We describe the multilingual customer solution by Language I/O in this paper. With data security and confidentiality ensured by the ISO 27001 certification, global corporations can turn monolingual customer support agents into efficient multilingual brand ambassadors in less than 24 hours. Our solution supports more than 100 languages.

QE Viewer is a web-based tool for visualizing results of a Machine Translation Quality Estimation (QE) system. It allows users to see information on the predicted post-editing distance (PED) for a given file or sentence, and highlighted words that were predicted to contain MT errors. The tool can be used in a variety of academic, educational and commercial scenarios.

The Google Patents is one of the main important sources of patents information. A striking characteristic is that many of its abstracts are presented in more than one language, thus making it a potential source of parallel corpora. This article presents the development of a parallel corpus from the open access Google Patents dataset in 74 language pairs, comprising more than 68 million sentences and 800 million tokens. Sentences were automatically aligned using the Hunalign algorithm for the largest 22 language pairs, while the others were abstract (i.e. paragraph) aligned. We demonstrate the capabilities of our corpus by training Neural Machine Translation (NMT) models for the main 9 language pairs, with a total of 18 models. Our parallel corpus is freely available in TSV format and with a SQLite database, with complementary information regarding patent metadata.

We will present our solution to replace the usage of publicly available machine translation (MT) services in companies where privacy and confidentiality are key. Our MT portal can translate across a variety of languages using neural machine translation, and supports an extensive number of file types. Corporations are using it to enable multilingual communication everywhere.

This paper provides useful insights into the practical design of bit allocation algorithms in multiantenna multiuser orthogonal frequency-division multiplexing (OFDM) systems. With the degrees of freedom obtained with multiple antennas and multiple subcarriers, the performance might be enhanced at the expense of a higher complexity. Since the scheduling with realistic integer mappings is an NP-complete combinatorial problem, suboptimum solutions based on the scalar product are shown to be good candidates for yielding a realizable scheduler at a wireless physical layer. Additionally, a power reuse strategy is proposed to lower the computational requirements of such a system. Besides the tradeoff between performance and complexity, there exists the tradeoff between performance and signaling. Therefore, we show that the required signaling might be reduced either by a user-subcarrier clustering or by using a scheme that forces an equal mapping for all the users at the same subcarrier, which might be well-suited if instantaneous fairness is required. The proposed strategies are evaluated for typical OFDM-based wireless LAN scenarios.

In multi-user communications, the access point (AP) has several alternatives for distributing the scarce resources among users. Since there exists a trade-off between the global performance and the individual needs, an analytical framework to study fairness is derived, which completes the scope given by the existing fairness indexes in the literature. The framework proposed in this paper is a way to interpret fairness that has been inspired by portfolio selection; basically, it analyzes the mean vs. standard deviation trade-off. In this work, the target application is a multi-antenna AP transmitting simultaneously to several single-antenna terminals, although this framework is valid to analyze other procedures in multi-user communications.

Given a zero forcing transmit beamforming, we focus on how the multi-antenna access point distributes the scarce resource (power) among the single-antenna terminals. Since there is a clear trade-off between the satisfaction of the individual needs and the global performance of the cell, several criteria are proposed, ranging from a classical physical layer point of view of capacity (rate) maximization to bit error rate (BER)-based cost functions, which are closer to the second layer of the protocol stack. Between two traditional techniques, namely the uniform power allocation and the equal BER and rate, a new one is proposed, which ultimately provides an intermediate performance. Then, we add BER (or signal to noise ratio) constraints so that the admission control problem has to be solved. Among traditional options, we propose a new mechanism to balance the above-mentioned trade-off between the total performance and the particular user behavior. The results in terms of fairness are presented by a mean vs. variance plot and by the Gini plot.

In this paper, we address the problem of transmit beamforming, power allocation, bit loading, and admission control in a multi-user scenario, where several single antenna mobile terminals are served by one base station equipped with an antenna array. In particular, we explore the impact of using either optimal or zero-forcing transmit beamforming schemes in combination with different bit allocation strategies, namely the maximization of the sum rate (MSR), the maximization of the minimum rate (MMR) and a modified version of the latter, for a finite set of modulation schemes. The performance evaluation is conducted by means of system-level simulations, using a realistic channel model for urban micro cells, with an emphasis not only on the aggregated cell throughput but also on the associated fairness issues.

