Accurate disease interpretation from radiology remains challenging due to imaging heterogeneity. Achieving expert-level diagnostic decisions requires integration of subtle image features with clinical knowledge. Yet major vision-language models (VLMs) treat images as holistic entities and overlook fine-grained image details that are vital for disease diagnosis. Clinicians analyze images by utilizing their prior medical knowledge and identify anatomical structures as important region of interests (ROIs). Inspired from this human-centric workflow, we introduce Anatomy-VLM, a fine-grained, vision-language model that incorporates multi-scale information. First, we design a model encoder to localize key anatomical features from entire medical images. Second, these regions are enriched with structured knowledge for contextually-aware interpretation. Finally, the model encoder aligns multi-scale medical information to generate clinically-interpretable disease prediction. Anatomy-VLM achieves outstanding performance on both in- and out-of-distribution datasets. We also validate the performance of Anatomy-VLM on downstream image segmentation tasks, suggesting that its fine-grained alignment captures anatomical and pathology-related knowledge. Furthermore, the Anatomy-VLM’s encoder facilitates zero-shot anatomy-wise interpretation, providing its strong expert-level clinical interpretation capabilities.

Deep reinforcement learning has demonstrated outstanding performance in autonomous driving (AD). However, independent single vehicle training struggles to cope with complex traffic environments, while collaborative training across multiple vehicles causes security risk in data sharing. To address these challenges, this paper proposes a dual-layer blockchain-enabled federated reinforcement learning algorithm (DBFRL) for personalized AD. The proposed DBFRL algorithm constructs a FRL architecture based on a dual-layer blockchain to ensure data security during training. Practical driving behaviors data from the HighD dataset are used to classify driving styles into three categories: timid, normal and aggressive. Correspondingly, the personalized multi-objective reward functions are designed to reflect individual driving preferences. Then, the improved TD3 algorithm with different experience replay buffers and prioritized experience replay mechanism are using in the local model training. Furthermore, the reputation values of connected autonomous vehicles are introduced to ensure high-quality global model aggregation. The effectiveness of the DBFRL algorithm is validated on the CARLA simulator. Simulation results confirm that it significantly improves training performance and preserves data safety simultaneously.

This paper proposes a novel computing architecture for high-precision, low-latency, low-power, and cost-effective computation of complex numberNth roots andNth powers. By integrating the high-precision properties of the coordinate rotation digital computer (CORDIC) algorithm with the low-latency benefits of piecewise linear (PWL) approximation, the architecture leverages the binary logarithm-antilogarithm relationship to compute roots and powers of arbitrary complex numbers. Specifically, TheNth roots andNth powers of the modulus of the input complex number are computed using normalization preprocessing and PWL, and the conversion between the plane coordinate and polar coordinate forms of the complex number is achieved using the CORDIC algorithm. The design is implemented in Verilog HDL and synthesized using 40nm CMOS technology at a frequency of 1GHz. The synthesis results show that the area consumption for the complexNth root computation is$24193.92\mu $m2, with a power consumption of 2.1352mW. The area consumption for the complexNth power computation is$20836.83\mu $m2, with a power consumption of 1.8658mW. Compared to the latest complexNth root design, the proposed architecture reduces the area by 11.67% and power consumption by 9.33%. The average accuracy exhibits only a slight reduction compared to the state-of-the-art design while remaining at the same order of magnitude. Furthermore, the computation delay for theNth root architecture is only 58.46% of the delay of the latest complexNth root design, while the delay for theNth power architecture is 94.87% of the delay of the existing real-valuedNth power design.

In multiple-input multiple-output (MIMO) - orthogonal frequency division multiplexing (OFDM) based communication systems, traditional pilot allocation schemes used for channel estimation may not be optimal for target positioning. This limitation motivates us to design a novel allocation scheme that flexibly fulfills requirements in integrated sensing and communication (ISAC) implementations. To address this, we first establish a unified channel state information (CSI) based ISAC model for single-site indoor positioning in uplink MIMO-OFDM systems. To quantify the impact of resource elements (REs) allocated to pilots—across both subcarrier and OFDM symbol dimensions—on positioning performance, we derive the Cramér-Rao lower bounds (CRLBs) of the target parameters for single-site positioning. Subsequently, we jointly optimize the number of pilot REs in the subcarrier dimension and the OFDM symbol dimension to minimize the squared position error bound (SPEB), while satisfying the requirements for communication capacity and velocity estimation. For the formulated mixed integer nonlinear programming (MINLP) problem, we propose an algorithm based on sequential convex approximation (SCA) and penalty functions to convert the non-convex problem into a convex one for efficient solution. Simulation results demonstrate that the proposed algorithm achieves superior SPEB performance compared to benchmark schemes, thereby maximizing time-frequency resource utilization in single-site indoor positioning systems.

Medical image segmentation remains challenging due to the vast diversity of anatomical structures, imaging modalities, and segmentation tasks. While deep learning has made significant advances, current approaches struggle to generalize as they require task-specific training or fine-tuning on unseen classes. We present Iris, a novel In-context Reference Image guided Segmentation framework that enables flexible adaptation to novel tasks through the use of reference examples without fine-tuning. At its core, Iris features a lightweight context task encoding module that distills task-specific information from reference context image-label pairs. This rich context embedding information is used to guide the segmentation of target objects. By decoupling task encoding from inference, Iris supports diverse strategies from one-shot inference and context example ensemble to object-level context example retrieval and in-context tuning. Through comprehensive evaluation across twelve datasets, we demonstrate that Iris performs strongly compared to task-specific models on in-distribution tasks. On seven held-out datasets, Iris shows superior generalization to out-of-distribution data and unseen classes. Further, Iris’s task encoding module can automatically discover anatomical relationships across datasets and modalities, offering insights into medical objects without explicit anatomical supervision.

MOTIVATION: Gene regulation inference, a key challenge in systems biology, is crucial for understanding cell function, as it governs processes such as differentiation, cell state maintenance, signal transduction, and stress response. Leading methods utilize gene expression, chromatin accessibility, transcription factor (TF) DNA binding motifs, and prior knowledge. However, they overlook the fact that TFs must be in physical proximity to facilitate transcriptional gene regulation. RESULTS: To fill the gap, we develop GRIP-Gene Regulation Inference by considering TF Proximity-a gene regulation inference method that directly considers the physical proximity between regulating TFs. Specifically, we use the distance in a protein-protein interaction (PPI) network to estimate the physical proximity between TFs. We design a novel Boolean convex program, which can identify TFs that not only can explain the gene expression of target genes (TGs) but also stay close in the PPI network. We propose an efficient algorithm to solve the Boolean relaxation of the proposed model with a theoretical tightness guarantee. We compare our GRIP with state-of-the-art methods (SCENIC+, DirectNet, Pando, and CellOracle) on inferring cell-type-specific (CD4, CD8, and CD 14) gene regulation using the PBMC 3k scMultiome-seq data and demonstrate its out-performance in terms of the predictive power of the inferred TFs, the physical distance between the inferred TFs, and the agreement between the inferred gene regulation and PCHiC data. AVAILABILITY AND IMPLEMENTATION: https://github.com/EJIUB/GRIP.

As unmanned robotic operations become prevalent in various fields, the presence of moving objects in complex scenes poses challenges to key technologies, such as environment mapping, obstacle avoidance, and trajectory prediction. In this article, we propose FP-MOS, a LiDAR-based dynamic object segmentation framework that integrates prediction information to continuously capture dynamic object information under constrained data conditions. First, a prediction module (PM) is employed to obtain predicted range images, enhancing the continuity of temporal information in the point cloud within the segmentation network. This improved continuity helps mitigate misjudgment issues caused by dynamic object turns and retrieves previously missed moving objects. Additionally, a motion attention weight guidance module is introduced, which accurately captures dynamic point cloud features through the transmission of point cloud weights between adjacent frames. Finally, we apply an improved adaptive local outlier factor (ALOF) method to filter out outliers in the segmentation results. Experimental results on the SemanticKITTI-MOS and Apollo datasets show that our method achieves leading Intersection-over-Union (IoU) scores of 79.3% and 79.0%, demonstrating the effectiveness of the FP-MOS method in network design and data optimization.

With the continuous development of 6G communication technology and artificial intelligence (AI), reconfigurable intelligent surface (RIS) technology and generative AI (GAI) have received widespread attention. Hence, combining these two techniques to solve the problem of nonlocalizability in single access point (AP) scenarios is promising. This article proposes a novel RIS-aided localization scheme via generative artificial intelligence (GAI). Specifically, first, considering the presence of a certain amount of noise and the low discriminative features of the original collected received signal strength (RSS) brought by integrated reflective elements of RIS, we construct a generative adversarial network (GAN) model named variational autoencoder-convolutional neural network (VAE-CNN). It can effectively perform noise reduction in the data preprocessing stage to reduce unnecessary redundant information, and extract more discriminative features through convolutional networks to improve the differentiation between data. Second, the target’s location in the spatial domain is formulated as a sparse vector, and then a sparse recovery model is introduced to solve the location estimation problem in RIS-aided localization scenarios. Finally, considering the case that RIS has multiple reflective elements, which will lead to a high dimension of the measurement matrix in the sparse recovery model, we further apply the semi-tensor product (STP) sparse recovery theory on the model to reduce the storage space and high time consumption. Experimental results show that our proposed methods outperform the traditional approaches and reduce the computational complexity simultaneously.

