Low-Latency Active Noise Control Using Attentive Recurrent Network

Processing latency is a critical issue for active noise control (ANC) due to the causality constraint of ANC systems. This paper addresses low-latency ANC in the context of deep learning (i.e. deep ANC). A time-domain method using an attentive recurrent network (ARN) is employed to perform deep ANC with smaller frame sizes, thus reducing algorithmic latency of deep ANC. In addition, we introduce a delay-compensated training to perform ANC using predicted noise for several milliseconds. Moreover, a revised overlap-add method is utilized during signal resynthesis to avoid the latency introduced due to overlaps between neighboring time frames. Experimental results show the effectiveness of the proposed strategies for achieving low-latency deep ANC. Combining the proposed strategies is capable of yielding zero, even negative, algorithmic latency without affecting ANC performance much, thus alleviating the causality constraint in ANC design.

A Low-Latency Optimization of a Rust-Based Secure Operating System for Embedded Devices

Critical systems such as drone control or power grid control applications rely on embedded devices capable of a real-time response. While much research and advancements have been made to implement low-latency and real-time characteristics, the security aspect has been left aside. All current real-time operating systems available for industrial embedded devices are implemented in the C programming language, which makes them prone to memory safety issues. As a response to this, Tock, an innovative secure operating system for embedded devices written completely in Rust, has recently appeared. The only downside of Tock is that it lacks the low-latency real-time component. Therefore, the purpose of this research is to leverage the extended Berkeley Packet Filter technology used for efficient network traffic processing and to add the low-latency capability to Tock. The result is a secure low-latency operating system for embedded devices and microcontrollers capable of handling interrupts at latencies as low as 60 µs.

A Novel Low-Latency and Energy-Efficient Task Scheduling Framework for Internet of Medical Things in an Edge Fog Cloud System

In healthcare, there are rapid emergency response systems that necessitate real-time actions where speed and efficiency are critical; this may suffer as a result of cloud latency because of the delay caused by the cloud. Therefore, fog computing is utilized in real-time healthcare applications. There are still limitations in response time, latency, and energy consumption. Thus, a proper fog computing architecture and good task scheduling algorithms should be developed to minimize these limitations. In this study, an Energy-Efficient Internet of Medical Things to Fog Interoperability of Task Scheduling (EEIoMT) framework is proposed. This framework schedules tasks in an efficient way by ensuring that critical tasks are executed in the shortest possible time within their deadline while balancing energy consumption when processing other tasks. In our architecture, Electrocardiogram (ECG) sensors are used to monitor heart health at home in a smart city. ECG sensors send the sensed data continuously to the ESP32 microcontroller through Bluetooth (BLE) for analysis. ESP32 is also linked to the fog scheduler via Wi-Fi to send the results data of the analysis (tasks). The appropriate fog node is carefully selected to execute the task by giving each node a special weight, which is formulated on the basis of the expected amount of energy consumed and latency in executing this task and choosing the node with the lowest weight. Simulations were performed in iFogSim2. The simulation outcomes show that the suggested framework has a superior performance in reducing the usage of energy, latency, and network utilization when weighed against CHTM, LBS, and FNPA models.

Low-Latency Spiking Neural Networks Using Pre-Charged Membrane Potential and Delayed Evaluation

Spiking neural networks (SNNs) have attracted many researchers' interests due to its biological plausibility and event-driven characteristic. In particular, recently, many studies on high-performance SNNs comparable to the conventional analog-valued neural networks (ANNs) have been reported by converting weights trained from ANNs into SNNs. However, unlike ANNs, SNNs have an inherent latency that is required to reach the best performance because of differences in operations of neuron. In SNNs, not only spatial integration but also temporal integration exists, and the information is encoded by spike trains rather than values in ANNs. Therefore, it takes time to achieve a steady-state of the performance in SNNs. The latency is worse in deep networks and required to be reduced for the practical applications. In this work, we propose a pre-charged membrane potential (PCMP) for the latency reduction in SNN. A variety of neural network applications (e.g., classification, autoencoder using MNIST and CIFAR-10 datasets) are trained and converted to SNNs to demonstrate the effect of the proposed approach. The latency of SNNs is successfully reduced without accuracy loss. In addition, we propose a delayed evaluation method (DE), by which the errors during the initial transient are discarded. The error spikes occurring in the initial transient is removed by DE, resulting in the further latency reduction. DE can be used in combination with PCMP for further latency reduction. Finally, we also show the advantages of the proposed methods in improving the number of spikes required to reach a steady-state of the performance in SNNs for energy-efficient computing.

