Search

Helical computed tomography (CT) scans are often performed to obtain three-dimensional images of an object that is longer than the detector. However, the existing quasi-exact and exact reconstruction methods, such as re-binning and Katsevich algorithm, generate interpolation errors or require high computational power. In this work, we propose a method to reconstruct helical CT projections by iteratively reducing helical artifacts. In each iteration, a convolutional neural network (CNN)-based denoising technique is used to accurately segment the prior image (bone and soft tissue image). The results indicate that the proposed algorithm reduces helical artifacts to a significantly greater extent than the existing single slice re-binning (SSR) and weighted filtered backprojection (W-FBP) methods.

Previous studies implicate extracellular adenosine signaling in attenuating myocardial ischemia and reperfusion injury (IRI). This extracellular adenosine signaling is terminated by its uptake into cells by equilibrative nucleoside transporters (ENTs). Thus, we hypothesized that targeting ENTs would function to increase cardiac adenosine signaling and concomitant cardioprotection against IRI. Mice were exposed to myocardial ischemia and reperfusion injury. Myocardial injury was attenuated in mice treated with the nonspecific ENT inhibitor dipyridamole. A comparison of mice with global Ent1 or Ent2 deletion showed cardioprotection only in Ent1–/– mice. Moreover, studies with tissue-specific Ent deletion revealed that mice with myocyte-specific Ent1 deletion (Ent1loxP/loxP Myosin Cre+ mice) experienced smaller infarct sizes. Measurements of cardiac adenosine levels demonstrated that postischemic elevations of adenosine persisted during reperfusion after targeting ENTs. Finally, studies in mice with global or myeloid-specific deletion of the Adora2b adenosine receptor (Adora2bloxP/loxP LysM Cre+ mice) implied that Adora2b signaling on myeloid-inflammatory cells in cardioprotection provided by ENT inhibition. These studies reveal a previously unrecognized role for myocyte-specific ENT1 in cardioprotection by enhancing myeloid-dependent Adora2b signaling during reperfusion. Extension of these findings implicates adenosine transporter inhibitors in cardioprotection against ischemia and reperfusion injury.

In a channel state information (CSI) based indoor positioning system, the positioning performance becomes susceptible to multipath fading effects especially in non-line-of-sight environments. We propose a transformer-based indoor positioning system (TIPS) to address this challenge. The proposed TIPS utilizes a self-attention mechanism to process the continuous WiFi CSI observed from predetermined routes as fingerprints in a given indoor environment. Each route is then considered a sentence, whereas the position along the route is treated as a word in terms of natural language processing. Consequently, the problem of predicting the position with the fingerprints can then be considered the task of predicting the current word with previous words, which can be efficiently solved using the proposed TIPS. In order to fully exploit the relations among positions, we propose embedding the information of the direction of arrival (DoA) on top of the collected CSI as inputs to the TIPS. Thus, the transformer of the proposed TIPS can better capture the dependencies of the positions in the route and significantly boost positioning accuracy. To exhibit the superiority of the proposed TIPS in a radio frequency (RF) environment, we demonstrate a hardware implementation of an RF testbed consisting of an emulator of WiFi access point and user equipment. Through extensive computer simulations and experimental tests, it is demonstrated that the proposed TIPS can reduce the positioning error down to 20 cm, which is a significant improvement compared to the current state-of-the-art models.

An integration scheme for sideslip angle estimation is proposed where a deep neural network and a simple kinematics-based model are combined in an unscented Kalman filter. The deep neural network contains two modules: a sensor filtering network designed to overcome the limitations of the kinematics-based model and a deep ensemble network to estimate the sideslip angle and its uncertainty. Both networks use recurrent neural networks with long short-term memory to analyze sequential sensor data. The networks were trained using only input signal sets that can be obtained from on- board sensor measurements. The filtering network reduces the noise and bias of the input signals to match the model used for the unscented Kalman filter. Next, the initial estimate and its uncertainty obtained from the deep ensemble network are utilized as a new measurement in the unscented Kalman filter, inducing an adaptive measurement variance. The algorithm was verified through both simulation and experiment, and the results demonstrate the effectiveness of the proposed algorithm.

Trajectory prediction of the ego vehicle is essential for advanced driver assistance systems to function properly. By recognizing various driving styles and predicting trajectories reflecting them, the prediction performance is enhanced, and a personalized trajectory can be generated. Therefore, we propose to combine driving style recognition and trajectory prediction tasks using only in-vehicle CAN-bus sensor data for possible application to normal vehicles. The DeepConvLstm network was utilized for driving style recognition, and a generative-based model was used for trajectory prediction. The classified driving style was added as a condition to the network. In addition, the past trajectory of the ego vehicle is estimated and utilized as an additional input for performance improvement. The performance of the proposed method is analyzed in terms of the root mean squared error (RMSE) and mean absolute error (MAE) compared with the case wherein the driving style and the past trajectory are not conditioned or given, respectively. The results demonstrate the effectiveness of the proposed method.