Three-dimensional (3D) imaging in low-light environments has broad application prospects. High-quality 3D imaging requires obtaining both the depth and reflectivity information of the target. With the rapid development of Quantum Information Technology (QIT), Quantum Imaging (QI) provides a new approach to achieve high-quality 3D imaging. However, traditional QI methods are limited to Two-dimensional (2D) imaging and cannot accurately extract the depth information from the target. These methods also suffer from insufficient precision in reflectivity calculations. Therefore, we introduce a novel quantum 3D imaging and reflectivity estimation method based on entangled photons. Specifically, we first utilize the spatiotemporal correlation characteristics of entangled photon pairs combined with coincidence measurement technology to accurately determine target depth. Second, we analyze the distribution of entangled photon numbers during free-space transmission to construct a photon-based detection probability model. Third, we use this model to derive the probability density function of coincidence counts. Finally, we estimate the surface reflectivity information through statistical analysis of coincidence counting. Experimental results demonstrate that the proposed method can accurately reconstruct both depth and reflectivity profiles of the target in low-light environments. These findings validate the effectiveness of our method and highlight the potential of quantum 3D imaging for target detection in low-light environments.

As an important part of underwater optical technology, underwater imaging plays a crucial role in accurately capturing underwater targets and environmental features. Facing the challenges of complex ocean environments and severe photon attenuation, quantum imaging breaks through the limitations of traditional optical imaging technology by utilizing the characteristics of two-photon entanglement and time-space correlation, thus offering a new perspective on ocean turbulence. To this end, we propose a new underwater entangled-photon quantum imaging system. Specifically, we first construct an entangled photon quantum imaging physical model through ocean long-exposure turbulence and then exploit an entangled light coincidence imaging reconstruction method to image the target object. Furthermore, in response to the problem that ocean environment has a great impact on photons, we develop a photon capture probability method based on entangled photon pairs to reduce the impact of the ocean turbulence noise on imaging and improve the underwater target imaging resolution. We experimentally demonstrate the effectiveness of our method by showing that even in harsh ocean environments, quantum imaging performs superior resolution capabilities over traditional light source imaging techniques.

The frequency-modulated continuous wave (FMCW) radar has received much attention in the field of noncontact vital signs monitoring. However, since vital signs are usually very weak, it can be easily buried by interference and noise, especially for the heartbeat signal. To tackle this challenge, this article proposes a novel systematic vital signs detection framework using the multiple-input multiple-output FMCW radar. First, the signal noise ratio of the vital signs signal is enhanced by combining the phase signals of multiple channels using the maximum ratio combining method. Then, to suppress noise and interference, we construct the vital signs signal with singular spectral analysis and propose a correlation-based selection criterion to select potential intrinsic mode functions of the respiration and heartbeat signals. Finally, a fast independent component analysis is applied to extract the respiration signal, and the second-order derivative based fast independent component analysis in conjunction with an infinite impulse response notch filter is further developed to extract the heartbeat signal. Simulations and experimental results validate the effectiveness of the proposed framework.

Extended object detection and tracking (EODT) is becoming a promising alternative for autonomous perception, which provides not only common motion states but also accurate spatial extent information, such as shape and size estimations. However, due to uncoordinated radar transmissions in zero-trust autonomous driving scenarios, radar-based EODT systems suffer from mutual radio frequency (RF) interference launched by attackers, leading to ghost targets and increased noise. On this account, a novel joint anti-interference detection and tracking system for weak extended targets is presented in this paper. In contrast to pioneering works that treat object detection and tracking as two separate steps, the proposed method handles them jointly by integrating a continuous detection process into tracking, improving the detectability of weak targets. More specifically, to accommodate the time-varying number and extended size of radar reflections, an adaptive spatial distribution model representing the deformable extents is incorporated to capture the contour evolution over time. The key insight is that by accumulating the reflected power, all backscattered points are regarded as one entity to match the real target so that the intractable data association problem can be circumvented in the proposed method. Unlike the prominent random matrix model-based approaches that split motion and extent states into independent parts, this study explores the interdependencies between the states and updates them simultaneously. In addition, the proposed system has been deployed on a low-cost automotive radar platform. Experimental results confirm that the proposed approach can achieve accurate and resilient EODT against RF interference attacks, especially in occlusion, dynamic motion switching, and complex multiple extended target tracking scenarios. A demonstration video with EODT results is available in the supplementary materials.

With the increasing popularity of smartphones, smartphone photography has become convenient and common in daily life, thus making visual localization technology receive widespread attention. Due that monocular cameras are mostly used on smartphones, the depth information from a single image cannot be obtained, so the image-matching-based localization technique is widely adopted. However, this method has the problems of high time consumption and vulnerable to environmental interference. Therefore, this article proposes a low overhead and robust indoor image positioning method, which mainly consists of two modules. First, an improved siamese neural network framework is introduced to train the similarity metric model between images, which significantly reduces the workload of labeling for large amounts of sample data. Moreover, it improves the robustness when the target environment contains similar image features in the matching stage. Second, to further estimate the user’s fine location, an adaptive random sample consensus algorithm is proposed to optimize the fundamental matrix in the classical EightPoint method, which could efficiently eliminate the matching outliers to solve the camera attitude, and dynamically adapt to more complex and changeable data situations. A large number of experimental results show that the positioning performance of this scheme is better than traditional schemes, and the average positioning error of 0.50 m can be achieved.

Neural network positioning technology, as one of the mainstream in indoor Wi-Fi positioning systems, is playing an increasingly important role in location-based services. The main challenge is that the samples are prone to be outdated as the indoor environment changes or the wireless signal varies over time, i.e., the samples’ Age of Information (AoI) is large, which leads to the trained model not being available. However, recollecting data to retrain the model is both time-consuming and labor-intensive. To address the above problem, this article proposes a fast adaptation approach based on Bayesian meta-learning that makes the pretrained model acquire a learned learning capability so that it can quickly learn new tasks based on the acquisition of existing knowledge. Specifically, first, a model-agnostic learning scheme is introduced to guide the learning process, which could automatically learn the optimal model parameters and hyperparameter settings. Second, to mitigate the effects of model uncertainty, especially to prevent the overfitting situation based on a limited number of samples, we combine the Stein variational gradient descent (SVGD) with the model-agnostic learning scheme, i.e., Bayesian meta-learning. Compared with traditional meta-learning algorithms, the proposed method makes the training more robust by inferring the Bayesian posterior from a probabilistic perspective. Extensive experimental results show that the proposed approach effectively overcomes the impact of large AoI on localization performance while decreasing labor consumption significantly.

Precise positioning is a key factor and enabler technology for many use cases on intelligent transportation systems (ITS) and connected and automated vehicles (CAVs). Recently, the quantum positioning system (QPS) based on quantum ranging has emerged as a novel way to improve security and precision. As a key process of QPS, the entangled photons based ranging technology has picosecond-level clock synchronization, and the ranging accuracy can reach the Heisenberg limit. If promising QPS is deployed in the ITS and CAVs, it will cause a profound change. However, the existing QPS still lacks accuracy and robustness in different scenarios. To solve this problem, we proposed an efficient and robust fusion positioning system based on entangled photons. In this system, we derive the ranging accuracy limit with many factors and propose a fast data grouping and selection algorithm to improve real-time performance. Furthermore, we propose a fusion extended fingerprint localization method for robust positioning in the dynamic environment. The effectiveness and robustness of the system are verified by extensive experiments. When the range is 15m, the ranging accuracy can be limited to 0.0018m. The proposed system achieves the probability of positioning errors 90% within 0.13m with only two APs.

In the intelligent logistics and warehouses, the autonomous mobile device (AMD) holds a key position as it is equipped with the ability to carry out functions like material transportation and inventory inspection. Nevertheless, the effective execution of these functions necessitates the location of the AMD. Given the increasing proliferation of networks like WiFi and 5G, leveraging these signals to achieve AMD localization is a desirable solution. Therefore, this paper proposes a channel state information (CSI) based system forthrough-the-wall (TTW) passive AMDdetection andlocalization, named T-DeLo. T-DeLo first establishes a reference channel and utilizes it to cancel the strong signal interference (SSI) and phase errors, ensuring that the reflections introduced by the AMD can be estimated. Built upon this core, it uses the proposed novel two-dimensional matrix pencil algorithm to estimate jointly the path length change rate (PLCR) and time of flight (ToF) of the AMD induced reflections, in the TTW scenario. Unlike existing algorithms, this algorithm aggregates multiple measurements to improve the estimation performance under conditions of low signal-to-noise ratio (SNR). Finally, leveraging the estimated ToF and PLCR, T-DeLo realizes TTW AMD detection and localization via statistical and geometric analysis, respectively. In the TTW glass and brick wall scenarios, the extensive experimental evaluation shows that the AMD detection accuracy of T-DeLo is 0.964 and 0.952, while the median localization errors are 1.65 m and 2.05 m, respectively, laying a solid foundation for practical and ubiquitous AMD passive detection and localization.

