Research Interests:
Research Interests:
Deep Learning has been widely applied in the area of image processing and natural language processing. In this paper, we propose an end-to-end communication structure based on autoencoder where the transceiver can be optimized jointly. A... more
Deep Learning has been widely applied in the area of image processing and natural language processing. In this paper, we propose an end-to-end communication structure based on autoencoder where the transceiver can be optimized jointly. A neural network roles as a combination of channel encoder and modulator. In order to deal with input sequences parallelly, we introduce block scheme, which means that the autoencoder divides the input sequence into a series of blocks. Each block contains fixed number of bits for encoding and modulating operation. Through training, the proposed system is able to produce the modulated constellation diagram of each block. The simulation results show that our autoencoder performs better than other autoencoder-based systems under additive Gaussian white noise (AWGN) and fading channels. We also prove that the bit error rate (BER) of proposed system can achieve an acceptable range with increasing the number of symbols.
Research Interests:
Research Interests:
Research Interests:
Research Interests:
ABSTRACT In this paper, we focus on the interference alignment precoder design for cellular system. Varying from the traditional constrained optimization method, we reformulate the optimization problem on the complex Grassmann manifold... more
ABSTRACT In this paper, we focus on the interference alignment precoder design for cellular system. Varying from the traditional constrained optimization method, we reformulate the optimization problem on the complex Grassmann manifold and derive a novel steepest descent algorithm to achieve perfect interference alignment. Moreover our proposed algorithm only requires the participation of transmitter. Thus it will alleviate the overhead induced by alternating between the up and down links significantly. Simulation results suggest the proposed algorithm has better convergence performance and higher system capacity comparing with previous methods.
Research Interests:
ABSTRACT Impulse radio ultra wideband (IR-UWB) technique has attracted interest in indoor localization thanks to its sub-nanosecond (ns) narrow pulse feature offering high timing resolution. However, measuring these short pulses demands... more
ABSTRACT Impulse radio ultra wideband (IR-UWB) technique has attracted interest in indoor localization thanks to its sub-nanosecond (ns) narrow pulse feature offering high timing resolution. However, measuring these short pulses demands high performance analog-to-digital converter (ADC), i.e. 4 giga Hz 8-bit ADC, which is hard to implement in real systems. In this paper, a finite-resolution (FR) quantization based localization method is proposed, revealing that high performance ADCs can be replaced with high speed comparators since a 2-bit quantization is good enough. At first we approximate the post-quantization signal as Gaussian distributed using Bussgang theorem, facilitating the following derivation. Then, practical TOA estimation and iterative Taylor-series (TS) localization algorithm are derived. Subsequently the analytical expressions of Cramer-Rao lower bound (CRLB) of proposed scheme is obtained, theoretically quantifying the performance loss caused by FR quantization. Finally, compared with the full-resolution and signal-strength (SS) approaches via simulation, we prove that our finite-resolution algorithm achieves almost the same performance as traditional full-resolution scheme while dramatically decreasing the complexity.
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
ABSTRACT Automatic repeat-request (ARQ) protocols for two-user cooperative diversity system employing Alamouti space-time coding are investigated in this paper. A whole cooperative transmit frame of the original cooperative system... more
ABSTRACT Automatic repeat-request (ARQ) protocols for two-user cooperative diversity system employing Alamouti space-time coding are investigated in this paper. A whole cooperative transmit frame of the original cooperative system consists of three sub-frames. According to different feedback schedules in a transmit frame at the destination, two basic ARQ protocols, namely post-cooperating and pre-cooperating, are proposed and analyzed. We show that both proposed ARQ protocols can improve the throughput in two-user cooperative system. Pre-cooperating protocol yields 50% gains in users' throughput than post-cooperating protocol if both users have high user-destination average received signal-to-noise ratio (SNR) at the destination. Nevertheless, if both of user-destination average received SNRs are low, post-cooperating protocol is more effective than pre-cooperating protocol. Proposed protocols provide useful basic tools for designing more complicated ARQ protocols. Finally, simulation results verify our analysis.
