the proposed approaches can substantially improve the throughput performance, as well as the throughput-and-hardware tradeoff, of decoders with semi-parallel architecture.

Novel Decision-Aided Channel Estimation for TDS-OFDM Systems

Accurate channel estimation is essential for coherent signal transmission. This paper presents a novel channel estimation algorithm for time-domain synchronous orthogonal frequency division multiplexing (TDS-OFDM) systems with identical pseudo-noise (PN) sequence padding. Simple cyclic reconstruction and tail cancellation methods are proposed to estimate the channel impulse response (CIR). Phase equalization and partial hard decision on the received data symbols are used to improve the estimation performance. Simulation results show that the proposed channel estimation algorithm brings good BER performance (approaching the ideal case) with fast convergence (no iteration required) over frequency-selective slowly fading channels.

Bit-Interleaved LDPC-Coded Modulation with Iterative Demapping and Decoding

Bit-interleaved coded modulation (BICM) is a sub- optimal scheme from the average mutual information (AMI) point of view due to independent demapping. However, this AMI loss can be neglected for signal constellations with Gray mapping at high coding rates. AMI of amplitude-phase shift keying (APSK) constrained additive white Gaussian noise (AWGN) channel is provided in this paper, which shows that the BICM scheme has considerable loss with APSK constellations since no Gray mapping exists. Bit-interleaved low-density parity-check (LDPC) coded modulation (BILCM) is an excellent scheme and has been employed in many broadcasting and communication systems, however, such scheme also suffers from the independent demapping loss. Therefore, BILCM with iterative demapping and decoding (BILCM-ID) is proposed in this paper to overcome such loss. Simulation results show that the BILCM-ID scheme outweighs BILCM scheme by about 0.5 and 0.3 dB over AWGN channel for 32-APSK constellation at coding rates 2/3 and 4/5, respectively.

Robust Frame Synchronization for Chinese DTTB System

Digital terrestrial multimedia/television broadcasting (DTMB), which is the recently released Chinese digital television terrestrial broadcasting (DTTB) standard, uses guard intervals padded with known pseudo-noise (PN) sequences for multi-carrier or single-carrier block transmission. Well designed synchronization algorithms are a prerequisite for achieving high-quality multimedia transmission. In this paper, a novel frame synchronization technique, which is robust to carrier frequency offset (CFO), is proposed for DTMB systems. This technique is derived through segmenting the local PN sequence into M contiguous blocks and performing multiple cross-correlations with the received signal. The metric for peak detection is defined as the non-coherent summation of the M cross-correlation magnitude outputs. Moreover, a low-complexity frame synchronizer based on segment-based full correlation for PN420 mode in DTMB system is proposed. It is shown through computer simulation that the proposed techniques generally outperform the conventional ones when CFO occurs, over both AWGN and frequency-selective fading channels. For a DTMB system using PN420 mode, the maximal allowed CFO to keep system still working normally of our proposed schemes are more than 3 times larger than that of the conventional method.

Loop Delay Correction for Adaptive Digital Linearization of Power Amplifiers

As one of the most common linearization techniques, adaptive digital predistortion minimizes the difference between the input signal of the predistorter and the feedback signal from the power amplifier. The loop delay between these two signals must be controlled precisely so that the convergence of the predistorter should be assured. In this paper a novel two-stage loop delay correction scheme is presented. In the first stage, the integer delay between the input signal and the feedback signal is estimated based on the cross-correlation of them. In the second stage, the fractional delay between these two signals are corrected by a delay lock loop (DLL) circuit, in which a novel timing error detector used as the discriminator is proposed. The proposed loop delay correction scheme is applied to an adaptive digital predistorter in an orthogonal frequency-division multiplexing (OFDM) system. Simulation results show that the proposed loop delay correction method eliminates the loop delay perfectly, so that nearly optimal power spectral density (PSD) performance is achieved and a high degree of linearization is realized.