Channel Memory Research Articles

Анализ и оптимизация схем кодирования для каналов с рэлеевскими замираниями

Conventional approach to perform error-correcting decoding in channels with memory is an adoption of interleaving, which increases the receiver complexity and delay. In order to avoid these limitations, approaches for adapting an error-correcting scheme for memory channels may be used. One approach is to use error-correcting codes with modified design and decoding procedure, taking into account the presence of error bursts in the channel. Another approach to adapting the coding scheme for burst error correction is to use product codes with iterative decoding algorithms. Component codes of product codes themselves may not be able to correct error bursts, but a two-dimensional structure of product codes operating as an artificial interleaver and iterative decoding allow for correction of grouping errors. Purpose: The purpose of the study is to analyze the methods of error-correcting code adaptation to burst error correction in channels with memory in order to reduce the probability of error. Results: The study considered methods of error-correcting code scheme adaptation for low-density parity-check codes, polar codes, and product codes for burst error correction in Gilbert-Elliott channel (GE) and correlated Rayleigh fading channel with different correlation coefficients. Degree distribution of low-density parity-check code was optimized for GE channel. The influence of selected parameters of product code component codes on burst error correction in the GE channel was analyzed. The structure of the polar code has been optimized using a genetic algorithm for the correlated Rayleigh channel, which outperforms the 5G polar code design in terms of error probability. Discussion: Existing error-correcting code schemes do not provide theoretically possible limits, so the question remains of developing coding and decoding schemes capable of reaching theoretical limits.

CAPACITY ESTIMATION METHOD WITH LINEAR COMPUTATIONAL COMPLEXITY OF NARROW-BAND IONOSPHERIC DECAMETER CHANNELS WITH PSK-N-SIGNALS

The paper presents a new algorithm with linear time complexity, depending on the effective channel memory, for estimating the resolution time and capacity of a frequency-selective communication channel using a linear receiver and a multi-position phase-shift keying signal. The main limitation of the method is the possibility of its application for frequency-selective communication channel with symmetrical amplitude-frequency response relative to the central frequency of the channel, which is equal to the frequency of the received signal. From a practical point of view, this method can be used in the construction of radio engineering systems with serial data transmission in narrow-band ionospheric decameter communication channels (3-12 kHz). The key features of this method are: 1) constant number of equations equal to 1; 2) linear dependence of the number of terms in the equation on effective memory; 3) a new, more accurate procedure for estimating the effective memory of the channel, which allows you to simultaneously determine the resolution time and capacity. These features provide an increase in frequency efficiency by reducing overhead costs for setting the parameters of a serial phase radio engineering data transmission system for HF communication channels with a bandwidth of 3-12 kHz.

Random walk with multiple memory channels.

A class of one-dimensional, discrete-time random walk models with memory, termed "random walk with n memory channels" (RWnMC), is proposed. In these models the information of n (n∈Z) previous steps from the walker's entire history is needed to decide a future step. Exact calculation of the mean and variance of position of the RW2MC (n=2) has been done, which shows that it can lead to asymptotic diffusive and superdiffusive behavior in different parameter regimes. A connection between RWnMC and a Pólya-type urn model evolving by drawing n balls at a time has also been reported. This connection for the RW2MC is discussed in detail and suggests the applicability of RW2MC in many population dynamics models with multiple competing species.

A thermal neutron detector array communication protocol

To assess hadronic and electromagnetic components from extensive air shower events (EAS), 16 6Li-based ZnS (Ag) scintillator detectors combine into a thermal neutron detector array. After a shower event, each scintillator screen can concurrently catch neutrons and electrons, producing a succession of electrical signals with different pulse shapes. These signals are sampled by 16 oscilloscope-like circuits and sent over a single serial link, where EAS-related sampling results may fill each channel's memory. However, there are presently no specific studies to address data congestion in such random signal digitizers. Here, we demonstrate that decreasing the sampling rate on the trailing edge can compress data size while preserving pulse shapes and that optimizing the read/write time of memories can enhance transmission speed between asynchronous clock domains. According to test results, the 100 MHz serial transceiver can send 1996 pulses per second when only receiving data from one channel at 50 MHz rates, and it can send 242 pulses per second when simultaneously receiving data from 32 channels (16 × 2). This method increases transmitting speeds 2.3 times over previously employed algorithms, and the compressed rate reaches 97.74%. In addition, we found that the total error rate is lower than 10–10. Our findings also show how memory controllers would probably operate in data flows of 32 channels, predicting the slower writing time and the faster reading time that can complete sending a packet of one channel in the shortest time. We anticipate that our assay can help more detector arrays that require waveform transmission.