Real-time, low-latency closed-loop feedback using markerless posture tracking

The ability to control a behavioral task or stimulate neural activity based on animal behavior in real-time is an important tool for experimental neuroscientists. Ideally, such tools are noninvasive, low-latency, and provide interfaces to trigger external hardware based on posture. Recent advances in pose estimation with deep learning allows researchers to train deep neural networks to accurately quantify a wide variety of animal behaviors. Here, we provide a new DeepLabCut-Live! package that achieves low-latency real-time pose estimation (within 15 ms, >100 FPS), with an additional forward-prediction module that achieves zero-latency feedback, and a dynamic-cropping mode that allows for higher inference speeds. We also provide three options for using this tool with ease: (1) a stand-alone GUI (called DLC-Live! GUI), and integration into (2) Bonsai, and (3) AutoPilot. Lastly, we benchmarked performance on a wide range of systems so that experimentalists can easily decide what hardware is required for their needs.

A Co-Design-Based Reliable Low-Latency and Energy-Efficient Transmission Protocol for UWSNs

Recently, underwater wireless sensor networks (UWSNs) have been considered as a powerful technique for many applications. However, acoustic communications in UWSNs bring in huge QoS issues for time-critical applications. Additionally, excessive control packets and multiple copies during the data transmission process exacerbate this challenge. Faced with these problems, we propose a reliable low-latency and energy-efficient transmission protocol for dense 3D underwater wireless sensor networks to improve the QoS of UWSNs. The proposed protocol exploits fewer control packets and reduces data-packet copies effectively through the co-design of routing and media access control (MAC) protocols. The co-design method is divided into two steps. First, the number of handshakes in the MAC process will be greatly reduced via our forwarding-set routing strategy under the guarantee of reliability. Second, with the help of information from the MAC process, network-update messages can be used to replace control packages through mobility prediction when choosing a route. Simulation results show that the proposed protocol has a considerably higher reliability, and lower latency and energy consumption in comparison with existing transmission protocols for a dense underwater wireless sensor network.

An Irregular Graph Based Network Code for Low-Latency Content Distribution

To fulfill the increasing demand on low-latency content distribution, this paper considers content distribution using generation-based network coding with the belief propagation decoder. We propose a framework to design generation-based network codes via characterizing them as building an irregular graph, and design the code by evaluating the graph. The and-or tree evaluation technique is extended to analyze the decoding performance. By allowing for non-constant generation sizes, we formulate optimization problems based on the analysis to design degree distributions from which generation sizes are drawn. Extensive simulation results show that the design may achieve both low decoding cost and transmission overhead as compared to existing schemes using constant generation sizes, and satisfactory decoding speed can be achieved. The scheme would be of interest to scenarios where (1) the network topology is not known, dynamically changing, and/or has cycles due to cooperation between end users, and (2) computational/memory costs of nodes are of concern but network transmission rate is spare.

Parallel imaging and convolutional neural network combined fast MR image reconstruction: Applications in low-latency accelerated real-time imaging

PURPOSE

To develop and evaluate a parallel imaging and convolutional neural network combined image reconstruction framework for low-latency and high-quality accelerated real-time MR imaging.

