Ahmed Masmoudi

In this paper, we investigate hybrid precoders for massive MIMO systems with the objective to reduce the number of radio-frequency (RF) chains. The users are first partitioned into groups that share similar angles of departure (AoDs) to design an angular-based RF-beamforming stage. Then a multi-user-Multiple Input Multiple Output (MIMO) baseband precoding stage is developed with reduced channel state information (CSI) dimension in massive MIMO systems. We develop a systematic approach to simultaneously determine the number of groups and to cluster the users accordingly. The partition is done by minimizing the inter-group interference using a similarity graph. We also incorporate a simple transfer block between the digital precoding and RF-beamforming stages to reduce the number of RF chains in massive MIMO systems. For a sufficiently large number of antennas and over a wide range of signal-to-noise ratio (SNR), the proposed hybrid RF/baseband precoding schemes can offer a sum-rate performance approaching that of the single-stage full digital optimal precoder, while keeping the number of RF chains equal to the number of users. Next, we propose a simple way to design the RF and baseband digital precoding stages in the case of multiple scatterings to fully exploit the multiple reflections that occur in dense urban environments. The proposed hybrid 2-stage RF/baseband schemes significantly outperform other hybrid RF/baseband precoding schemes using orthogonal multiplexing.

This paper proposes a new user grouping algorithm and three-dimensional (3D) angular-based hybrid precoding (AB-HP) scheme for massive multi-user multiple-input multiple-output (MU-MIMO) systems using uniform rectangular arrays (URA). At first, the users clustered in multiple spots are efficiently grouped according to the proposed user grouping algorithm, which only utilizes the user angle-of-departure (AoD) information and does not require prior knowledge of the number of user groups. By employing the AoD support of the user groups, the RF-beamformer of AB-HP is designed to reduce the inter-group interference, the channel state information (CSI) overhead, and the number of RF chains. Then, the digital baseband precoder of AB-HP is constructed via regularized zero-forcing (RZF) using the effective channel seen from baseband to simultaneously serve the users clustered in multiple groups, by considering three approaches: joint-group-processing (JGP), per-group-processing (PGP) and common-group-processing (CGP). For each approach, the signal-to-interference-plus-noise ratio (SINR) expressions as well as their tight deterministic approximations are derived. To further reduce the number of RF chains, we also propose a new transfer block design, which reduces the number of RF chains down to the number of independent data streams without penalizing the sum-rate performance. Illustrative results reveal that the proposed AB-HP schemes with the relaxed CSI estimation overhead and reduced hardware cost/complexity can closely approach to the sum-rate performance of the single-stage fully-digital precoding (FDP). myGreenFurthermore, AB-HP has considerably higher energy efficiency performance compared to FDP due to the reduced number of RF chains. We show through simulation that the proposed AB-HP can offer significantly better performance than existing HP techniques. The computational complexity of AB-HP is also analyzed.

This paper addresses the self-interference (SI) cancellation for full-duplex (FD) operation in the presence of imperfect radio-frequency (RF) components. In particular, we develop a new scheme to jointly estimate and cancel the IQ mixer imbalance, power amplifier (PA) nonlinearities, up-/down-conversion phase-noise and the SI-channel. First, we develop a detailed baseband model that captures the most significant transceiver RF imperfections, for both separate- and common-oscillator structures used in the up- and down-conversions. Then, a basis expansion model (BEM) is derived to approximate the time-varying phase-noise and to transform the problem of estimating the time-varying phase-noise into the estimation of a set of time-invariant coefficients. Subsequently, the likelihood function is derived in the presence of the unknown intended signal to formulate the joint estimation of the intended channel, SI-channel, nonlinear impairments and phase-noise, under the maximum likelihood (ML) criterion. An iterative procedure is developed to find the ML estimate of the different parameters based on the known transmitted data, the known pilot symbols, and the statistics of the unknown intended signal received from the intended transmitter. The full use of the received signal significantly reduces the required number of pilot symbols as compared to training-based techniques. We consider the two pilot-insertion structures used in LTE for the frequency-multiplexed pilots and the time-multiplexed pilots. Simulation results indicate that the proposed ML algorithms can offer a superior SI-cancellation performance with the resulting intended-signal-to-SI-and-noise ratio (SINR) very close to the intended-signal-to-noise ratio (SNR).