Deep Neural Network Memory Performance and Throughput Modeling and Simulation Framework

Deep neural networks (DNNs) are widely used in various artificial intelligence applications and platforms, such as sensors in internet of things (IoT) devices, speech and image recognition in mobile systems, and web searching in data centers. While DNNs achieve remarkable prediction accuracy, they introduce major computational and memory bandwidth challenges due to the increasing model complexity and the growing amount of data used for training and inference. These challenges introduce major difficulties not only due to the constraints of system cost, performance, and energy consumption, but also due to limitations in currently available memory bandwidth. The recent advances in semiconductor technologies have further intensified the gap between computational hardware performance and memory systems bandwidth. Consequently, memory systems are, today, a major performance bottleneck for DNN applications. In this paper, we present DRAMA, a deep neural network memory simulator. DRAMA extends the SCALE-Sim simulator for DNN inference on systolic arrays with a detailed, accurate, and extensive modeling and simulation environment of the memory system. DRAMA can simulate in detail the hierarchical main memory components—such as memory channels, modules, ranks, and banks—and related timing parameters. In addition, DRAMA can explore tradeoffs for memory system performance and identify bottlenecks for different DNNs and memory architectures. We demonstrate DRAMA’s capabilities through a set of experimental simulations based on several use cases.

Low-Complexity Near-Optimum Symbol Detection Based on Neural Enhancement of Factor Graphs

We consider the application of the factor graph framework for symbol detection on linear inter-symbol interference channels. Based on the Ungerboeck observation model, a detection algorithm with appealing complexity properties can be derived. However, since the underlying factor graph contains cycles, the sum-product algorithm (SPA) yields a suboptimal algorithm. In this paper, we develop and evaluate efficient strategies to improve the performance of the factor graph-based symbol detection by means of neural enhancement. In particular, we consider neural belief propagation and generalizations of the factor nodes as an effective way to mitigate the effect of cycles within the factor graph. By applying a generic preprocessor to the channel output, we propose a simple technique to vary the underlying factor graph in every SPA iteration. Using this dynamic factor graph transition, we intend to preserve the extrinsic nature of the SPA messages which is otherwise impaired due to cycles. Simulation results show that the proposed methods can massively improve the detection performance, even approaching the maximum a posteriori performance for various transmission scenarios, while preserving a complexity which is linear in both the block length and the channel memory.

Advanced nonlinearity equalizer with TC-NL-MLSE for transmitting beyond 200-Gb/s PAM-8 in IM/DD systems.

The severe band-limited effect resulted from the low-cost optical transceiver increases the channel memory length and the number of taps of the equalizers. Besides, the interaction of fiber dispersion and square-law detection introduce nonlinear distortions in intensity modulation and direct-detection (IM/DD) transmission systems. The serious band-limited effect and nonlinear distortions degrade the transmission performance and bring challenges to current equalizers for low-complexity implementation. In this paper, we propose a trellis-compression nonlinear maximum likelihood sequence estimation (TC-NL-MLSE) algorithm to compensate the linear and nonlinear distortions with lower complexity. In the TC-NL-MLSE, we introduce a polynomial nonlinear filter (PNLF) to partly compensate both the linear distortions and nonlinear distortions. Then, we establish a look-up-table (LUT) to calculate the nonlinear branch metric (BM). To simplify the calculation, two or three levels with the highest probabilities are selected according to decision thresholds for each symbol to compress the state-trellis graph (STG). This significantly reduces computational complexity on BM calculations especially for high-order modulations. We conduct experiments to transmit beyond the 200-Gb/s PAM-8 signal over 2-km standard single mode fiber (SSMF) at C-band. The TC-NL-MLSE outperforms the reduced-state MLSE with PNLF, and can reach the 7%-overhead hard-decision forward error correction threshold. Moreover, the TC-NL-MLSE reduces the complexity by 97% compared with standard LUT-MLSE, limiting the multipliers around 100 at the expense of only 0.2-dB receiver sensitivity penalty.