In this letter, we propose an Integrated Sensing and Communications (ISAC) waveform design method based on the fusion of Orthogonal Time Frequency Space (OTFS) and Faster-Than-Nyquist (FTN). The objective is to maximize the communication data transmission rate while minimizing the sensing performance impact on the target parameter estimation. We first map the FTN symbols to OTFS waveform time domain for realizing symbol spacing compression and transmit them in time-varying channels. Then, an equalizer based on the Minimum Mean Square Error (MMSE) algorithm is used to eliminate the interference generated by the FTN. Simulation results show taking into account the system bit-error rate, the proposed method achieves an increase in the throughput as well as an improvement in the distance and velocity estimation of the target compared to the existing methods.

With the advent of the new era of 6G, new applications of smart factories and intelligent transportation systems based on real-time wireless sensing technology will confront great demands and challenges. In the intelligent transportation system, it is essential to realize services such as localization and intrusion detection for intelligent vehicles. To build a wide range of positioning network based on large-scale wireless networks, it is of great challenge to simultaneously solve the problem of unacceptable delay and bandwidth requirements caused by a large number of channel state information (CSI) data transmission. Therefore, we propose a novel algorithm, named PAOFIT, where a projection transformation aided CSI curve fitting compression algorithm is firstly proposed to decrease data distortions by improving the orthogonality of signal subspace and noise subspace, and an adaptive weighted average fitting order judgment algorithm is proposed to calculate the fitting order needed in the curve fitting process. Then, localization parameter, time of flight (ToF) are estimated by CSI reconstruction and parameter estimation. Finally, the location of the target is obtained by substituting these parameters into time difference of arrival (TDoA) wireless localization technology. Extensive experimental results verify that, compared with the existing compression algorithms, the proposed PAOFIT has a better performance in terms of compression ratio, median positioning error, residual and execution time.

Due to the small radar cross section (RCS), pedestrians are typical low-observable traffic participants for radar-based automotive perception systems. The early detection and understanding of pedestrians’ activities are of great significance to automotive safety. To this end, this paper presents an end-to-end joint tracking and activity estimation (RETA) system based on 4D automotive radar, which deals in particular with pedestrian activity identification under cluttered real-world scenes. Firstly, a novel integrated detection and tracking algorithm is proposed to guarantee positioning accuracy, in which all unthresholded 4D radar measurements are incorporated to explore the spatial coherent information across multiple frames, avoiding weak target information loss. After that, to discriminate continuous activities with varying durations in sequential trajectories, this paper innovatively presents a decomposed connectionist recurrent convolutional neural network, which facilitates fused temporal-spatial motion feature extraction. Especially, the labor-consuming activity pre-segmentation problem is circumvented with the help of a connectionist temporal classification algorithm in the proposed neural network. At last, RETA can be implemented for real end-to-end perception applications. Extensive experiment results highlight its superiority and effectiveness by attaining a continuous recognition accuracy of 94.8%. To the best of our knowledge, this is the first end-to-end activity recognition system specific for low-observable pedestrians. A demonstration video recorded in challenging practical traffic scenarios has been uploaded in the supplementary materials.

As indoor applications grow in diversity, wireless sensing, vital in areas like localization and activity recognition, is attracting renewed interest. Indoor wireless sensing relies on signal processing, particularly channel state information (CSI) based signal parameter estimation. Nonetheless, regarding reflected signals induced by dynamic human targets, no satisfactory algorithm yet exists for estimating the acceleration of dynamic path length change (DPLC), which is crucial for various sensing tasks in this context. Hence, this paper proposes DP-AcE, a CSI based DPLC acceleration estimation algorithm. We first model the relationship between the phase difference of adjacent CSI measurements and the DPLC’s acceleration. Unlike existing works assuming constant speed, DP-AcE considers both speed and acceleration, yielding a more accurate and objective representation. Using this relationship, an algorithm combining scaling with Fourier transform is proposed to realize acceleration estimation. We evaluate DP-AcE via the acceleration estimation and acceleration-based fall detection with the collected CSI. Experimental results reveal that, using distance as the metric, DP-AcE achieves a median acceleration estimation percentage error of 4.38%. Furthermore, in multi-target scenarios, the fall detection achieves an average true positive rate of 89.56% and a false positive rate of 11.78%, demonstrating its importance in enhancing indoor wireless sensing capabilities.

Frequent interactions between individuals are a fundamental challenge for pose estimation algorithms. Current pipelines either use an object detector together with a pose estimator (top-down approach), or localize all body parts first and then link them to predict the pose of individuals (bottom-up). Yet, when individuals closely interact, top-down methods are ill-defined due to overlapping individuals, and bottom-up methods often falsely infer connections to distant bodyparts. Thus, we propose a novel pipeline called bottom-up conditioned top-down pose estimation (BUCTD) that combines the strengths of bottomup and top-down methods. Specifically, we propose to use a bottom-up model as the detector, which in addition to an estimated bounding box provides a pose proposal that is fed as condition to an attention-based top-down model. We demonstrate the performance and efficiency of our approach on animal and human pose estimation benchmarks. On CrowdPose and OCHuman, we outperform previous state-of-the-art models by a significant margin. We achieve 78.5 AP on CrowdPose and 48.5 AP on OCHuman, an improvement of 8.6% and 7.8% over the prior art, respectively. Furthermore, we show that our method strongly improves the performance on multi-animal benchmarks involving fish and monkeys. The code is available at https://github.com/amathislab/BUCTD

Entangled photon quantum imaging technology has broad application prospects in ocean imaging due to its high resolution, anti-turbulence and anti-interference characteristics. Turbulence and optical attenuation caused by seawater will seriously limit the traditional imaging system, and quantum imaging based on entangled light source can overcome the limitations of low imaging resolution, large impact on marine environment and low confidentiality in traditional underwater target imaging. Based on the propagation characteristics of weak turbulent environment, we propose a new photon capture probability algorithm for marine environment based on entangled photon pairs. Finally, the relevant experimental environment is constructed according to the designed imaging optical path. We conduct extensive experiments to evaluate the effectiveness of our proposed quantum imaging method.

The process of quantifying and analyzing animal behavior involves translating the naturally occurring descriptive language of their actions into machine-readable code. Yet, codifying behavior analysis is often challenging without deep understanding of animal behavior and technical machine learning knowledge. To limit this gap, we introduce AmadeusGPT: a natural language interface that turns natural language descriptions of behaviors into machine-executable code. Large-language models (LLMs) such as GPT3.5 and GPT4 allow for interactive language-based queries that are potentially well suited for making interactive behavior analysis. However, the comprehension capability of these LLMs is limited by the context window size, which prevents it from remembering distant conversations. To overcome the context window limitation, we implement a novel dual-memory mechanism to allow communication between short-term and long-term memory using symbols as context pointers for retrieval and saving. Concretely, users directly use language-based definitions of behavior and our augmented GPT develops code based on the core AmadeusGPT API, which contains machine learning, computer vision, spatio-temporal reasoning, and visualization modules. Users then can interactively refine results, and seamlessly add new behavioral modules as needed. We used the MABe 2022 behavior challenge tasks to benchmark AmadeusGPT and show excellent performance. Note, an end-user would not need to write any code to achieve this. Thus, collectively AmadeusGPT presents a novel way to merge deep biological knowledge, large-language models, and core computer vision modules into a more naturally intelligent system. Code and demos can be found at: https://github.com/AdaptiveMotorControlLab/AmadeusGPT

Deciphering 3D genome conformation is important for understanding gene regulation and cellular function at a spatial level. The recent advances of single cell Hi-C technologies have enabled the profiling of the 3D architecture of DNA within individual cell, which allows us to study the cell-to-cell variability of 3D chromatin organization. Computational approaches are in urgent need to comprehensively analyze the sparse and heterogeneous single cell Hi-C data. Here, we proposed scDEC-Hi-C, a new framework for single cell Hi-C analysis with deep generative neural networks. scDEC-Hi-C outperforms existing methods in terms of single cell Hi-C data clustering and imputation. Moreover, the generative power of scDEC-Hi-C could help unveil the differences of chromatin architecture across cell types. We expect that scDEC-Hi-C could shed light on deepening our understanding of the complex mechanism underlying the formation of chromatin contacts.

Industrial Internet of Things (IIoT) usually utilizes 2.4-GHz unlicensed frequency band, which is also heavily used by many other communication systems, such as ZigBee, WiFi, Bluetooth, etc. Therefore, the lack of spectrum resources has become a key technical bottleneck to restrict the development of IIoT. Integrating cognitive radio (CR) into IIoT, Cognitive IIoT (CIIoT) can cope with the spectrum resource shortage by accessing the frequency bands licensed to primary user (PU). However, spectrum sensing and access control must be performed to avoid bringing severe interference to the PU. In this article, an integrated cooperative spectrum sensing (CSS) and access control model is proposed to improve the transmission performance of the CIIoT while guaranteeing the CSS’s detection probability and controlling the interference to the PU. This model is optimized to maximize the total throughput of IIoT in each frame by jointly optimizing sensing time, the number of sensing nodes and the transmit power for each node under the constraints of the minimum detection probability, the total power control, the interference control, and the minimum rate for each node. The optimization problem is solved by the joint optimization of spectrum sensing and access control. A simultaneous CSS and access control model is also proposed to increase the communication time by using one time slot to perform CSS and access control simultaneously. The simulation results show that there exist optimal sensing and control parameters to maximize the total throughput of CIIoT.