In this paper, we address the problem of time delay estimation (TDE) from single-carrier (SC) or multi-carrier (MC) direct-sequence spread spectrum (DSSS) multipath transmissions in the presence of multiple transmit and/or receive antennas that will characterize future 5G radio interface technologies (RITs), such as coded-domain nonorthogonal multiple access. We derive for the first time a closed-form expression for the Cramer-Rao lower bound (CRLB) and develop two maximum likelihood (ML) multipath TDEs for SC DSSS single-input multiple-output (SIMO) in the non-data-aided (NDA) case. The first TDE, based on iterative expectation maximization (EM), provides accurate estimates whenever a good initial guess of the parameters is available at the receiver. The second TDE implements the ML criterion in a non-iterative way and finds the global maximum of the compressed likelihood function using the importance sampling (IS) technique without requiring any initialization. We also extend both the SC DSSS SIMO CRLB and the two new SC DSSS SIMO ML NDA TDEs to MC DSSS RITs and to multiple-input multiple-output structures with any diversity versus multiplexing pre-coding type before generalizing them all to the data-aided (DA) case. Simulations suggest that the EM TDE is suitable for large observation in space, time, and/or frequency, whereas the IS TDE is preferred in the opposite case of very short data records. Moreover, we show in the NDA case, both analytically and by simulations, that spatial (transmit and receive), temporal, and frequency samples interchangeably have the same impact on estimation accuracy and performance bound regardless of the channel correlation type and amount present in each dimension. Furthermore, we are able to properly cope with such channel correlations that do indeed arise in practice and, hence, become very challenging both in estimation and CRLB derivation in the DA case, but that have been so far overlooked in previous works.

This paper addresses the self-interference (SI) cancellation at baseband for full-duplex MIMO communication systems in consideration of practical transmitter imperfections. In particular, we develop a subspace-based algorithm to jointly estimate the SI and intended channels and the nonlinear distortions. By exploiting the covariance and pseudo-covariance of the received signal, we can increase the dimension of the received signal subspace while keeping the dimension of the signal subspace constant, and hence, the proposed algorithm can be applied to most of full-duplex MIMO configurations with arbitrary numbers of transmit and receive antennas. The channel coefficients are estimated, up to an ambiguity term, without any knowledge of the intended signal. A joint detection and ambiguity identification scheme is proposed. Simulation results show that the proposed algorithm can properly estimate the channel with only one pilot symbol and offers superior SI cancellation performance.

This paper presents two self-interference (SI) cancellation methods using active signal injection (ASI) for full-duplex MIMO-OFDM systems. Specifically, the ASI approach considers adding an appropriate cancelling signal to the transmitted signal to reduce the SI at the receiver input to avoid overloading the receiver low-noise amplifier (LNA) and analog-to-digital converter (ADC) while ensuring proper signal detection. In the first method, the cancelling signal uses some reserved subcarriers that are not used for data transmission. In the second method, the constellation points are dynamically extended within the cancellation boundary in order to minimize the SI. The proposed techniques appear to be simple to implement and do not require any change on the receiver structure.

Full-duplex operation requires effective self-interference (SI) cancellation that in turn needs reliable SI channel estimation. In this paper, we develop two estimation algorithms suitable for a 2-stage SI cancellation structure. By exploiting the sparsity of the SI channel, we first derive a compressed sensing-based SI channel estimation algorithm to be used in the first SI cancellation stage at radio-frequency (RF) to reduce the SI. We then develop a subspace-based algorithm to jointly estimate the residual SI channel, the intended channel and the transmitter nonlinearities for the second SI cancellation stage at baseband. Including the intended received signal in the estimation process is the main advantage of the proposed algorithm as compared to previous works that assume it as additive noise. Simulation results show that the proposed algorithms outperform the least-square (LS) algorithm and offer higher signal-to-residual-interference-and-noise ratio (SINR) over a large received signal-to-noise ratio (SNR) range.

This paper addresses the problem of time delay estimation of minimum shift keying (MSK)-modulated signals. We derive for the first time the analytical expression of the stochastic Cramer-Rao lower bound (CRLB) for non-data-aided (NDA) timing recovery, from which the effects of the shaping pulse are easily observed. It is assumed that the transmitted symbols are completely unknown at the receiver. Moreover, the carrier phase and frequency are considered as unknown nuisance parameters.We emphasize the penalty resulting from considering the transmitted data as unknown in the estimation of the time delay.