Optimization design of high-speed data acquisition system based on DMA double cache mechanism

The large-capacity on-chip cache often causes timing problems or erroneous transmission in the high-speed acquisition of massive data. However, many factors always affect the communication performance of the system. In this paper, a high-speed data acquisition system (DAS) based on DMA double cache mechanism is introduced. Aurora protocol is used for optical fiber transmission, and the line rate dynamic adaptive design of the X1 channel is realized. In the SDRAM controller, the DMA dual channel is targeted optimized, including memory ping-pong operation and channel arbiter. In addition, collecting data from the lower computer or transmitting data to the upper computer are both based on software instructions in our design, which is more flexible and stable than receiving data without instructions. Finally, to optimize the transmission process of cached data, we cascaded FIFOs in different modules. The results show that the system can receive optical fiber data at the rate of 7.57Gbps, and the average DMA communication efficiency reaches 76.9%.

Memory effects on bidirectional teleportation

In this contribution, we have investigated the bidirectional quantum teleportation (BQT) of single qubit states using a Bell state influenced by decoherence channels with memory, dephasing and amplitude damping channels. The expressions of the partial transposed density operator, as a measure of the entanglement remaining in the BQT quantum channel, the teleportation fidelities and the quantum Fisher information are also evaluated. We find that both these last quantities depend on the survival amount of entanglement in the BQT quantum channel, on the decoherence factor and on the correlation degree of the decoherence channel. We show that in the Markovian regime, the Negativity, the teleportation average fidelities and the quantum Fisher information are slightly enhanced by considering the classical channel correlations. Besides, in the non-Markovian regime, these three quantities could be improved for a long period of time.

Tuning Ambipolarity of the Conjugated Polymer Channel Layers of Floating-Gate Free Transistors: From Volatile Memories to Artificial Synapses.

Three-terminal synaptic transistor has drawn significant research interests for neuromorphic computation due to its advantage of facile device integrability. Lately, bulk-heterojunction-based synaptic transistors with bipolar modulation areproposed to exempt the use of an additional floating gate. However, the actual correlation between the channel's ambipolarity, memory characteristic, and synaptic behavior for a floating-gate free transistor has not been investigated yet. Herein, by studying five diketopyrrolopyrrole-benzotriazole dual-acceptor random conjugated polymers, a clear correlation among the hole/electron ratio, the memory retention characteristic, and the synaptic behavior for the polymer channel layer in a floating-gate free transistor is described. It reveals that the polymers with balanced ambipolarity possess better charge trapping capabilities and larger memory windows; however, the high ambipolarity results in higher volatility of the memory characteristics, namely poor memory retention capability. In contrast, the polymer with a reduced ambipolarity possesses an enhanced memory retention capability despite showing a reduced memory window. It isfurther manifested that this enhanced charge retention capability enables the device to present artificial synaptic characteristics. Theresults highlight the importance of the channel's ambipolarity of floating-gate free transistors on the resultant volatile memory characteristics and synaptic behaviors.

On the Capacity of Gaussian MIMO Channels With Memory

The operational capacity of Gaussian MIMO channels with memory was obtained by Brandenburg and Wyner (1974) under certain mild assumptions on the channel impulse response and its noise covariance matrix, which essentuially require channel memory to be not too strong. This channel was also considered by Tsybakov (2006) and its information capacity was obtained in some cases. It was further conjectured, based on numerical evidence, that these capacities are the same in all cases. This conjecture is proved here. An explicit closed-form expression for the optimal input power spectral density matrix is also given. The obtained result is further extended to the case of joint constraints, including per-antenna and interference power constraints as well as energy harvesting constraints. These results imply the information-theoretic optimality of OFDM-type transmission systems for such channels with memory.

Memory Channel Models of a Hybrid PLC-VLC Link for a Smart Underground Mine

In this article, discrete memory channel models of a low-cost, visible light communication (VLC) and a hybrid amplify-and-forward (AF) powerline communication (PLC) and VLC channel are derived using data obtained from an underground mining testbed. The testbed is set up close to the mining face, where blasting occurs. Then, the noise distributions as a result of the connection of a blasting equipment and the mains switchboard supply on the powerline channel are modeled. The baseband signals are communicated over software-defined radios, using multicarrier modulation in the VLC and hybrid AF PLC-VLC channels. Aided by a Fritchman 3-state semi-hidden Markov model (SHMM) for channel state classification, the channels are modeled using a first-order (FO) Markov process, which considers only one previous state in the channel. In addition, the block diagonal (BD) Markov model, which groups similar successive states of the channel, is implemented in order to reduce the training complexity. The models for the error distribution are then derived for both VLC and hybrid PLC-VLC channels. Results show that the probability of error obtained from the trained models closely matches the probability of error obtained from the channel measurements, validating the suitability of the SHMM for modeling VLC and hybrid PLC-VLC channels. In addition, the modified BD Markov model, which groups the channel states shows better performance with reduced complexity when compared with the FO Markov model.