The sparsity of the localization problem makes the compression sensing (CS) theory suitable for indoor localization in wireless local area networks (WLANs). However, in practice, we find that the location errors and computing complexity increase significantly as the dimensionality of the sparse vector and measurement matrix are high in a large environment, so most CS-based localization techniques are accompanied by coarse localization and access point (AP) selection stages. Therefore, in this article, we first deduced the relationship between the number of APs and the dimensionality of the sparse vector theoretically to give the guideline that the number of subdatabases and APs should be obtained. Then an adaptive intuitionistic fuzzy C-ordered mean (AIFCOM) clustering is designed for the data with outliers in the environment with multipath effects. Finally, in the fine localization stage, we propose a semi-tensor product CS (STP-CS) model to construct the measurement matrix, compared with the traditional CS model, our model not only remains more number of APs, but also decreases the dimensionality of measurement matrix, which can reduce the storage space and improve localization accuracy simultaneously.

Robust pedestrian trajectory-tracking is an essential prerequisite to traffic accident prevention. However, it is a challenging task in urban autonomous driving, since the weak backscattered signals from pedestrians with small radar cross-section may be submerged in strong background clutters, especially under adverse weather conditions. On this account, this article presents an integration of detection and tracking (iDT) toward multipedestrian with a low signal-to-noise ratio (SNR). In particular, in contrast to conventional methods, in which the detection and tracking are treated as two separate processes, we address them jointly to ensure the accuracy of continuous detection and tracking in low SNR conditions. Another distinguishing element is that to accommodate the time-varying number of targets, the Bayesian framework is tailored by augmenting the state vector with a multipedestrian evolutional indicator. The advantage is that all targets can be tracked simultaneously by searching the global likelihood ratio of a spectrum once, rather than assigning an individual tracker to each target in conventional methods. Furthermore, through the proposed integrated framework, the data association problem is circumvented because there is no explicit measurement-target assignment process in our approach. In addition, a commercial automotive multiple-input-multiple-output millimeter-wave radar sensor is employed to validate the proposed method. Consequently, numerous simulation and experiment results turn out that iDT shows unique advantages in low-observable multipedestrian tracking compared with traditional methods.

Some published WiFi-based passive human tracking systems have achieved sub-meter accuracy. However, they require the location of WiFi devices to be known in advance for passive human tracking, which limits their application in practical indoor scenarios. In this paper, we propose WiSen, a novel passive human tracking system using one pair of commodity WiFi devices with unknown locations. First, we introduce a signal power model for human-related signal extraction and multi-dimensional parameter estimation. Due to low-resolution parameter estimates and noise, we further design a confidence-aware-based path pruning method that combines the distribution of path parameters from successive windows to select reliable paths of interest. Before that, we adopt a data augmentation method to increase the number of available paths to learn parameter distributions better. Then, we statistically estimate the transmitter's location using a kernel density estimation method and ultimately yield the user's location using an improved Gaussian Sum filter approach. We validate the performance of WiSen in real-life indoor environments. The experimental results show that WiSen can realize the sub-meter level accuracy for passive human tracking and device localization.

Wireless sensing attracts significant attention in recent years, due to its ubiquitous nature, individual privacy-preserving ability, and potential for future applications, such as home security and the Internet of Things (IoT). Among the sensed information, the dynamic human target acceleration, which can be used for gait analysis and passive localization, plays an irreplaceable role. Considering the target velocity is composed of initial velocity and acceleration, this paper proposes a model to describe the relationship between the acceleration of the dynamic target and the CSI phase change between adjacent CSI packets. Based on the proposed model, the target acceleration can be estimated directly from CSI via the fractional Fourier transform, which is essentially different from the existing algorithm that derives the acceleration from target velocity. The real-world evaluation shows that the median acceleration estimation error can reach about 0.69 m/s2, verifying the effectiveness of the proposed model.

Abstract Background Drug resistance is a critical obstacle in cancer therapy. Discovering cancer drug response is important to improve anti-cancer drug treatment and guide anti-cancer drug design. Abundant genomic and drug response resources of cancer cell lines provide unprecedented opportunities for such study. However, cancer cell lines cannot fully reflect heterogeneous tumor microenvironments. Transferring knowledge studied from in vitro cell lines to single-cell and clinical data will be a promising direction to better understand drug resistance. Most current studies include single nucleotide variants (SNV) as features and focus on improving predictive ability of cancer drug response on cell lines. However, obtaining accurate SNVs from clinical tumor samples and single-cell data is not reliable. This makes it difficult to generalize such SNV-based models to clinical tumor data or single-cell level studies in the future. Results We present a new method, DualGCN, a unified Dual Graph Convolutional Network model to predict cancer drug response. DualGCN encodes both chemical structures of drugs and omics data of biological samples using graph convolutional networks. Then the two embeddings are fed into a multilayer perceptron to predict drug response. DualGCN incorporates prior knowledge on cancer-related genes and protein–protein interactions, and outperforms most state-of-the-art methods while avoiding using large-scale SNV data. Conclusions The proposed method outperforms most state-of-the-art methods in predicting cancer drug response without the use of large-scale SNV data. These favorable results indicate its potential to be extended to clinical and single-cell tumor samples and advancements in precision medicine.

As one of the key technologies of the 5G, the device-to-device (D2D) can realize communication between terminals without any base station, thus achieving more convenience of cooperative positioning. In this article, we propose a D2D cooperative positioning approach via matrix completion and anchor selection, which tackles the positioning problem with inadequate distance information. Specifically, first, an incomplete Euclidean distance matrix (EDM) is constructed by using insufficient distance information between nodes, and then the singular value thresholding (SVT) algorithm is used to recovery this EDM to obtain completed information. Second, multidimensional scaling (MDS) is performed to reduce dimensions of recovered EDM, which aims to obtain the relative positions of nodes while maintaining the distance relationship among them. Third, a set of suitable anchor nodes is selected by using the Hodges–Lehmann (HL) test for position transformation. Finally, we apply the procrustes analysis (PA) to transform the relative positions to the global ones according to the selected set of suitable anchor nodes. From the extensive experimental results, it is evident that the proposed approach has high positioning accuracy even when a large proportion of elements are missing in the EDM.

With the development of the radar sensing technology, hand gesture sensing and recognition has attracted much attention. This letter adopts a frequency-modulated continuous wave (FMCW) radar to achieve short-range hand gesture sensing and recognition. Specifically, the range, Doppler, and angle parameters of hand gestures are measured by fast Fourier transformation (FFT) and multiple signal classification (MUSIC) algorithm, respectively. The mixup (MP) algorithm combined with augmentation (AU) algorithm using a weight factor is applied to expand the hand gesture data. Then, a complementary multidimensional feature fusion network-based hand gesture recognition (CMFF-HGR) is designed to extract the features and achieve HGR. Finally, a series of experiments are carried out to verify the effectiveness of the proposed approach, and the results show that the recognition accuracy is higher than the existing alternatives with low computational complexity.

Different from the social network which only focuses on the interaction between people, the integration of social networks and the internet of things (IoT) leads to multi-directional interactions of human to human, human to thing, and thing to thing. The social IoT is composed of a large number of heterogeneous devices, which can improve the scalability of resource and service. Meanwhile, the heterogeneous devices have uneven computing power and different location information measurement types (e.g., the distance, angle, and hop count). Therefore, a localization approach with easy-to-obtain measurement data and low requirements on the computing power is needed. In this article, we propose a localization approach for the social IoT by combining the fuzzy rough set theory and the ridge regression extreme learning machine (RRELM). First of all, a location fingerprint database is constructed. Different from the traditional location fingerprint database, the location fingerprint database here stores the minimum hop counts between the reference node (RN) and the anchor node (AN) instead of the received signal strength (RSS). Second, the fuzzy rough set theory is used to compute the significant degree of each AN, and the ANs that contribute little to the positioning result are removed. This approach not only relieves the storage pressure of the location fingerprint database but also reduces the computational complexity of user position estimation. Third, the RRELM is trained by using the samples in the location fingerprint database. Finally, by inputting the newly collected minimum hop counts from the user to each AN into the trained RRELM, the user’s position is estimated. From the extensive experimental results, the proposed approach has high positioning accuracy and low computational complexity, which is suitable for the social IoT.

