Abstracts:Linear minimum mean-square error (MMSE) detection achieves near-optimal performance in large-scale multiple-input multiple-output (LS-MIMO) systems but entails high computational complexity due to large matrix inversion operations. In this Letter, a novel computationally efficient algorithm based on second-order Richardson method is proposed to solve the LS-MIMO detection problem. While no a priori information for the first iteration of the second-order Richardson method is available, the conjugate gradient scheme is exploited that greatly reduces the number of iterations to achieve the desired performance. Moreover, the eigenvalue-based acceleration parameters are proposed to further accelerate the convergence rate. Numerical results demonstrate that the proposed detector outperforms the existing methods and approaches the performance of MMSE with a small number of iterations.

Abstracts:In this Letter, the authors consider a multi-unicast flow wireless communication system with a retransmission scheme. The effective capacity (EC) is used as a performance metric in order to make use of the recurrence relation approach. They also take advantage of network coding (NC) as combined with automatic repeat request (ARQ) protocol (NC-ARQ), for more robust communication. The authors' main contribution, in this work, is the derivation of the EC for <italic>M</italic>-user NC-ARQ using only the opportunity of XoR-ing <italic>M</italic> packets. They then, focus on a 3-user system and derive the EC using the opportunity of XoR-ing packets for 3 and 2 users. An interesting result is the fact that, contrary to expectation, as the number of users in a network that uses NC-ARQ increases, the performance in terms of EC increases as well, depending on how the sender schedules and manages the retransmissions.

Abstracts:Proliferation of smart internet of things (IOTs) devices has boosted the improvement of multiple networking functions which have a different capability in terms of capacity and access delay. Herein, the networking function of IoT devices should be properly selected to fully utilise the capabilities of the different types of networking technologies. In this Letter, a reinforcement learning-based self-organising scheme is proposed for the IOTs. A node selects an adequate IoT network function and adapts its topology by learning channel circumstance. To verify the performance of the proposed learning-based scheme, simulations reflect a multiple number of heterogeneous IoT networks and show that the average latency of IoT devices can be efficiently reduced compared to the conventional benchmark networks (Wi-Fi and narrow band IoT).

Abstracts:The data payload from the medium access control layer to the physical layer is referred to as the transport block (TB). In fifth-generation new radio, the TB size is determined by using a formula due to large variations of available resource elements (REs) per resource block (RB). To ensure that a matching TB size can easily be scheduled in the retransmission instance, the reference number of REs is used to determine the TB size. As a result, the actual code rate may be much higher than the target code rate. This degrades the performance of the channel coding and increases the likelihood of the decoding failure. To solve these problems, the authors quantise the number of REs within the scheduled duration that are not available for the data transmission and the demodulation reference signal transmission. The difference between the reference number of REs of the traditional scheme and the quantised number is used to determine the TB size in the proposed scheme. Compared to the traditional scheme, the proposed scheme reduces the difference between the actual code rate and the target code rate.

Abstracts:In the integrated system of underwater single-node detection and communication, active self-interference cancellation requires high accuracy of time delay estimation. To solve this problem, this Letter proposes a high accuracy delay estimation method based on phase correction. Initially, the normalised frequency correction is calculated according to the phase difference and time delay of the signal. Then, the discrete spectrum is reconstructed, and the phase is obtained by the least square method. Finally, the time delay is estimated by the phase difference between the received signal and the reference signal. This method corrects the phase and frequency errors caused by the Fourier transform and improves the accuracy of time delay estimation. Simulation results confirm the efficacy of the proposed method and achieve the accuracy requirement of the active self-interference cancellation.