In state-of-the-art mobile devices (MDs), non-radio contexts (i.e., non-real-time data) provided by various sensors have been excluded in determining the modem configuration because of the problems in applying them in real-time modem operations. This paper presents a reconfiguration procedure and hardware implementation of a European Telecommunications Standards Institute (ETSI)-standard reconfigurable MD with a multi-antenna system. Borrowing the ETSI-standard architecture, the proposed MD can exploit a high-level abstraction of configuration-related operations, which consequently provides its application processor operating in non-real-time domain a uniform way of managing the multi-antenna-related hardware platform operating in real time. The configuration of the proposed MD is determined in accordance with non-radio and radio context information for opportunistically utilizing multiple antenna resources. From an implemented multi-antenna MD system operating in an indoor layout and computer simulations of the multi-antenna MD operating in a modeled outdoor layout, we verify that the MD configuration can be optimized in accordance with radio and non-radio contexts not only for fixed indoor but also for mobile outdoor environments. This enables the multi-antenna MD to efficiently utilize its antenna resources, such that the throughput maximization and/or power consumption minimization can be achieved.

This paper addresses how to achieve efficient programmability and software portability in the data plane of a software-defined radio access network (SDRAN). We assume a cloud RAN environment that builds on multi-vendor hardware components. Recent literature on SDRAN data plane indicates that software portability remains an issue in terms of efficient execution of software, even if the software is abstracted from the underlying hardware. In addition, software interfaces typically vary across different hardware components in the SDRAN data plane, leading to platform-dependent software management. Generalizing the European Telecommunications Standards Institute approach for a mobile device architecture, this paper presents a novel SDRAN data plane framework, providing efficient hardware platform-independent programmability and software portability. First, to resolve the software portability issue, the proposed data plane framework employs a specific (radio) virtual machine as well as a radio library; the heterogeneous hardware platforms are abstracted, enabling the joint optimization of the radio application code and hardware platform. Second, to achieve platform-independent software management, the proposed data plane framework adopts a double-layered structure enabling users to exploit high-level software management for the SDRAN data plane. Third, the feasibility of the proposed data plane framework is verified through a proof-of-concept (PoC) system with the proposed double-layered structure. Based on this PoC system, we show that users can efficiently perform software management. According to the numerical results obtained from the PoC system, the proposed double-layered structure introduces negligible additional footprint in terms of computational resources, memory requirements, and latency.

We propose a method for minimizing global buffer access within a deep learning accelerator for convolution operations by maximizing the data reuse through a local register file, thereby substituting the local register file access for the power-hungry global buffer access. To fully exploit the merits of data reuse, this study proposes a rearrangement of the computational sequence in a deep learning accelerator. Once input data are read from the global buffer, repeatedly reading the same data is performed only through the local register file, saving significant power consumption. Furthermore, different from prior works that equip local register files in each computation unit, the proposed method enables sharing a local register file along the column of the 2D computation array, saving resources and controlling overhead. The proposed accelerator is implemented on an off-the-shelf field-programmable gate array to verify the functionality and resource utilization. Then, the performance improvement of the proposed method is demonstrated relative to popular deep learning accelerators. Our evaluation indicates that the proposed deep learning accelerator reduces the number of global-buffer accesses to nearly 86.8%, consequently saving up to 72.3% of the power consumption for the input data memory access with a minor increase in resource usage compared to a conventional deep learning accelerator.

Several sinogram inpainting based metal artifact reduction (MAR) methods have been proposed to reduce metal artifact in CT imaging. The sinogram inpainting method treats metal trace regions as missing data and estimates the missing information. However, a general assumption with these methods is that data truncation does not occur and that all metal objects still reside within the field-of-view (FOV). These assumptions are usually violated when the FOV is smaller than the object. Thus, existing inpainting based MAR methods are not effective. In this paper, we propose a new MAR method to effectively reduce metal artifact in the presence of data truncation. The main principle of the proposed method involves using a newly synthesized sinogram instead of the originally measured sinogram. The initial reconstruction step involves obtaining a small FOV image with the truncation artifact removed. The final step is to conduct sinogram inpainting based MAR methods, i.e., linear and normalized MAR methods, on the synthesized sinogram from the previous step. The proposed method was verified for extended cardiac-torso simulations, clinical data, and experimental data, and its performance was quantitatively compared with those of previous methods (i.e., linear and normalized MAR methods directly applied to the originally measured sinogram data). The effectiveness of the proposed method was further demonstrated by reducing the residual metal artifact that were present in the reconstructed images obtained using the previous method.