The cognitive industrial Internet of Things (CIIoT) can improve transmission performance by utilizing the spectrum licensed to a primary user (PU), providing that the normal communication of the PU is not disturbed. However, the traditional spectrum access schemes for the CIIoT are difficult to adapt to the various communication environments. In this article,$Q$-learning-based dynamic spectrum access is proposed for the CIIoT to intelligently utilize the spectrum resources in three access scenarios: orthogonal multiple access (OMA), underlay spectrum access, and nonorthogonal multiple access (NOMA). In the OMA scheme, the CIIoT learns to access the idle channels to avoid distributing the PUs, but its communication continuity cannot be guaranteed when most of the channels are occupied by the PUs. In the underlay scheme, the CIIoT learns to utilize the busy channels to ensure the communication continuity by limiting its transmit power within the tolerance of the PU. However, the interference to the PU cannot be eliminated, which will decrease the PU’s throughput. In the NOMA scheme, however, the CIIoT can utilize the busy channels by canceling the interference to the PU with successive interference cancellation, which will guarantee the transmission performance of both the CIIoT and the PU. A$Q$-learning-based spectrum access algorithm is proposed to improve the transmission performance of the CIIoT in the three schemes. The simulation results have shown the advantages of the$Q$-learning-based NOMA scheme in terms of guaranteeing the throughput of the CIIoT nodes and decreasing the interference to the PUs.

In the Internet of Things (IoT), with the development of channel state information (CSI) based wireless sensing technologies, such as intrusion detection, activity recognition, and indoor localization, a tremendous amount of CSI data need to be transmitted simultaneously, resulting in unacceptable time delay and requirement for additional bandwidth. Therefore, this paper proposes a CSI compression algorithm based on projection transformation and curve fitting (PCFIT), which realizes high-compression-ratio transmission and high-accuracy reconstruction. Concretely, projection matrixes are firstly constructed to perform projection transformation on the original CSI. Next, an adaptive weighted average fitting order judgment algorithm is designed to get the fitting order. And then, the Levenberg-Marquardt (LM) algorithm is used to perform curve fitting on the CSI based on the linear combination of a few sine waves to obtain the fitting parameters. Finally, CSI is reconstructed with these fitting parameters. By analyzing compression indicators such as residual, compression ratio, and parameter estimation accuracy, experimental results show that the PCFIT compression algorithm achieves a better compression ratio than other existing compression algorithms under the same residual and parameter estimation accuracy.

Wi-Fi based passive human detection has attracted numerous research interests recently. For the real-world application, however, the human detection under the through-the-wall (TTW) scenario needs to be addressed. In this paper, we consider the signal spatial distribution from a statistical perspective and propose TWPad, a unified scheme for TTW stationary and moving human detection based on Wi-Fi channel state information (CSI). Specifically, TWPad first extracts the angle of arrival (AoA) of the multipath signals under the TTW scenario and conduct the Jarque-Bera (JB) test on AoA to analyze the normality of the signal spatial distribution. Then, a novel joint Mann-Whitney U (for non-normal distribution) and T-test (for normal distribution) hypothesis test algorithm is proposed to monitor changes in signal spatial distribution. By doing this, TWPad can capture the disturbance in the spatial distribution of multipath signals caused by the moving or stationary human and realize detection under the TTW scenario. The experimental evaluation shows that the TWPad’s F1-measure of stationary human detection can reach about 0.975 and 0.967, under the TTW scenario of glass and brick wall, respectively, outperforming the state-of-the-art solutions and shedding promising lights on ubiquitous human detection in practice.

Low-dose CT has been a key diagnostic imaging modality to reduce the potential risk of radiation overdose to patient health. Despite recent advances, CNN-based approaches typically apply filters in a spatially invariant way and adopt similar pixel-level losses, which treat all regions of the CT image equally and can be inefficient when fine-grained structures coexist with non-uniformly distributed noises. To address this issue, we propose a Structure-preserving Kernel Prediction Network (StructKPN) that combines the kernel prediction network with a structure-aware loss function that utilizes the pixel gradient statistics and guides the model towards spatially-variant filters that enhance noise removal, prevent over-smoothing and preserve detailed structures for different regions in CT imaging. Extensive experiments demonstrated that our approach achieved superior performance on both synthetic and non-synthetic datasets, and better preserves structures that are highly desired in clinical screening and low-dose protocol optimization.

With the development of wireless sensing technologies, position-based services are widely applied. The amount of channel state information (CSI) data increases sharply in the large-scale wireless network system, and the cost of storage and transmission of CSI becomes more extensive, which will affect the CSI transmission rate and the real-time performance of the target location. Therefore, this paper constructs an indoor positioning system based on CSI compression and reconstruction, realizing CSI compression transmission by the curve fitting algorithm and analyzing the compression and positioning performance. Firstly, the original CSI data is collected and preprocessed. Then the CSI is compressed and reconstructed by the curve fitting algorithm. Next, positioning parameters are estimated by parameter estimation algorithms. Finally, the time difference of arrival (TDoA) algorithm is applied to locate the target. There are CSI compression indexes, such as the residual and compression ratio. The experimental results show that the compression ratio based on the curve fitting algorithm is considerable, up to 22:1, when the residual error is the same. Additionally, the positioning accuracy is guaranteed compared with other algorithms that the positioning error by the curve fitting algorithm is about 1.326m. Furthermore, the experimental results also verify the effectiveness of the CSI compression algorithm based on curve fitting and the feasibility of an indoor positioning system based on CSI compression reconstruction.

With the in-depth development of the Internet of Things (IoT) and the constant discussion of 6G visions, the future development of indoor location-based services (LBSs) in the 6G-enabled IoT is attracting people’s attention. In fact, due to the continuous updating of network techniques, the complexity of the indoor environment and the number of connected wireless access points (APs) will increase dramatically, which leads to the diversity of the received signal strength (RSS) on the performance of signal propagation distance estimation, resulting in low positioning accuracy and poor robustness. In response to this problem, this article proposes a multisource information fusion-based indoor positioning approach via Gaussian kernel density estimation. First, the heuristic distribution model is used to establish the mathematical relationship between the RSS from different Wi-Fi APs and the signal propagation distance, and then the Gaussian kernel density estimation approach is applied to estimate the signal propagation distance distribution. Second, the normalized signal propagation distance distribution is set as the basic probability assignment in the Dempster–Shafer (D–S) evidence theory, and then the multisource RSS information is fused according to D–S evidence synthesis rules. Meanwhile, the trust function synthesized by the D–S evidence theory is used to select matching reference points (RPs). Finally, the fuzzy decision algorithm is used to select the ideal matching RPs from the matching RPs for the positioning. Experimental results show that the proposed approach has higher positioning accuracy and stronger positioning robustness compared to existing ones.

Channel state information (CSI) can provide phase and amplitude of multichannel subcarrier to better describe signal propagation characteristics. Therefore, CSI has become one of the most commonly used features in indoor Wi-Fi localization. In addition, compared to the CSI geometric localization method, the CSI fingerprint localization method has the advantages of easy implementation and high accuracy. However, as the scale of the fingerprint database increases, the training cost and processing complexity of CSI fingerprints will also greatly increase. Based on this, this article proposes to combine backpropagation neural network (BPNN) and adaptive genetic algorithm (AGA) with CSI tensor decomposition for indoor Wi-Fi fingerprint localization. Specifically, the tensor decomposition algorithm based on the parallel factor (PARAFAC) analysis model and the alternate least squares (ALSs) iterative algorithm are combined to reduce the interference of the environment. Then, we use the tensor wavelet decomposition algorithm for feature extraction and obtain the CSI fingerprint. Finally, in order to find the optimal weights and thresholds and then obtain the estimated location coordinates, we introduce an AGA to optimize BPNN. The experimental results show that the proposed algorithm has high localization accuracy, while improving the data processing ability and fitting the nonlinear relationship between CSI location fingerprints and location coordinates.

With the rapid increase of industrial systems, industrial spectrum is stepping into the era of big data, and at the same time spectrum resources are facing serious shortage. Cognitive industrial system (CIS) based on cognitive radio can improve spectrum utilization by accessing the idle spectrum licensed to primary user. However, the CIS must find enough idle channels by performing spectrum sensing. In this article, a reinforcement learning-based multislot double-threshold spectrum sensing with Bayesian fusion is proposed to sense industrial big spectrum data, which can find required idle channels faster while guaranteeing spectrum sensing performance. Double thresholds are set to guarantee both high detection probability and spectrum access probability, and weighed energy detection is proposed to maximize detection probability when the energy statistic falls into the confusion area between the double thresholds. Bayesian fusion is proposed to get a final decision on the channel availability by combining the local sensing decisions of all the time slots. A prediction and selection algorithm for idle channels is proposed to predict the idle probability of each channel and find required idle channels from the sorted channel set. From simulation results, the proposed spectrum sensing scheme outperforms cooperative spectrum sensing and energy detection, which can predict idle channels accurately and get needed idle channels with fewer sensing operations.

Existing domain adaptation methods aim at learning features that can be generalized among domains. These methods commonly require to update source classifier to adapt to the target domain and do not properly handle the trade-off between the source domain and the target domain. In this work, instead of training a classifier to adapt to the target domain, we use a separable component called data calibrator to help the fixed source classifier recover discrimination power in the target domain, while preserving the source domain's performance. When the difference between two domains is small, the source classifier's representation is sufficient to perform well in the target domain and outperforms GAN-based methods in digits. Otherwise, the proposed method can leverage synthetic images generated by GANs to boost performance and achieve state-of-the-art performance in digits datasets and driving scene semantic segmentation. Our method also empirically suggests the potential connection between domain adaptation and adversarial attacks.