At a multi-antenna base station, the multiple antennas are used to enhance the scheduling task. After applying a zero forcing transmit beamforming, the scheduling distributes the resources among the users. Several criteria are presented and their performance in the high signal-to-noise ratio (SNR) range is analyzed. The next issue is to take into account the SNR requirements from the users and perform admission control accordingly. An algorithm that yields the optimum performance is proposed, together with a new criterion that falls between the optimization of the best global performance and the satisfaction of the individual needs. Simulation results are provided to show the performance of the techniques.

This paper addresses the problem of the spatial scheduling of users in a cell for simultaneous downlink transmission from a base statimultiple antennas on (BS) having multiple antennas under a perspective of joint Physical and Medium Access Control (PHY-MAC) design. First of all, we compute the transmit beamvectors for each group according to a zero forcing (ZF) criterion, which gives a simple closed-form solution. We show first that it is equivalent to the minimization of the maximum bit error rate (BER). In this paper, the main contribution lies on the resolution of the NP-complete combinatorial problem that comes up as a cost function if we want to minimize the total transmit power. The solution of the NP-complete problem is performed by the stochastic technique simulated annealing (SA). Additionally, we present two heuristic algorithms that may enable a real-time implementation of this scheduling approach.

This paper studies the application of the minimum mean square error (MMSE) beamformer to orthogonal frequency division multiplexing (OFDM)-based wireless local area networks. The questions here addressed are mainly the design with finite-length data and the choice of the OFDM signal domain where the beamformer is applied, either frequency or time. As OFDM signals need more samples than other modulations to stabilize the estimation of the signal statistics, how to exploit the finite-length training sequence provided for the design of equalizers becomes an important issue. The paper also shows that the usual frequency processing in OFDM is not always the best choice for the spatial beamforming, mainly for channels with a very high delay spread. Then, time processing turns out to be the best suited approach in terms of the tradeoff between performance and complexity. Additionally, novel modifications of the MMSE spatial filter are proposed to improve the raw bit-error rate performance: 1) a temporal semiblind approach that exploits the cyclic prefix and 2) windowing in the frequency domain.

Adaptive antenna arrays can be used to improve the performance of Orthogonal Frequency Division Multiplexing systems. They are usually applied in the frequency domain, after the FFT demodulation. In this paper, the selected algorithms are applied in time domain. A blind method is firstly proposed for the maximization of the temporal Signal to Noise plus Interference Ratio (SNIR) in frequency selective channels exploiting the temporal redundancy introduced by the cyclic prefix (CP). This algorithm may cope well with non-stationary interference that appears during the burst transmission. This blind technique is extended so as to take advantage of the preamble that precedes the useful data in the HIPERLAN/2 (HL/2) specification. Simulations show that this latter semiblind approach improves the overall performance of the system with respect to the traditional Sample Matrix Inversion (SMI) algorithm even in interference stationary scenarios. Additionally, a moving pilot structure is proposed to perform more accurate channel estimation, which further improves the performance of the semiblind technique.

The performance of very high data rate wireless local area networks (WLAN) is degraded by several problems such as long delay spread in outdoor environments or interference that may arise both from other WLAN or from other systems sharing the same band. The current standards allow the use of multiple antennas at the receiver in a single input multiple output (SIMO) configuration. Traditionally, algorithms have been applied in the frequency domain, but this paper presents a comparison of three approaches to be applied also in the time domain. It is shown that a significant uncoiled bit error rate (BER) performance improvement can be achieved by working before the FFT at the receiver side, especially for channels with large delay spreads.

The sample matrix inversion (SMI) algorithm has been widely applied in the filed of adaptive antenna arrays. However, in time division multiple access (TDMA) systems with known data sequences, a low cost solution would imply the computation of the beamformer based exclusively on this data. In some cases, this might not be possible since the known data could not be sufficient to get a stable estimate of both the sample covariance matrix and the steering vector needed for the SMI spatial filter. In this paper, windowing techniques are applied in order to calculate a beamformer per subcarrier for orthogonal frequency division multiplexing (OFDM) systems with very limited known training. This approach is compared to the subcarrier grouping approach, and the presented results show that the former method provides a further performance gain, even in channels with high delay spread.