Milutin Pajovic

Design of a 1 Tb/s Superchannel Coherent Receiver

We describe the design of a trained and pilot-aided digital coherent receiver, capable of detecting a 1 Tb/s superchannel with a single optical front-end. Algorithms for receiver training are described, which calculate the equalizer coefficients, subchannel SNRs, and centroids of the transmitted constellations. Algorithms for pilot-aided operation are then described in detail, providing pilot-aided constant modulus equalization and joint carrier-phase estimation over several coherent subchannels. We demonstrate the detection of a superchannel with net bit rate in excess of 1 Tb/s with a single coherent receiver. An 11 × 10 GBd DP-64QAM Nyquist superchannel is used, with 1.32 Tb/s gross bit rate.

Performance Analysis and Optimal Design of Multichannel Equalizer for Underwater Acoustic Communications

Adaptive equalization is a widely used method of mitigating the effects of multipath propagation and Doppler spreading in underwater acoustic communication channels. While the structure of a multichannel equalizer and least-squares-based adaptation algorithm are extensively used in practice, little is known in how to choose the number of sensors, separation between them, and lengths of the constituent filters such that the equalization performance is optimized. This paper studies the problem of optimal multichannel equalizer design in the context of time-varying underwater acoustic communication channels. In the first part, the paper presents a theoretical characterization of the equalization performance when the number of symbols that can be received in the time period over which the channel can be considered time invariant is limited. This result is then used to develop an understanding that the optimal number of equalizer coefficients is a tradeoff between the minimum mean squared error (MMSE) requirement for longer constituent filters and the insight that the limit on the number of stationary observations also limits the number of filter coefficients that can be effectively adapted. In the second part, the paper develops a theoretical model for wideband arrivals impinging upon an array of sensors of the multichannel equalizer. This model is used to develop an understanding that the optimal sensor separation is a tradeoff between the requirement for long aperture which improves resolution, and the fact that the grating lobes, caused by spatial undersampling, limit the equalizers ability to estimate the transmitted signal from the received signal.

Millimeter wave communications channel estimation via Bayesian group sparse recovery

We consider the problem of channel estimation for millimeter wave communications (mmWave). We formulate channel estimation as a structured sparse signal recovery problem, in which the signal structure is governed by a priori knowledge of the channel characteristics. We develop a Bayesian group sparse recovery algorithm which takes into account for several features unique to mmWave channels, such as spatial (angular) spreads of received signals and power profile of rays impinging on the receiver array. We validate the developed method via numerical simulations and demonstrate an improved estimation performance relative to the existing methods.

Experimental demonstration of multi-pilot aided carrier phase estimation for DP-64QAM and DP-256QAM

We present a statistical inference based multi-pilot aided CPE algorithm and analyze its performance via simulations. We experimentally verify LDPC coded back-to-back performance using 10 GBd DP-64QAM and DP-256QAM modulation, with transmitter and receiver linewidths of 100 kHz.

Battery State-of-Charge Estimation Based on Regular/Recurrent Gaussian Process Regression

This paper presents novel machine-learning-based methods for estimating the state of charge (SoC) of lithium-ion batteries, which use the Gaussian process regression (GPR) framework. The measured battery parameters, such as voltage, current, and temperature, are used as inputs for regular GPR, whereas the SoC estimate at the previous sample is fed back and incorporated into the input vector for recurrent GPR. The proposed methods consist of two parts. In the first part, training is performed wherein the optimal hyperparameters of a chosen kernel function are determined to model data properties. In the second part, online SoC estimation is carried out according to the trained model. One of the practical advantages of a GPR framework is to quantify estimation uncertainty and, hence, to enable reliability assessment of the battery SoC estimate. The performance of the proposed methods is evaluated by using a simulated dataset and two experimental datasets, one with constant and the other with dynamic charge and discharge currents. The simulations and experimental results show the superiority of the proposed methods in comparison to state-of-the-art techniques including a support vector machine, a relevance vector machine, and a neural network.

Online battery state-of-charge estimation based on sparse gaussian process regression

This paper presents a new online method for state-of charge (SoC) estimation of Lithium-ion (Li-ion) batteries based on sparse Gaussian process regression (GPR). Building upon sparse approximation of the regular GPR, the proposed method is computationally more efficient. The battery SoC is estimated based on measured voltage, current and temperature. The accuracy of the proposed method is verified using LiMn2O4/hard-carbon battery data collected from a constant-current discharge test. In addition, the estimation performance of the proposed method is compared with a SoC estimation method using regular GPR with different covariance functions.

Online state of charge estimation for Lithium-ion batteries using Gaussian process regression

This paper presents an application of Gaussian process regression (GPR) to estimate a state of charge (SoC) of Lithium-ion (Li-ion) batteries with different kernel functions. One of the practical advantages of using GPR is that uncertainties in the estimates can be quantified, which enables reliability assessment of the SoC estimate. The inputs of GPR are voltage, current and temperature measurements of the battery and the output is an estimate of SoC. First, training is performed in which optimal hyperparameters of a kernel function are determined to model data properties. Then, the battery SoC is estimated online based on the trained model. The kernel function is the key element in the GPR model since it encodes the prior assumptions about the properties of the function being modeled. Therefore, the impact of kernel function selection on the estimation performance is analyzed using both simulated data and experimental data collected from a LiMn2O4/hard-carbon battery with a nominal capacity of 4.93Ah operating under constant charge and discharge currents.

Detection of a 1 Tb/s superchannel with a single coherent receiver

We demonstrate detection of a superchannel with net bit rate in excess of 1 Tb/s with a single coherent receiver. Novel, pilot-aided equalization and carrier recovery algorithms enable detection of an 11 × 10 GBd DP-64QAM Nyquist superchannel (1.32 Tb/s gross bit rate).

Sparse channel estimation in millimeter wave communications: Exploiting joint AoD-AoA angular spread

In this paper, channel estimation in millimeter wave (mmWave) communication systems is considered. In contrast to prevailing mmWave channel estimation methods exploiting the sparsity nature of the channel, we move one step further by exploiting the joint AoD-AoA angular spread. By formulating the channel estimation as a block-sparse signal recovery with an underlying two-dimensional cluster feature, we propose a two-dimensional sparse Bayesian learning method without a priori knowledge of two-dimensional angular spread patterns. It essentially couples the channel path power at one angular direction with its two-dimensional AoD-AoA neighboring directions. Compared with existing sparse mmWave channel estimation methods, the proposed method is numerically verified to reduce the training overhead and channel estimation error.

Spatial Scattering Modulation for Uplink Millimeter-Wave Systems

In this letter, a new spatial scattering modulation (SSM) is proposed for uplink millimeter-wave (mmWave) systems that support a single user terminal (UT). By utilizing the analog and hybrid beamforming with a large antenna array and phase shifters for mmWave communications systems, an architecture, where the UT has a single radio-frequency (RF) chain, whereas the base station has more than one RF chain, is adopted. In this architecture, the proposed SSM modulates some information bits on the spatial directions of scattering clusters in the angular domain, so that a higher spectral efficiency can be achieved with the use of a lower order modulation. For a particular number of scattering clusters and a number of RF chains, a closed-form expression for the upper bound on the bit-error rate (BER) is derived for the proposed SSM. Monte Carlo simulations are also conducted to verify the achievable BER performance.