Device-Free Localization and Tracking (DFLT) acts as a key component for the contactless awareness applications such as elderly care and home security. However, the random phase errors in WiFi signal and weak target echoes submerged in background clutter signals are mainly obstacles for current DFLT systems. In this paper, we propose the design and implementation of MuTrack, a multiparameter based DFLT system using commodity WiFi devices with a single link. Firstly, we select an antenna with maximum reliability index as the reference antenna for signal sanitization in which the conjugate operation removes the random phase errors. Secondly, we design a multi-dimensional parameters estimator and then refine path parameters by optimizing the complete data of path components. Finally, the Hungarian Kalman Filter based tracking method is proposed to derive accurate locations from low-resolution parameter estimates. We extensively validate the proposed system in typical indoor environment and these experimental results show that MuTrack can achieve high tracking accuracy with the mean error of 0.82 m using only a single link.

MOTIVATION: Accurate prediction of cancer drug response (CDR) is challenging due to the uncertainty of drug efficacy and heterogeneity of cancer patients. Strong evidences have implicated the high dependence of CDR on tumor genomic and transcriptomic profiles of individual patients. Precise identification of CDR is crucial in both guiding anti-cancer drug design and understanding cancer biology. RESULTS: In this study, we present DeepCDR which integrates multi-omics profiles of cancer cells and explores intrinsic chemical structures of drugs for predicting CDR. Specifically, DeepCDR is a hybrid graph convolutional network consisting of a uniform graph convolutional network and multiple subnetworks. Unlike prior studies modeling hand-crafted features of drugs, DeepCDR automatically learns the latent representation of topological structures among atoms and bonds of drugs. Extensive experiments showed that DeepCDR outperformed state-of-the-art methods in both classification and regression settings under various data settings. We also evaluated the contribution of different types of omics profiles for assessing drug response. Furthermore, we provided an exploratory strategy for identifying potential cancer-associated genes concerning specific cancer types. Our results highlighted the predictive power of DeepCDR and its potential translational value in guiding disease-specific drug design. AVAILABILITY AND IMPLEMENTATION: DeepCDR is freely available at https://github.com/kimmo1019/DeepCDR. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: Mining drug-disease association and related interactions are essential for developing in silico drug repurposing (DR) methods and understanding underlying biological mechanisms. Recently, large-scale biological databases are increasingly available for pharmaceutical research, allowing for deep characterization for molecular informatics and drug discovery. However, DR is challenging due to the molecular heterogeneity of disease and diverse drug-disease associations. Importantly, the complexity of molecular target interactions, such as protein-protein interaction (PPI), remains to be elucidated. DR thus requires deep exploration of a multimodal biological network in an integrative context. RESULTS: In this study, we propose BiFusion, a bipartite graph convolution network model for DR through heterogeneous information fusion. Our approach combines insights of multiscale pharmaceutical information by constructing a multirelational graph of drug-protein, disease-protein and PPIs. Especially, our model introduces protein nodes as a bridge for message passing among diverse biological domains, which provides insights into utilizing PPI for improved DR assessment. Unlike conventional graph convolution networks always assuming the same node attributes in a global graph, our approach models interdomain information fusion with bipartite graph convolution operation. We offered an exploratory analysis for finding novel drug-disease associations. Extensive experiments showed that our approach achieved improved performance than multiple baselines for DR analysis. AVAILABILITY AND IMPLEMENTATION: Source code and preprocessed datasets are at: https://github.com/zcwang0702/BiFusion.

Being essential for many emerging applications, the device-free localization systems have gained increasing interest, of which the through-the-wall device-free localization is of great challenge. This paper presents the design and implementation of TWPalo, a through-the-wall device-free localization system based on Wi-Fi channel state information (CSI). To this end, we first develop an algorithm for three dimensional joint estimation of angle of arrival (AoA), time of flight (ToF) and Doppler frequency shift (DFS). Combining with this algorithm, we then separate the CSI and obtain the parameters of each propagation path through the iteration of parameter estimation, channel reconstruction and cancellation. At last, the human induced reflection is found out and its relevant parameters are translated into the precise location of the human behind the wall. Our implementation and evaluation on commodity Wi-Fi devices demonstrate that TWPalo is better than existing systems in the form of AoA estimation and localization accuracy under the through-the-wall scenario.

Recently, hand gesture detection (HGD) system have become increasingly interesting to researchers in the field of human-computer interfaces. However, the traditional HGD has low robustness and detection accuracy, as well as privacy protection problem. Therefore, we present Rammar, a residual attention module (RAM) assisted Mask R-CNN, for frequency modulated continuous wave (FMCW) sensor based on HGD system. Firstly, by analyzing the time domain and frequency domain of the FMCW sensor signal, the three-dimensional feature maps of Range-Time-Map (RTM), Doppler-Time-Map (DTM) and Angle-Time-Map (ATM) of each hand gesture are obtained, respectively, avoiding insufficient information of single dimension parameter. Secondly, RTM, DTM and ATM images of each hand gesture are simultaneously sent to Rammar for training. To focus on the features of gesture, RAM in Rammar employs average-pooling and max-pooling to extract time and spatial features. Finally, the extracted three-dimensional feature maps are merged in the fully connected layer. The experimental results show that Rammar not only makes the average detection accuracy of the hand gestures to 98.1%(increased by 5%), but also reduces the detection time effectively.

Due to the rapid development of indoor Unmanned Aerial Vehicles (UAVs) in recent years, the localization of indoor UAVs has become a focus of attention in UAVs applications. Among them, the Wireless Local Area Network (WLAN) based localization approach has become an effective means to achieve indoor localization due to the widely-deployed WLAN infrastructure. At the same time, with the increased use of WLAN module in the state-of-the-art mobile devices, various types of mobile WLAN Access Points (APs) exist in indoor environment. In this circumstance, the mobile WLAN APs deteriorates localization accuracy since their associated Received Signal Strength (RSS) data become unstable with the variation of locations. To address this problem, a new approach based on the Density-based Spatial Clustering of Applications with Noise (DBSCAN) is proposed to detect mobile WLAN APs for manifold alignment localization of UAVs. Specifically, first of all, the DBSCAN is conducted at the Reference Points (RPs) on motion paths to detect mobile APs. Second, the RSS data from mobile APs are removed from the database to enhance the location-dependency of RSS data used for the localization. Third, the concept of augmentation process is considered in manifold alignment to achieve satisfactory localization accuracy. Finally, the extensive experimental results show that the proposed system performs better in localization accuracy compared with the existing CIMLoc and WILL under the presence of mobile WLAN APs.

With the fast growth of demand for the ubiquitous, precise, and instant indoor location information, the Received Signal Strength (RSS) based Wi-Fi indoor localization has been greeted with an avalanche of publicity. The studies in this field so far rarely consider the diversity of Wi-Fi signals, and thereby the RSS measures involving gross error on account of the complicated indoor environment deteriorate localization accuracy. In response to this compelling problem, we propose to use the concept of Asymptotic Relative Efficiency (ARE) to design a new synergetic test of Wi-Fi signal for indoor localization. Specifically, first of all, the Jarque-Bera (JB) test is conducted to test the normality of Wi-Fi signals at each Reference Point (RP). Second, the result of JB test is fed into the synergetic Mann-Whitney U and T test to construct the set of matching RPs corresponding to the newly-collected RSS data. Finally, the location coordinate of the target is obtained by calculating the K-nearest neighbor of matching RPs. Furthermore, the experimental results demonstrate that the proposed approach is featured with higher localization accuracy compared with the existing Wi-Fi indoor localization approaches.

With the widespread adoption of Wireless Local Area Network (WLAN) in indoor environment, indoor WLAN intrusion detection has become a key technique in various fields with the advantage of accomplishing intrusion detection without any requirement of special device or collaboration from the target. However, this technique is suffered by a serious problem that the offline database construction normally leads to high manpower and time cost especially for the large-scale indoor environment. To address this problem, a new indoor WLAN intrusion detection approach with low effort is proposed in this paper. Specially, first of all, the difference between the Received Signal Strength (RSS) data in source and target domains at the same locations is reduced by intra-class transfer learning with the purpose of applying the relations between the offline RSS data and their labels to the online RSS data. Second, the RSS data in target domain are classified by using the classifier trained from the relations of RSS data and the corresponding labels in source domain. Third, the iterative transfer learning between source and target domains is conducted to obtain the labels of RSS data in target domain. Finally, the experimental results demonstrate that the proposed approach is able to achieve high detection accuracy as well as the strong robustness to the number of RSS data used for database construction.

