Search

In this paper, a fast pursuit algorithm based on a Bayesian approach is proposed for pruning behavioral models of power amplifiers (PAs). This approach, referred to as sparse Bayesian pursuit (SBP), is compared to two previously reported pruning algorithms: orthogonal matching pursuit (OMP) and doubly OMP (DOMP). The three approaches are benchmarked with experimental data provided by a class AB PA driven by a 30-MHz 5G New Radio signal. The new sparse Bayesian pursuit approach outperforms the OMP in both modeling and linearization capabilities, while it reduces the computational cost with respect to the DOMP algorithm.

A bivariate Volterra series perspective is adopted for a generic power amplifier (PA) modeling. This approach is motivated by the presence of a PA mechanism responsible of an internal signal generation. The bivariate resulting model extends a conventional Volterra model by including terms containing cross products of the input signal and the new internal variable. The approach is developed considering a simple univariate Volterra model and the signal envelope as the internal variable. The enriched structure incorporates new terms with the envelope raised to odd powers. A particular case of this bivariate Volterra series approach is the generalized memory polynomial model. The convenience of the bivariate terms in the model structure of a class J PA is experimentally demonstrated for the amplifier driven by a 30-MHz 5G New Radio signal, over a range of more than 13 dB output power levels. Results for the digital predistortion of the class J PA are also provided, illustrating as well the benefits of the bivariate model for the linearization performance.

This Innovative Practice Work-In-Progress presents the application of low-cost technology for the creation of practical demonstrations in the validation of theoretical lessons concepts. The creation of new activities is based on the fact that equipment related to laboratory sessions is costly and therefore difficult to access for the students. With this methodology, we expect an increase in the motivation and satisfaction of the students as well as a tool for themselves to experiment individually. Hardware as the well-known Raspberry Pi working with devices such as a software-defined radio (SDR) dongle and a sensor board were used in the experiments. Preliminary results show the students acceptance, becoming a promising technique in undergraduate-level courses.

This work presents the application of the Random Forest technique to behavioral modeling component selection. Several digital predistorters (DPDs) with a random structure are attained to obtain a sequence of the regressors importance with respect to their impact in the linearization error of a power amplifier (PA). The approach has been validated in the DPD of a commercial PA operating under a 5G-NR signal, showcasing the ability of the algorithm to sort and prune the components.

This work presents a novel power amplifier behavioral model, referred to as power dependent Volterra (PDV) model, which is devoted to keep track on the average power of the input signal. The behavior of the parameters under varying power conditions is theoretically investigated using circuit-level knowledge including electrical and thermal subcircuits, and the dependence on average power is demonstrated. Experimental results employing a GaN HEMT power amplifier show that the model performance at $\mathrm{P}_{\mathrm{o}}=+33.8\mathrm{dBm}$ is maintained over a range of about 15 dB.

This work presents a simplified version of the doubly orthogonal matching pursuit (DOMP) algorithm to estimate the coefficients of a power amplifier (PA) Volterra behavioral model. The kernels are sequentially selected from the complete model, which requires the pseudoinversion of the measurement matrix after adding a new column. The proposed enhancement avoids the matrix inversions, reducing the computation complexity.

This communication presents a new technique for the digital predistortion of power amplifiers (PAs) based on sparse behavioral models. The subspace pursuit algorithm formulation is adapted to work in the nonlinear series framework. Experiments driven on a test bench based on a GaN PA driven by a 15-MHz filter bank multicarrier (FBM C) signal were conducted in order to validate the algorithm. Experimental results in a digital predistortion scenario and the comparison with the orthogonal matching pursuit highlight the enhancement of this pruning method.

This paper describes a new educational robot called ROBOBO which was developed in order to provide an engaging and natural interactive experience with children. The objective is to motivate children to explore and cooperate with the robots while they learn concepts and abilities related to STEM education (Science, Technology, Engineering and Mathematics). The robot and the educational approach has been tested in different school environments corresponding to different educational levels with very successful results.

This paper reports the identification of nonlinear models for wireless communications systems. The procedure relies on a novel complex-valued Volterra series (CVS) representation to provide a sparse representation based on statistical hypothesis testing and the Bayesian information criterion (BIC). The approach has been experimentally evaluated with the front-end of a communications transmitter taking into account the possible nonlinear impairments associated with the I/Q modulator. The proposed identification method showed an excellent performance in terms of model complexity reduction and dynamic range.

Campus networks — with his high diversity of users, traffic requirements, fast changing roles, and openness of equipment (BYOD) — present multiple unique challenges difficult to meet with traditional networking technologies. SDN/OpenFlow provides a new toolset to rethink and redesign a campus network with aims at higher scalability, flexibility, and lower cost. This paper presents PUCPLight, an SDN/OpenFlow controller for a layer 2 campus network of tens of thousands of users. PUCPLight is a first approximation to a vision of a pure SDN scalable, flexible campus network that seamlessly supports from administrative services, to high performance computing, to experimental research. This paper describes the drives behind PUCPLight design, some of the design choices, as well as initial results proving scalability to a subnet of 10K nodes.

This paper presents a simple and efficient algorithm for improving the sparse modeling of wireless communications nonlinear devices. This technique consists on a modified greedy algorithm that sorts the regressors of the model by its importance and a maximum likelihood method for the selection of the optimum number of coefficients. This approach has been applied to obtain a model with a reduced number of parameters while upholding its performance capabilities for a commercial power amplifier (PA) driven with an orthogonal frequency division multiplexing signal, following the LTE standard. Results show that the presented algorithm achieves the same level of performance than a reference compressed-sensing algorithm while providing a further reduction in the number of coefficients.

A CMOS interface circuit to transfer a digital signal between two circuits of different supply voltages is described. The interface can be used, for example, between 3 volt and 5 volt or higher voltage families. The principal characteristics of the interface circuit are: no static power dissipation, high speed, and high speed buffering.

The behavior of digital circuits in a noisy environment in mixed-signal smart-power systems is described in this paper. Several models and mechanisms explaining the process in which substrate noise affects on-chip digital circuits as well as the noise immunity behavior of digital circuits are presented and discussed. The models and mechanisms are demonstrated by simulations and by extensive test chip-based experimental data.