Research on Efficient Channel Decoding Algorithm for Memory Channel and Short Packet Transmission in Smart Grid

Many smart factories use the smart grid for power system automation, and its wireless control technology requires low-time-delay and high-reliability communication. Guessing random additive noise decoding algorithm has outstanding short packet error correction performance. In the decoding process, the order of noise parameter combination affects the decoding delay. Aiming at the communication problem of the smart grid in the process of factory power supply and distribution, this study analyzes the characteristics of the original noise parameter ranking algorithm. When the steady-state flip probability is large, more search times are required to obtain the correct combination of noise parameters, which means that greater delay is required for decoding in the time-varying channel. To solve the aforementioned problems, this study optimizes the noise parameter ranking before the noise error mode arrangement and proposes a noise parameter ranking algorithm for predicting the symbol string. First, the channel perception is completed by edge computing. Then, the algorithm uses the obtained soft information to rank the channel noise parameters. Simulation results show that the proposed algorithm has better search performance than the original sorting algorithm, especially when the channel parameter b is greater than 0.5. Finally, by comparing the BM Decoding Algorithm of BCH with different noise parameter ranking algorithms of decoding, the results show that the noise parameter ranking algorithm proposed in this study has better decoding performance in the environmental channel of the smart factory, so as to improve the reliability of the smart grid in the process of factory power supply and distribution.

Deep Hybrid Neural Network-Based Channel Equalization in Visible Light Communication

In this letter, the channel impairments compensation of visible light communication is formulated as a time sequence with memory prediction. Then we propose efficient nonlinear post equalization, using a combined long-short term memory (LSTM) and deep neural network (DNN), to learn the complicated channel characteristics and recover the original transmitted signal. We leverage the long-term memory parameters of LSTM to represent the sequence causality within the memory channel and refine the results by DNN to improve the reconstruction accuracy. Results demonstrate that the proposed scheme can robustly address the overall channel impairments and accurately recover the original transmitted signal with fairly fast convergence speed. Besides, it can achieve better balance between performance and complexity than that of the conventional competitive approaches, which demonstrates the potential and validity of the proposed methodology for channel equalization.

Ship Pitch Prediction Based on Bi-ConvLSTM-CA Model

When a ship is sailing at sea, its pitch angle will be affected by ship motions such as turning angle, relative wind speed, relative wind direction, velocity in surge and velocity in sway of the ship. Due to the randomness of ship motion attitude and the difficulty of capturing the motion rules, traditional machine learning models, statistical learning models and single deep learning models cannot accurately capture the correlation information between multiple variables, which results in poor prediction accuracy. To solve this problem, the bidirectional convolutional long short-term memory neural network (Bi-ConvLSTM) and channel attention (CA) for ship pitch prediction are used to build a Bi-ConvLSTM-CA model in this paper. The Bi-ConvLSTM-CA prediction model can simultaneously extract the time information and spatial information of the ship motion data, and use the channel attention mechanism to process the output of different time steps to obtain the corresponding weight of each channel. Using the weights to do dot product with the output of Bi-ConvLSTM, the resulting attention mechanism output is processed to produce predicted value by the fully connected layer. Compared with other models, the RMSE index of Bi-ConvLSTM-CA model decreased by at least 28.20%; the MAPE index decreased by at least 29.39%; the MAE index decreased by at least 22.68%. The experimental results of real ship data show that the proposed Bi-ConvLSTM-CA model has a significant reduction in mean absolute percentage error (MAPE), mean square error (MSE) and mean absolute error (MAE) compared with other advanced models, which verifies the effectiveness of the Bi-ConvLSTM-CA model in predicting ship pitch angle.