Research Interests:
ABSTRACT Sparse equalizers which have fewer nonzero coefficients can significantly reduce the cost of the implementation and computational complexity. The performance of the sparse equalizers heavily depends on the tap selection... more
ABSTRACT Sparse equalizers which have fewer nonzero coefficients can significantly reduce the cost of the implementation and computational complexity. The performance of the sparse equalizers heavily depends on the tap selection algorithms. In this paper, a low complexity algorithm based on the forward greedy selection is proposed. The objective of the sparse equalizer design is to reduce the number of nonzero equalizer coefficients under a quadratic constraint with a given tolerance of the performance loss. Furthermore, the performance of the proposed algorithm is analyzed and we obtain the terminal condition for the iteration number of the forward selection algorithm. Simulation results show that the proposed algorithm can reach the same performance or outperform the conventional sparse equalizer schemes, offering a low computational complexity at the same time.
Research Interests:
ABSTRACT High-speed high-resolution analog-to-digital converter (ADC) is a key bottleneck in large-bandwidth systems such as 60 GHz communication of wireless personal area networks (WPANs), due to its large power consumption and high... more
ABSTRACT High-speed high-resolution analog-to-digital converter (ADC) is a key bottleneck in large-bandwidth systems such as 60 GHz communication of wireless personal area networks (WPANs), due to its large power consumption and high complexity. Hence, monobit ADC has been previously proposed to address this problem. In this paper, we propose a receiver architecture with monobit sampling and carrier phase rotation for quadrature-phase-shift-keying (QPSK) modulation. The optimal rotation phase is derived from the achievable-rate point of view. Then, a suboptimal but low-complexity monobit receiver is obtained, and the impact of different rotation phases is investigated. The optimal rotation phase is found to be π/4. Simulation results show that there is only about 1dB signal-to-noise ratio (SNR) loss with a rotation phase deviation of 15° from the optimal value, providing a pretty good robustness to implementation imperfection.
Research Interests:
Research Interests:
ABSTRACT In this paper, the effect of finite-level quantization on UWB time-of-arrival (TOA) estimation is investigated. The scheme of optimized quantization threshold combined with the post-quantization processing is derived, which is... more
ABSTRACT In this paper, the effect of finite-level quantization on UWB time-of-arrival (TOA) estimation is investigated. The scheme of optimized quantization threshold combined with the post-quantization processing is derived, which is shown to provide satisfactory gains in the system performance. The TOA estimation errors of several low-resolution sampling approaches are compared via Monte Carlo simulation, where the tri-level quantizer is of particular interest due to its simplicity and capability. We demonstrate that the tri-level sampling receiver, with use of the proposed scheme provides an outstanding performance in TOA estimation with an affordable cost and low complexity.
Research Interests:
ABSTRACT Future communication system requires large bandwidths to achieve high data rates, thus rendering analog-to-digital conversion (ADC) a bottleneck due to its high power consumption. In this paper, we consider monobit receivers for... more
ABSTRACT Future communication system requires large bandwidths to achieve high data rates, thus rendering analog-to-digital conversion (ADC) a bottleneck due to its high power consumption. In this paper, we consider monobit receivers for QPSK. The optimal monobit receiver under Nyquist sampling is obtained and its performance is analyzed. Then, a suboptimal but low-complexity receiver is proposed. The effect of imbalances between In-phase (I) and Quadrature (Q) branches is carefully examined. To combat the performance loss due to IQ imbalances, monobit receivers based on double training sequences and eight-sector phase quantization are proposed. Numerical simulations show that the low-complexity suboptimal receiver suffers 3dB signal-to-noise-ratio (SNR) loss in additive white Gaussian noise (AWGN) channels and only 1dB SNR loss in multipath channels compared with matched-filter monobit receiver with perfect channel state information (CSI). It is further demonstrated that the amplitude imbalance has essentially no effect on monobit receivers. In AWGN channels, receivers based on double training sequences can efficiently compensate for the SNR loss without complexity increase, while receivers with eight-sector phase quantization can almost completely eliminate the SNR loss caused by IQ imbalances. In dense multipath channels, the effect of imbalances on monobit receivers is slight.
Research Interests:
ABSTRACT In this paper, the effect of finite-resolution analog-to-digital converter (ADC) quantization on ultra-wideband (UWB) time of arrival (TOA) estimation is investigated. The deflection criterion is introduced for optimizing the... more
ABSTRACT In this paper, the effect of finite-resolution analog-to-digital converter (ADC) quantization on ultra-wideband (UWB) time of arrival (TOA) estimation is investigated. The deflection criterion is introduced for optimizing the nonuniform quantization. The training based TOA estimation algorithm using maximum likelihood (ML) rule is proposed. Compared with the full-resolution and energy-detection (ED) estimators via simulation, we demonstrate that our training based 3-level quantization scheme is satisfactory because it can approach a full-resolution estimator while decreasing the complexity greatly.