Minglu Jin

Denoising Detection for the Generalized Spatial Modulation System Using Sparse Property

Generalized spatial modulation (GSM) is a novel scheme developed from the conventional single-active antenna spatial modulation (SA-SM). The challenge of the GSM lies at the receiving end. Since more than one antenna is activated, the complexity of ML detection for GSM is much higher than that for SA-SM, especially if the number of antennas is large. Low complexity sub-optimal detections, such as ZF detection or MMSE detection, can be used to detect GSM. However, the performance of these detections is poor and cannot yet be applied in an underdetermined system. Spatial modulation has an inherent property of sparsity. However, to the best of our knowledge, this property has not been used for GSM. In this paper, we exploit this property and propose a sub-optimal detection algorithm. The traditional algorithm using sparsity is sensitive to noise and not suitable for the application to communication systems. Therefore, we employ a denoising scheme to counter the effect of the background noise.

Further Complexity Reduction Using Rotational Symmetry for EDAS in Spatial Modulation

Euclidean distance antenna selection (EDAS) is an effective transmit antenna selection technique for spatial modulation (SM). However, since an exhaustive search over both the antenna set and the set of conventional constellation points is executed, the computational complexity required to find the optimal solution of EDAS is too high for practical applications. In general, digital modulation constellations have rotational symmetry. In this letter, it is shown that the symmetric constellation points have the same value as EDAS. Hence, only one search is enough for the symmetrical constellation set. Using this proposition, we propose a fast algorithm to find the optimal EDAS solution, which exhibits lower computational complexity than the conventional EDAS solution.

Optimal Transceiver Design for Interference Alignment Based Cognitive Radio Networks

Interference alignment (IA) can effectively eliminate the interference among users in cognitive radio (CR) networks. Nevertheless, to further increase the transmission rate of the primary user (PU) and guarantee its priority, we propose an optimal transceiver design (OTD) scheme for IA-based CR networks in this letter. In the OTD scheme, two partial-IA based algorithms are proposed to design the precoding matrices of SUs, in which the interference is aligned at the primary receiver with much lower overhead and complexity. Then the PUs precoding matrix is designed to maximize its throughput accordingly. To further improve the performance of SUs, their decoding matrices are re-designed. Simulation results are presented to show the effectiveness of the proposed scheme.

A Modified LMS Algorithm with Turbo-Like Scheme

LMS algorithm has low computation complexity; however, its convergence rate is slow. In this paper, a novel implementation scheme for LMS algorithm is proposed to form a modified LMS algorithm. The scheme is similar to the structure of turbo decoder, which consists of one feedback module called as converse-speediness module, two LMS algorithm modules, and one speediness module which connects two LMS algorithm modules. Meanwhile, two LMS algorithm modules can be implemented parallel, as well as the speediness module and converse-speediness module. Simulation results show that the scheme makes the convergence rate of the proposed algorithm over 4 times faster than that of LMS algorithm, and almost the same as that of RLS algorithm. Moreover, compared to LMS algorithm and RLS algorithm, the mean square error (MSE) of the proposed algorithm decreases greatly.

FQAM/FPSK modulation for spatial modulation systems

Spatial Modulation (SM) is a recently developed transmit technique, and it has attracted wide interests for it gains higher transmit rate than single-input-single-output (SISO) system and provides simpler transceiver structure and lower power consumption compared with traditional MIMO system. In this paper, we propose the spatial modulation systems with frequency and quadrature amplitude modulation (FQAM)/frequency and phase shift keying (FPSK), which combine frequency shift keying (FSK) and quadrature amplitude modulation (QAM)/phase shift keying (PSK). Simulation results show that the proposed scheme (i.e. FQAM/FPSK-SM) achieves significantly better bit error performance (BER) than traditional SM system and FQAM/FPSK system using maximum likelihood (ML) detector, and it can assign information flexibly among the constellation modulation, the index of the active antenna and the index of the active frequency. We also compare and analyze system performance and complexity of optimal detector and sub-optimal detector.

Penalty Function Based Detector for Generalized Space Shift Keying Massive MIMO Systems

Generalized space shift keying (GSSK) has recently attracted much interest in the study of emerging massive MIMO systems. The maximum likelihood (ML) detection of GSSK can be posed as a 0-1 quadratic program with an equality constraint. The penalty function method is a common way to transform a constrained programming into an unconstrained one. However, determining an ideal penalty factor is not a trivial problem and has not been addressed adequately. In this letter, we prove a lemma to show that the ML detection of GSSK can be converted into an equivalent 0-1 quadratic program if the penalty factor is greater than a small constant. Based on the proposed lemma, we also present an algorithm to determine the penalty factor and finally recover the transmitted signals of GSSK. Simulation results substantiate the performance of the proposed detector.

Detection of Generalized Space Shift Keying Signal With Sparse Reconstruction

In this paper, we propose a novel generalized space shift keying (GSSK) detection algorithm by exploiting the inherent sparse property of GSSK signal. In particular, we formulate the GSSK detection into a sparse convex optimization problem. The key contribution of the proposed algorithm lies in the strategical adoption and transformation of the sparse reconstruction (SR) from image processing to GSSK detection. The proposed SR detector achieves better performance than all existing suboptimal solutions with comparable computational complexity. In order to further improve the performance of the SR algorithm, we develop another approach, i.e., the iterative SR algorithm, which utilizes the result of the SR algorithm as the start point and enhances the detection performance gradually in each iteration. The simulation results confirm the efficiency of both of the proposed algorithms.

Lagrangian Detection for Generalized Space-Shift Keying MIMO Systems

Generalized space-shift keying (GSSK) has recently established itself as a promising technology for massive multiple-input multiple-output (MIMO) systems. However, the computational complexity of maximum likelihood (ML) detection is too high, and it increases significantly as the number of transmit antennas and active antennas increases. In this correspondence, we propose a low-complexity suboptimal detection for massive GSSK-MIMO systems. The ML detection of GSSK can be posed as a 0–1 quadratic programming with an equality constraint. First, we employ the Lagrange multiplier to transform the 0–1 quadratic programming with a linear equality constraint into a standard 0–1 quadratic programming. Most of the conventional methods for determining the Lagrange multiplier are derived from Karush–Kuhn–Tucker (KKT) conditions, which are usually valid for continuous variable programming rather than the discrete one. However, in our problem, the optimization variables are binary. Therefore, we propose a theorem that can determine the Lagrange multiplier iteratively by an 1-D binary search rather than KKT conditions and, finally, detect the GSSK transmission symbols. Simulation results demonstrate that the proposed method can achieve an excellent signal detection performance for massive GSSK-MIMO systems with low computational complexity.