Driver vigilance detection for high-speed rail using fusion of multiple physiological signals and deep learning

With the popularity of high-speed rail in China and with faster operation speeds, the safety of railways has also received more attention. This paper proposes a vigilance detection method for high-speed rail drivers based on wireless wearable multi-physiological signals fusion and deep learning. The intended method consists of three components: wireless wearable signal acquisition equipment, physiological signal preprocessing and driver vigilance detection. In the initial stage, a wireless wearable device based on open source brain–computer interfaces was used to collect electroencephalogram, electrocardiogram and electromyogram signals in the high-speed rail simulation environment. Secondly, linear filtering, fast independent component analysis and wavelet filtering are performed on the three kinds of signals, and reasonable slicing is performed to make a dataset. Finally, a convolutional recurrent neural network with channel attention mechanism and memory ability is proposed. Multiple physiological signals from wireless wearable devices are used to train the network. This network improves the ability to recognize the vigilance of drivers and verifies the effectiveness of the combination of squeeze-and-excitation block and long short-term memory with convolutional neural network. Furthermore, the vigilance detection effectiveness was evaluated under different signal combinations; the testing set verified the accuracy of the network as 98.11%. Results prove the feasibility of the high-speed rail driver vigilance detection method based on multiple physiological signals and deep learning, which can help to avoid high-speed rail accidents.

Single-image Deraining via a channel memory network

Single-image Deraining via a channel memory network

Decorrelation of optical critical dimensions in mid-infrared ellipsometric spectroscopy of high aspect ratio etch profiles

Monitoring the high aspect ratio etch profiles in state-of-the-art three-dimensional NAND memory fabrication processes has pushed metrology technologies to new limits. Here, we discuss how a mid-infrared ellipsometric measurement can yield angstrom level discrimination in critical dimension changes of memory channel hole (CH) profiles across such a memory chip. Using finite-difference time-domain and rigorous coupled-wave analysis simulations, we demonstrate how dispersion mitigated mid-infrared beam penetration into these memory structures permits parameter decorrelation and the measurement of the full CH profile.

Noise Error Pattern Generation Based on Successive Addition-Subtraction for GRAND-MO

Guessing random additive noise decoding (GRAND) is a capacity-approaching universal algorithm. The GRAND Markov order (GRAND-MO) variant is effective to correct burst errors in a Markov chain modeled memory channel, and the core of GRAND-MO is to generate putative noise error patterns in MO effectively. For purpose of a uniform noise error pattern generation scheme of GRAND-MO, this letter proposes a successive addition-subtraction scheme to generate noise/zero error permutations for predefined MO parameters. Detailed procedures of the proposed scheme are presented, and its correctness is also proofed through theoretical derivation. By embedding the “1” and “0” bursts alternately, the proposed scheme can generate all noise error patterns in MO, which is helpful to improve the error correction capability of GRAND-MO decoders for flexible applications.

RPkNN: An OpenCL-Based FPGA Implementation of the Dimensionality-Reduced kNN Algorithm Using Random Projection

Due to the so-called curse of dimensionality and increase in the size of databases, there is an ever-increasing demand for computing resources and memory bandwidth when performing the k-nearest neighbors (kNNs) algorithm, resulting in a slow-down to process large datasets. This work presents an OpenCL-based framework for accelerating the kNN algorithm on field-programmable gate arrays (FPGAs) benefiting from the random projection dimensionality reduction. The proposed RPkNN framework includes two compute modules implementing a throughput-optimized hardware architecture based on random projection and the kNN algorithm and a host program facilitating easy integration of the compute modules in the existing applications. RPkNN also utilizes a new buffering scheme tailored to random projection and the kNN algorithm. The proposed architecture enables parallel kNN computations with a single memory channel and takes advantage of the sparsity features of the input data to implement a highly optimized and parallel implementation of random projection. We employ a computation storage device (CSD) to directly access the high-dimensional data on non-volatile memory express (NVMe) solid state drive (SSD) and store and reuse the compressed and low-dimensional data on the FPGA dynamic random access memory (DRAM), hence eliminating data transfers to the host DRAM. We compare RPkNN implemented on the Samsung SmartSSD CSD with the kNN implementation of the scikit-learn library running on an Intel Xeon Gold 6154 CPU. The experimental results show that the proposed RPkNN solution achieves, on average, <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$26\times $ </tex-math></inline-formula> and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$46\times $ </tex-math></inline-formula> higher performance across different dimensions per a single kNN computation for the SIFT1M and GIST1M databases, respectively. Finally, RPkNN is <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$1.7\times $ </tex-math></inline-formula> faster than the similar FPGA-based reference method.