METHODS

Conventional Parallel Imaging reconstruction resolved as gradient descent steps was compacted as network layers and interleaved with convolutional layers in a general convolutional neural network. All parameters of the network were determined during the offline training process, and applied to unseen data once learned. The proposed network was first evaluated for real-time cardiac imaging at 1.5 T and real-time abdominal imaging at 0.35 T, using threefold to fivefold retrospective undersampling for cardiac imaging and threefold retrospective undersampling for abdominal imaging. Then, prospective undersampling with fourfold acceleration was performed on cardiac imaging to compare the proposed method with standard clinically available GRAPPA method and the state-of-the-art L1-ESPIRiT method.

RESULTS

Both retrospective and prospective evaluations confirmed that the proposed network was able to images with a lower noise level and reduced aliasing artifacts in comparison with the single-coil based and L1-ESPIRiT reconstructions for cardiac imaging at 1.5 T, and the GRAPPA and L1-ESPIRiT reconstructions for abdominal imaging at 0.35 T. Using the proposed method, each frame can be reconstructed in less than 100 ms, suggesting its clinical compatibility.

CONCLUSION

The proposed Parallel Imaging and convolutional neural network combined reconstruction framework is a promising technique that allows low-latency and high-quality real-time MR imaging.

High-Throughput Line Buffer Microarchitecture for Arbitrary Sized Streaming Image Processing

Parallel hardware designed for image processing promotes vision-guided intelligent applications. With the advantages of high-throughput and low-latency, streaming architecture on FPGA is especially attractive to real-time image processing. Notably, many real-world applications, such as region of interest (ROI) detection, demand the ability to process images continuously at different sizes and resolutions in hardware without interruptions. FPGA is especially suitable for implementation of such flexible streaming architecture, but most existing solutions require run-time reconfiguration, and hence cannot achieve seamless image size-switching. In this paper, we propose a dynamically-programmable buffer architecture (D-SWIM) based on the Stream-Windowing Interleaved Memory (SWIM) architecture to realize image processing on FPGA for image streams at arbitrary sizes defined at run time. D-SWIM redefines the way that on-chip memory is organized and controlled, and the hardware adapts to arbitrary image size with sub-100 ns delay that ensures minimum interruptions to the image processing at a high frame rate. Compared to the prior SWIM buffer for high-throughput scenarios, D-SWIM achieved dynamic programmability with only a slight overhead on logic resource usage, but saved up to 56 % of the BRAM resource. The D-SWIM buffer achieves a max operating frequency of 329.5 MHz and reduction in power consumption by 45.7 % comparing with the SWIM scheme. Real-world image processing applications, such as 2D-Convolution and the Harris Corner Detector, have also been used to evaluate D-SWIM's performance, where a pixel throughput of 4.5 Giga Pixel/s and 4.2 Giga Pixel/s were achieved respectively in each case. Compared to the implementation with prior streaming frameworks, the D-SWIM-based design not only realizes seamless image size-switching, but also improves hardware efficiency up to 30 × .

In-DRAM Cache Management for Low Latency and Low Power 3D-Stacked DRAMs

Recently, 3D-stacked dynamic random access memory (DRAM) has become a promising solution for ultra-high capacity and high-bandwidth memory implementations. However, it also suffers from memory wall problems due to long latency, such as with typical 2D-DRAMs. Although there are various cache management techniques and latency hiding schemes to reduce DRAM access time, in a high-performance system using high-capacity 3D-stacked DRAM, it is ultimately essential to reduce the latency of the DRAM itself. To solve this problem, various asymmetric in-DRAM cache structures have recently been proposed, which are more attractive for high-capacity DRAMs because they can be implemented at a lower cost in 3D-stacked DRAMs. However, most research mainly focuses on the architecture of the in-DRAM cache itself and does not pay much attention to proper management methods. In this paper, we propose two new management algorithms for the in-DRAM caches to achieve a low-latency and low-power 3D-stacked DRAM device. Through the computing system simulation, we demonstrate the improvement of energy delay product up to 67%.

