We study a classic problem in wireless data communications in which receivers generally rely on training pilots for equalization of multi-path dispersive channel distortions. Traditional equalizers often require a substantial number of pilot symbols for effective channel compensation. To conserve limited channel bandwidth, we investigate novel equalization algorithms that can make more efficient use of fewer pilot symbols. We present a linear programming algorithm that takes better advantage of the QAM constellation. We further exploit the information inherent in the LDPC forward error correction codes within the linear programming framework to improve the equalization performance and convergence properties.

In this paper, we consider a two-tier heterogeneous network that locally consists of a multi-antenna macrocell base station and a multi-antenna femtocell base station (FBS) each serving separate single-antenna users. We investigate an optimal transmission strategy with maximum degree of freedom for transmit power minimization of the FBS under outage-based quality-of-service (QoS) constraints for the femtocell user equipment (FUE) and macrocell user equipment (MUE). Specifically, we examine the scenario that the FBS receives no instantaneous channel estimate from the FUE, and relies on only statistical information of downlink multiple-input single-output (MISO) channels. Although the outage constrained problem has no closed-form probabilistic constraints and may not be convex in general, we propose a transmission strategy and prove its optimality under a given condition. Our simulation results also demonstrate that the proposed transmission strategy can significantly save power compared to beamforming strategies.

Conventional approaches to linear equalization of data transmissions under frequency selective channels simply treat the problem as a system identification issue. However, practical data communication systems impose additional conditions on the channel input signals such as LDPC error correction coding and known data constellation. We study a unified framework to integrate these known constraints into the channel equalization procedure and propose a linear programming-based approach to channel equalization for QAM signals that can succeed with significantly reduced pilots. In addition to utilizing the information of QAM constellation, our method exploits the code information of LDPC forward error correction coding to improve the equalization performance and convergence properties.

To improve wireless heterogeneous network service via macrocell and femtocells that share certain spectral resources, this paper studies the transmit beamforming design for femtocell base station (FBS), equipped with multiple antennas, under an outage-based quality-of-service (QoS) constraint at the single-antenna femtocell user equipment characterized by its signal-to-interference-plus-noise ratio. Specifically, we focus on the practical case of imperfect downlink multiple-input single-output (MISO) channel state information (CSI) at the FBS due to limited CSI feedback or CSI estimation errors. By characterizing the CSI uncertainty probabilistically, we formulate an outage-based robust beamforming design. This nonconvex optimization problem can be relaxed into a convex semidefinite programming problem, which reduces to a power control problem when all CSI vectors are independent and identically distributed. We also investigate the performance gap between the optimal transmission strategy (that allows maximum transmission degrees of freedom (DoF) equal to the number of transmit antennas) and the proposed optimal beamforming design (with the DoF equal to one) and provide some feasibility conditions, followed by their performance evaluation and trade-off through simulation results.

This work studies the reconstruction of gene regulatory networks by the means of network component analysis (NCA). We will expound a family of convex optimization-based methods for estimating the transcription factor control strengths and the transcription factor activities (TFAs). The approach taken in this work is to decompose the problem into a network connectivity strength estimation phase and a transcription factor activity estimation phase. In the control strength estimation phase, we formulate a new subspace-based method incorporating a choice of multiple error metrics. For the source estimation phase we propose a total least squares (TLS) formulation that generalizes many existing methods. Both estimation procedures are noniterative and yield the optimal estimates according to various proposed error metrics. We test the performance of the proposed algorithms on simulated data and experimental gene expression data for the yeast Saccharomyces cerevisiae and demonstrate that the proposed algorithms have superior effectiveness in comparison with both Bayesian Decomposition (BD) and our previous FastNCA approach, while the computational complexity is still orders of magnitude less than BD.

Practical transmission of orthogonal frequency division multiplexing (OFDM) signals suffers from high peak-to- average power ratio (PAPR) at the transmit antenna(s). Among various solutions, tone injection is an effective approach to mitigate this problem without incurring bandwidth loss. However, due to its computational complexity, finding the optimal tone injection becomes intractable for OFDM systems with a large number of subcarriers. In this work we present a novel convex optimization approach to numerically determine the near-optimal tone injection solution based on a semi-definite relaxation. This approach leads to significant PAPR reduction with polynomial complexity. An STBC-OFDM extension is also proposed.