Abstracts:An effective and robust hybrid algorithm consisting of particle swarm optimisation (PSO) and limited memory Broyden&#x2013;Fletcher&#x2013;Goldfarb&#x2013;Shanno (L-BFGS) method based on artificial neural network (ANN) is proposed for modelling flexible metal-oxide thin-film transistors (TFTs). The L-BFGS method as an optimiser is exploited to update the parameters of ANN and speed up the training process. A mutation strategy for PSO is derived to enhance the searching ability further. With the great global searching ability, PSO is implemented to find a hopeful initial position in solution space for the next ANN model. The simulation result shows a high accuracy not only in <italic>I&#x2013;V</italic> curve fitting but also in small-signal parameter (<inline-formula><alternatives><tex-math notation="TeX">$g_m$</tex-math><mml:math overflow="scroll"><mml:msub><mml:mi>g</mml:mi><mml:mi>m</mml:mi></mml:msub></mml:math><inline-graphic xlink:href="EL.2019.4001.IM1.gif" /></alternatives></inline-formula>, <inline-formula><alternatives><tex-math notation="TeX">$g_d$</tex-math><mml:math overflow="scroll"><mml:msub><mml:mi>g</mml:mi><mml:mi>d</mml:mi></mml:msub></mml:math><inline-graphic xlink:href="EL.2019.4001.IM2.gif" /></alternatives></inline-formula>, etc.) predictions, which have not been exposed in the training process. The measured DC characteristics of In&#x2013;Zn&#x2013;O TFTs are used to verify the proposed ANN model, which has the benefits of rapid fitting from the L-BFGS algorithm and universal searching ability from PSO.

Abstracts:In this Letter, two comparator-based voltage-mode hysteretic controller designs are proposed to enhance dynamic response for single-inductor multiple-output (SIMO) switching converters to minimise over-/under-shoot voltages due to self- and cross-regulations in fast load transients, especially with high output power and large current steps. The proposed converters are designed and simulated in a 65 nm CMOS process with standard I/O devices. With the proposed techniques, the output voltage fluctuations, including steady-state voltage ripples and over-/under-shoot voltages due to self- and cross-regulations during a 1.8 A load transient in 1 ns, are kept within 60 mV. The proposed converter delivers more than 3 A maximum load current per output and a maximum total output power above 6 W, with a peak efficiency above 90%. Compared to state-of-the-art SIMO converters, this work achieves a significantly higher output power capacity and faster dynamic response thus smaller self- and cross-regulations within 0.033 mV/mA.

Abstracts:In this Letter, a novel filtering power divider with second-order quasi-elliptic bandpass response is proposed, which has two transmission zeros near to the passband for sharp frequency selectivity. In addition, it is featured by ultra-wide stopband owing to specific topology. For demonstration, a filtering power divider example operating at 3.1 GHz is designed, fabricated and measured. Its 3 dB fractional bandwidth is about 11.9%, and the in-band isolation between two output ports exceeds 20 dB. Specifically, more than 21 dB out-of-band suppression is achieved from the upper side of the passband to 4.8<italic>f</italic>₀, and more than 12 dB suppression is achieved to at least 20<italic>f</italic>₀. The measurement agrees well with the simulation.

Abstracts:In order to study the effect of shape on graphene pressure sensors, a theoretical model of polymerisation degree is proposed in this work. According to the theoretical model of polymerisation degree, it is found that the regular polygons have better sensitivity characteristics. The more the number of sides for the regular polygons, the larger the polymerisation degree, and the better the sensitivity of the graphene pressure sensors. According to the theoretical model, the polymerisation degree of the positive triangle, regular quadrilateral, regular pentagon, and circle shape is 0.5, 0.71, 0.81, and 1, respectively. The measured results show that the relative resistance change of positive triangle, regular quadrilateral, regular pentagon, and circle shape are 8.56, 9.24, 9.61, and 10.4%, respectively. The experimental results are consistent with the theoretical results. Therefore, the theoretical model of polymerisation degree can provide effective quantitative guidance for the graphene pressure sensors.

Abstracts:In this Letter, the authors introduce a novel approach to learn representations for sentence-level paraphrase identification (PI) using BERT and ten natural language processing tasks. Their method trains an unsupervised model called BERT with two different tasks to detect whether two sentences are in paraphrase relation or not. Unlike conventional BERT, which fine tunes the target task such as PI to pre-trained BERT, twice fine-tuning deep neural networks first fine tune each task (e.g. general language understanding evaluation tasks, question answering, and paraphrase adversaries from word scrambling task) and second fine tune target PI task. As a result, the multi-fine-tuned BERT model outperformed the fine-tuned model only with Microsoft Research Paraphrase Corpus (MRPC), which is paraphrase data, except for one case of Stanford Sentiment Treebank - 2 (SST-2). Multi-task fine-tuning is a simple idea but experimentally powerful. Experiments show that fine-tuning just PI tasks to the BERT already gives enough performance, but additionally, fine-tuning similar tasks can affect performance (3.4% point absolute improvement) and be competitive with the state-of-the-art systems.