NFBLab-A Versatile Software for Neurofeedback and Brain-Computer Interface Research

Neurofeedback (NFB) is a real-time paradigm, where subjects learn to volitionally modulate their own brain activity recorded with electroencephalographic (EEG), magnetoencephalographic (MEG) or other functional brain imaging techniques and presented to them via one of sensory modalities: visual, auditory or tactile. NFB has been proposed as an approach to treat neurological conditions and augment brain functions. Although the early NFB studies date back nearly six decades ago, there is still much debate regarding the efficiency of this approach and the ways it should be implemented. Partly, the existing controversy is due to suboptimal conditions under which the NFB training is undertaken. Therefore, new experimental tools attempting to provide optimal or close to optimal training conditions are needed to further exploration of NFB paradigms and comparison of their effects across subjects and training days. To this end, we have developed open-source NFBLab, a versatile, Python-based software for conducting NFB experiments with completely reproducible paradigms and low-latency feedback presentation. Complex experimental protocols can be configured using the GUI and saved in NFBLab's internal XML-based language that describes signal processing tracts, experimental blocks and sequences including randomization of experimental blocks. NFBLab implements interactive modules that enable individualized EEG/MEG signal processing tracts specification using spatial and temporal filters for feature selection and artifacts removal. NFBLab supports direct interfacing to MNE-Python software to facilitate source-space NFB based on individual head models and properly tailored individual inverse solvers. In addition to the standard algorithms for extraction of brain rhythms dynamics from EEG and MEG data, NFBLab implements several novel in-house signal processing algorithms that afford significant reduction in latency of feedback presentation and may potentially improve training effects. The software also supports several standard BCI paradigms. To interface with external data acquisition devices NFBLab employs Lab Streaming Layer protocol supported by the majority of EEG vendors. MEG devices are interfaced through the Fieldtrip buffer.

Sensing and Connection Systems for Assisted and Autonomous Driving and Unmanned Vehicles

The special issue, “Sensors, Wireless Connectivity and Systems for Autonomous Vehicles and Smart Mobility” on MDPI Sensors presents 12 accepted papers, with authors from North America, Asia, Europe and Australia, related to the emerging trends in sensing and navigation systems (i.e., sensors plus related signal processing and understanding techniques in multi-agent and cooperating scenarios) for autonomous vehicles, including also unmanned aerial and underwater ones.

A Fine-Grained and Privacy-Preserving Query Scheme for Fog Computing-Enhanced Location-Based Service

Location-based services (LBS), as one of the most popular location-awareness applications, has been further developed to achieve low-latency with the assistance of fog computing. However, privacy issues remain a research challenge in the context of fog computing. Therefore, in this paper, we present a fine-grained and privacy-preserving query scheme for fog computing-enhanced location-based services, hereafter referred to as FGPQ. In particular, mobile users can obtain the fine-grained searching result satisfying not only the given spatial range but also the searching content. Detailed privacy analysis shows that our proposed scheme indeed achieves the privacy preservation for the LBS provider and mobile users. In addition, extensive performance analyses and experiments demonstrate that the FGPQ scheme can significantly reduce computational and communication overheads and ensure the low-latency, which outperforms existing state-of-the art schemes. Hence, our proposed scheme is more suitable for real-time LBS searching.

A Type of Low-Latency Data Gathering Method with Multi-Sink for Sensor Networks

To balance energy consumption and reduce latency on data transmission in Wireless Sensor Networks (WSNs), a type of low-latency data gathering method with multi-Sink (LDGM for short) is proposed in this paper. The network is divided into several virtual regions consisting of three or less data gathering units and the leader of each region is selected according to its residual energy as well as distance to all of the other nodes. Only the leaders in each region need to communicate with the mobile Sinks which have effectively reduced energy consumption and the end-to-end delay. Moreover, with the help of the sleep scheduling and the sensing radius adjustment strategies, redundancy in network coverage could also be effectively reduced. Simulation results show that LDGM is energy efficient in comparison with MST as well as MWST and its time efficiency on data collection is higher than one Sink based data gathering methods.