Partial sensing is used to reduce the power consumption of pedestrian user equipment (P-UE) that operates in the signal environment of a mode-2 sidelink. However, because the data trans-mission is allowed only for the window duration of each corresponding P-UE, the throughput of the P-UE decreases by the ratio between the width of the window and the entire data period. This paper presents a novel method for enhancing the throughput of the P-UE that operates with partial sensing in the mode-2 sidelink. The proposed technique employs an additional UE, denoted the roadside unit (RSU), to collect the sensing results from each P-UE that operates with partial sensing. The proposed RSU sequentially aligns all of the partial sensing windows, such that the combination of each partial sensing window can eventually provide an almost complete sensing result. In this study, extensive computer simulations were performed. The results reveal that the proposed method enhances the throughput of each P-UE operating with partial sensing almost to that of full sensing without increasing the required power consumption.

In a Wi-Fi indoor positioning system (IPS), the performance of the IPS depends on the channel state information (CSI), which is often limited due to the multipath fading effect, especially in indoor environments involving multiple non-line-of-sight propagation paths. In this paper, we propose a novel IPS utilizing trajectory CSI observed from predetermined trajectories instead of the CSI collected at each stationary location; thus, the proposed method enables all the CSI along each route to be continuously encountered in the observation. Further, by using a generative adversarial network (GAN), which helps enlarge the training dataset, the cost of trajectory CSI collection can be significantly reduced. To fully exploit the trajectory CSI’s spatial and temporal information, the proposed IPS employs a deep learning network of a one-dimensional convolutional neural network–long short-term memory (1DCNN-LSTM). The proposed IPS was hardware-implemented, where digital signal processors and a universal software radio peripheral were used as a modem and radio frequency transceiver, respectively, for both access point and mobile device of Wi-Fi. We verified that the proposed IPS based on the trajectory CSI far outperforms the state-of-the-art IPS based on the CSI collected from stationary locations through extensive experimental tests and computer simulations.

This paper presents a novel calibration method that equalizes the impulse responses of all the Radio Frequency (RF) modules of an antenna array system operating in Long-Term Evolution (LTE) evolved NodeB (eNB). The proposed technique utilizes the Zadoff-Chu (Z-C) sequence of the Primary Synchronization Signal (PSS) and Sounding Reference Signal (SRS) that are available in every LTE data frame for downlink and uplink, respectively, for estimating and compensating the differences in the impulse responses among the RF modules. The proposed calibration method is suitable for wide bandwidth signal environments of LTE because it equalizes the impulse response of each RF module, which is ultimately equivalent to compensate the phase and amplitude differences among RF modules for the entire frequency band. In addition, the proposed method is applicable while the target eNB is transmitting or receiving a data stream. From various experimental tests obtained from a test-bed implemented with 2 RF modules, it has been verified that the proposed method provides a reliable calibration for Release 10 Time Division Duplex (TDD) LTE signals. Phase errors after the calibration in our test-bed have been found to be about 2.418° and 2.983° for downlink and uplink, respectively.

C. elegans undergoes periods of behavioral quiescence during larval molts (termed lethargus) and as adults. Little is known about the circuit mechanisms that establish these quiescent states. Lethargus and adult locomotion quiescence is dramatically reduced in mutants lacking the neuropeptide receptor NPR-1. Here, we show that the aroused locomotion of npr-1 mutants results from the exaggerated activity in multiple classes of sensory neurons, including nociceptive (ASH), touch sensitive (ALM and PLM), and stretch sensing (DVA) neurons. These sensory neurons accelerate locomotion via both neuropeptide and glutamate release. The relative contribution of these sensory neurons to arousal differs between larval molts and adults. Our results suggest that a broad network of sensory neurons dictates transitions between aroused and quiescent behavioral states.

We propose a novel precoding algorithm that is a zero-forcing (ZF) method combined with adaptive beamforming in the Worldwide Interoperability for Microwave Access (WiMAX) system. In a Multiuser Multiple-Input Multiple-Output (MU-MIMO) system, ZF is used to eliminate the Multiple Access Interference (MAI) in order to allow several users to share a common resource. The adaptive beamforming algorithm is used to achieve the desired SNR gain. The experimental system consists of a WiMAX base station that has 2 MIMO elements, each of which is composed of three-array antennas and two mobile terminals, each of which has a single antenna. Through computer simulations, we verified that the proposed method outperforms the conventional ZF method by at least 2.4 dB when the BER is 0.1%, or 1.7 dB when the FER is 1%, in terms of the SNR. Through a hardware implementation of the proposed method, we verified the feasibility of the proposed method for realizing a practical WiMAX base station to utilize the channel resources as efficiently as possible.

We present an implementation of a mobile-WiMAX (m-WiMAX) base station (BS) that supports smart antenna (SA) functionality. To implement the m-WiMAX SA BS, we must address a number of key issues in baseband signal processing related to symbol-timing acquisition, the beamforming scheme, and accurate calibration. We propose appropriate solutions and implement an m-WiMAX SA BS accordingly. Experimental tests were performed to verify the validity of the solutions. Results showed a 3.5-time (5.5 dB) link-budget enhancement on the uplink compared to a single antenna system. In addition, the experimental results were consistent with the results of the computer simulation.