The Internet of Things (IoT) has gradually changed the way of people’s lives due to its ability of connecting everything together, and meanwhile the accurate location sensing plays a crucial role in achieving this goal. Up to now, as one of the most representative outdoor localization systems, the global positioning system has been widely used, but its performance may be dramatically declined in indoor environment due to the serious multipath effect and signal attenuation caused by the complicated indoor structure. At the same time, the location fingerprint-based localization approach has become a popular one in indoor environment, and meanwhile the corresponding calibrated signal simplification in location database construction has been primarily considered due to its significant practical meaning in avoiding the blind signal sampling. In this paper, we propose to use an information-theoretic lens to construct the energy-efficient location fingerprint database for the localization in IoT. Interestingly, by analyzing the information loss in signal sampling, we analogize the database construction process into the information propagation process in a lossy channel, and then formulate the relations of sample capacity and localization error from an information-theoretic view. After that, by selecting an appropriate time interval to sample the independent and nonredundant signal, the minimum number of sampled signal under the given expected localization accuracy is determined. Finally, the extensive experimental results show that compared with the state-of-the-art approaches, the proposed one can effectively simplify the calibrated data for the energy-efficient location database construction in different wireless localization networks.

To facilitate green and secure communications, the energy efficiency (EE) and physical (PHY) layer security are desirable for cloud radio access network (C-RAN). However, the problem of EE optimization with guaranteed PHY layer security in C-RAN is challenging, especially in the presence of channel state information (CSI) uncertainty of both information receivers (IRs) and eavesdropper receivers (ERs). Thereby, in this paper, we investigate the worst-case EE maximization problem in a downlink C-RAN, subject to limited power budget of each BS and infinite number of PHY layer security constraints, which is a non-convex fractional programming problem and is NP-hard even in the event of perfect CSI. To solve this non-trivial problem, a fast-converging algorithm is developed. Specifically, based on successive convex approximation (SCA) technique, we transform the original problem into a semi-definite program (SDP) one, allowing for solving it iteratively. The tightness of the SDR is proved, and the SCA algorithm is also proved to converge to a Karush- Kuhn-Tucker point. Extensive simulations are carried out to verify the effectiveness of the proposed algorithm.

In this paper, we present a Real-time Interacting with Hand Description system via Millimeter-wave Sensor (Riddle) for human-computer interaction. Firstly, we describe a new approach to developing a radar-based system. When hand motions are captured by millimeter- wave radar sensor, the unique range information can be observed in the spectrogram. Compared to traditional hand gesture recognition systems based on optical sensors, the radar-based system avoids the influence of ambient light conditions. Secondly, we employ deep neural networks combined with connectionist temporal classification algorithm to recognize diverse hand gestures in real-time. Besides, we visualize the feature maps extracted from different layers to understand the deep neural networks. The deep neural networks are powerful to extract hand gesture features as well as class boundaries through a training process. Finally, we demonstrate that Riddle is capable of detecting six hand gestures and achieving high recognition accuracy of 96%.

The rapid development of ubiquitous and high-speed wireless communication technology has driven the increasingly serious demand for the Location-based Services (LBSs). In this circumstance, we propose a new crowd-sourced calibration-free and inertial sensor- independent indoor pedestrian localization approach, namely Mann-Whitney rank-sum testing via segments labeling (Marvel). In concrete terms, first of all, the motion paths are modeled by using the A* algorithm with the floor plan provided by the merchant, and then each motion path is segmented according to the preset expected localization accuracy. Second, by setting the signal similarity threshold, the Received Signal Strength (RSS) sequences which are collected by the human subjects following their daily routines in target environment are also segmented. Third, the proposed Marvel is adopted to cluster the motion path segments as well as RSS sequence segments respectively to construct the physical and signal logic graphs. Finally, by using the concept of backbone nodes diffusion mapping to establish the mapping relations between the physical and signal spaces, the pedestrian localization and the related motion analysis are conducted by the server. Furthermore, the extensive experimental results show that the proposed approach is capable of achieving higher localization accuracy compared with the current state-of-the-art approaches.

The Internet of Thing (IoT) based applications can provide various services and be widely applied in intelligent home. With the tendency of house safety protection, the detecting accuracy and privacy of intrusion detection system, which detects the human motion in indoor environment, has become a continuing concern. Up to now, there are emerging many intrusion detection systems which employ different devices such as camera and infrared. However, the poor privacy and deployment of specialized devices are mainly disadvantages of the afore-mentioned systems for deploying in home environment. In this paper, we propose WLID, a whole-home level intrusion detection system based on RSSI (Received Signal Strength Indicator) measurements of WiFi in indoor complex environment. In order to expand the area of human presence detection, WLID cooperates with WiFi-enabled IoT devices such as smart TV, air conditioner and other smart devices. The detection system constructs a detection algorithm with the non-parametric statistical method by only using RSSI and realize whole-home level real-time detection by using software implementation. The experimental results show that WLID can achieve the consistent detection rate close to 100% in a practical home environment.

To address low efficiency of traditional fingerprint database construction approach, we propose a fingerprint database expansion approach based on Multiquadric Radial Basis Function (RBF) interpolation. First of all, multi-directional fingerprints are collected dynamically, and sparse fingerprint database is generated by combining Received Signal Strength (RSS) and coordinates. Subsequently, RSS of each new Reference Point (RP) is estimated by using optimized RBF approach. In particular, Genetic Algorithm (GA) is applied to optimize shape parameter, so as to improve interpolation accuracy. Extensive experimental results show that the proposed approach is able to achieve high localization accuracy as well as significantly reduce fingerprint database construction effort.

Applications on Location Based Services (LBSs) have attracted significant attention due to its personalized, convenient, and smart user experience, and meanwhile the accurate mapping and localization algorithm plays a crucial role in satisfying the LBSs. At the same time, motivated by the widely-deployed Wi-Fi network, the Wi-Fi signal based localization has become one of the superior positioning techniques in indoor environment, and the corresponding sample capacity involved in positioning database establishment should be given much attention due to its significant guidance meaning in practice. In this paper, we propose a new sample capacity optimization approach for indoor Wi-Fi localization from the information-theoretic view, namely Sample Capacity Optimization for Positioning database Establishment (SCOPE) in indoor Wi-Fi environment. Interestingly, we analogize the positioning database establishment process in indoor Wi-Fi environment into the information propagation process in a lossy channel, and meanwhile formulate the relations between the sample capacity and localization error. Experimental result shows that the proposed SCOPE can accurately estimate the minimum sample capacity with a given expected localization accuracy under different Access Point (AP) combination.

The indoor localization systems based on wireless local area network received signal strength (RSS) have been widely applied due to the simplicity of system deployment as well as easy implementation on various mobile devices like the smartphones. However, they are often suffered by the major drawback of the extensive effort for location fingerprinting which is significantly labor-intensive and time-consuming. In response to this compelling problem, we design an improved manifold alignment approach to construct a cost-efficient radio map which consists of the sparsely collected location fingerprints and crowdsourcing RSS data with the purpose of reducing the overall fingerprints calibration effort. A new graph construction scheme which is proved to be the optimal choice to model the smoothness assumption in semi-supervised learning is proposed to explore the informativeness conveyed by location fingerprints during the process of radio map construction. In addition, the concept of execution characteristic function is considered to minimize the RSS sample capacity at each reference point to reduce fingerprints calibration effort further. Finally, the extensive experimental results demonstrate the performance improvement by the proposed system with the probability of localization errors within 3 m, 79.60%, which is at most 26.30 percentages higher than the one by the existing systems using location fingerprints solely.

Low cost and high efficiency, defined as energy efficiency (EE) and spectral efficiency (SE), have raised more and more attention in the fifth generation (5G) communication systems due to steadily rising hardware cost, energy consumption, and mobile traffic. This paper studies the hybrid architecture of multiuser massive MIMO systems, where the digital domain utilizes the zero‐forcing (ZF) precoding scheme and the analog domain uses discrete Fourier transform (DFT) processing that significantly reduces hardware cost and energy consumption. We derive analytical expressions on the total achievable SE and EE, as well as offering insight into some engineering parameters in the system performance. Our aim is to achieve low cost and high efficiency massive MIMO system, with constraints on the overall transmit power, the number of users, and the number of radio frequency (RF) chains. Results exhibit that the total achievable SE of the hybrid architectures with DFT precessing is inferior to the full digital architectures and hybrid architectures with the ideal phase shifters, but the performance attenuation can be compensated by providing the more input SNR and higher number of RF chains. Moreover, we find that the total achievable EE of hybrid architectures with DFT precessing outperforms other massive MIMO architectures that include a full digital implementation, ideal phase shifters, and a switched network.

Applications on Location Based Services (LBSs) have driven the increasing demand for indoor localization technology. The conventional location fingerprinting based localization involves heavy time and labor cost for database construction, while the well‐known Simultaneous Localization and Mapping (SLAM) technique requires assistant motion sensors as well as complicated data fusion algorithms. To solve the above problems, a new pedestrian motion learning based indoor Wireless Local Area Network (WLAN) localization approach is proposed in this paper to achieve satisfactory LBS without the demand for location calibration or motion sensors. First of all, the concept of pedestrian motion learning is adopted to construct users’ motion paths in the target environment. Second, based on the timestamp relation of the collected Received Signal Strength (RSS) sequences, the RSS segments are constructed to obtain the signal clusters with the newly defined high‐dimensional linear distance. Third, the PageRank algorithm is performed to establish the hotspot mapping relations between the physical and signal spaces which are then used to localize the target. Finally, the experimental results show that the proposed approach can effectively estimate the target’s locations and analyze users’ motion preference in indoor environment.

(First paragraph) In the first edition of the special issue titled “Wireless Sensor Networks for Smart Communications”, a total of 22 manuscripts were received and 6 of these were accepted. This issue demonstrated that network congestion, user mobility, and adjacent spectrum interference are the main reasons for the degradation ofcommunication quality inWireless Sensor Networks (WSNs).

Knowing whether the propagation path between transmitter and receiver is Line-Of-Sight (LOS) or No-Line-Of-Sight (NLOS) propagation is a important factor for improving the performance of communication services and vast of mobile computing applications. Several promising systems in the current commodity WiFi networks analyze frequency- dependent amplitude and phase of Channel State Information (CSI) to achieve a LOS/NLOS recognition scheme, but robustness not be considered enough since there are many frequently-used wireless channels. With the development of Multiple-Input- Multiple-Output (MIMO), the spatial properties of channel can be obtained effortlessly. Consequently, it provides a potential to utilize the unchanged spatial information of multipath signals to get a robustness LOS recognition system. Here, we propose Wi-vision, an accurate and robust LOS recognition system based on Single-Input-Multiple-Output (SIMO) measurements in indoor environment. We creatively introduce the Hopkins statistic to measure the distribution of Angle-Of-Arrival (AOA) and relative Time-Of-Flight (TOF) of multipath under LOS and NLOS condition, respectively. The test results show that Wi-vision possesses consistent LOS and NLOS detection rate of above 91% and 87% respectively with different system configurations.

The indoor positioning system based on Micro Electro Mechanical Systems (MEMS) sensors is featured with short-term high accuracy, whose current positioning performance depends on the historical positioning result. Therefore, MEMS positioning has long-time error accumulation. The fingerprint positioning of Bluetooth Low Energy (BLE) is independent of accumulative error, but there is an irregular jump error in the positioning result, which limits the positioning accuracy. Furthermore, the actual commercial positioning systems generally require the consecutive positioning in multi-floor environment. Based on this, this paper proposes a data fusion algorithm based on BLE and MEMS for indoor cross-floor positioning. Firstly, we denoise the fingerprint database by clustering, outlier detection, and filtering algorithms. Then, the extended Kalman filter is employed to complete the optimal estimation of the two-dimensional target position according to the robust M estimation. Finally, the barometer and geographical position information are used to achieve the height estimation of the target. This paper also carries out a large number of engineering verification. The experimental results show that the algorithm can suppress the cumulative error effectively caused by low-cost MEMS sensors, and solve the problem of irregular jump error caused by Received Signal Strength Indicator (RSSI) jitter. In the indoor multi-layer environment, the proposed system achieves the horizontal and vertical positioning Root Mean Square (RMS) errors less than 0.9 m and 0.35 m respectively. In addition, we have verified the stability of the designed system through the long-time test.

With the rapid development of Wireless Local Area Network (WLAN) technique, the indoor WLAN localization has caught significant attention. In this paper, a novel indoor WLAN localization approach by using the high-dimensional manifold alignment with limited calibration load is proposed. Different from the conventional dimension-reduction based manifold alignment approach which preserves a limited part of the Received Signal Strength (RSS) data information, we first construct an innovative objective function from the augmented physical locations and the corresponding RSS data. Second, the closed-form solution to the objective function is obtained by applying the Lagrange multiplier approach. Finally, the target location is estimated at the closest point in the manifold. Furthermore, we present some preliminary analysis towards the generalization of the proposed objective function to the scenario with multiple types of measurements used for the localization. The extensive analytical and experimental results demonstrate that the performance of the proposed approach is well with limited calibration load and can be further improved by using more calibrated locations with known RSS data.

Accurate lung nodule segmentation from computed tomography (CT) images is of great importance for image-driven lung cancer analysis. However, the heterogeneity of lung nodules and the presence of similar visual characteristics between nodules and their surroundings make it difficult for robust nodule segmentation. In this study, we propose a data-driven model, termed the Central Focused Convolutional Neural Networks (CF-CNN), to segment lung nodules from heterogeneous CT images. Our approach combines two key insights: 1) the proposed model captures a diverse set of nodule-sensitive features from both 3-D and 2-D CT images simultaneously; 2) when classifying an image voxel, the effects of its neighbor voxels can vary according to their spatial locations. We describe this phenomenon by proposing a novel central pooling layer retaining much information on voxel patch center, followed by a multi-scale patch learning strategy. Moreover, we design a weighted sampling to facilitate the model training, where training samples are selected according to their degree of segmentation difficulty. The proposed method has been extensively evaluated on the public LIDC dataset including 893 nodules and an independent dataset with 74 nodules from Guangdong General Hospital (GDGH). We showed that CF-CNN achieved superior segmentation performance with average dice scores of 82.15% and 80.02% for the two datasets respectively. Moreover, we compared our results with the inter-radiologists consistency on LIDC dataset, showing a difference in average dice score of only 1.98%.

Decisions about cervical cancer diagnosis and classification currently require microscopic examination of cervical tissue by an expert pathologist. In the present study, which focused on full automation of this approach, we solely use nucleus-level features to classify tissues as normal or cancer. We propose Adaptive Nucleus Shape Modeling (ANSM) algorithm for nucleus-level analysis which consists of two steps to capture the nucleus-level information: adaptive multilevel thresholding segmentation; and shape approximation by ellipse fitting. After applying the proposed algorithm, the features are extracted for tissue classification. Experiments show that ANSM can achieve an accuracy of 93.33% with a false negative rate of zero in classifying cancer and healthy cervical tissues using nucleus texture features. This provides evidence that nucleus-level analysis is valuable in cervical histology image analysis.

Finding appropriate feature representations from radiological images is a vital task for prediction and diagnosis. Deep convolutional neural networks have recently achieved state-of-the-art performance in classification problems from several different domains. Research has also shown the feasibility of using a pre-trained deep neural network as a feature extractor when only a small dataset is available. This paper proposes a novel image feature extraction method for predicting survival time from brain tumor magnetic resonance images using pretrained deep neural networks. Since all tumors are different sizes, we also explore different image resizing methods in the paper. We demonstrate that deep features can result in better survival time prediction with the highest accuracy of 95.45% versus conventional feature extraction methods from magnetic resonance images of the brain.

Device-free Passive (DfP) localization is an emerging technology that uses the widely deployed wireless networks to detect and localize the people and other entities in target environment. The existing DfP localization systems realize the detection and localization under the WLAN indoor environment, but the low transmission power of the Access Points (APs) restricts their application. In this paper, we propose a novel DfP motion detection system based on the cellular network with the purpose of achieving the accurate, robust, low-overhead, and long-distant motion detection capability. To overcome the poor detection performance resulted from the time-varying signal, the signal strength difference at different timestamp is adopted as the characteristic parameter of the system. We apply the non-parametric kernel density estimation technique to calculate the optimum detection threshold of each signal stream. Furthermore, a joint detection mechanism is introduced to reduce the impact of noisy readings, as well as enhance the detection performance of the system using the cellular network. The results in two typical test beds show that the proposed system can achieve high detection accuracy with false negative (FN) rate 0.8% and false positive (FP) rate 4.6%, while require significantly lower deployment overhead and be more robust to the environmental changes compared with the WLAN DfP detection system.

Access Point (AP) optimization is one of the most important components in indoor Wireless Local Area Network (WLAN) localization technique since the AP number and locations have significant impact on the variations of Received Signal Strength (RSS) in target environment. Different from the conventional AP optimization approaches, we propose to use the concept of adaptive channel power allocation to construct a water-filling model, and then conduct AP optimization based on the weights of candidate AP locations which are calculated by the fast discrete water-filling algorithm. The experimental results demonstrate that the proposed approach is able to achieve high localization precision, and meanwhile consume low time overhead.

Multi-path fading, environmental shadowing and channel interference always result in the significant temporal and spatial variations of Received Signal Strength (RSS), and eventually lead to the low accuracy in Wireless Local Area Networks (WLAN) fingerprint based indoor localization. Motivated by this, we focus on deriving out the positioning error bound which can be applied to characterize the theoretical relationship between the positioning errors and signal distributions by using Fisher Information Matrix (FIM). Furthermore, the positioning error bound is recognized as an effective criterion of designing a more beneficial WLAN deployment with higher positioning accuracy. Extensive simulations are conducted in a regular Lineof-sight (LOS) environment as well as in a complex irregular Non-line-of-sight (NLOS) environment.

In recent decade, the Received Signal Strength (RSS) based indoor localization has caught significant attention, but it always suffers from the time-consuming and labor intensive fingerprint calibration. At the same time, the Simultaneous Localization and Mapping (SLAM) technique is considered with the low time and laboring cost, whereas the dedicated hardware is often required. To solve these problems, a novel indoor WLAN SLAM approach by using the Edge Detection based Gene Sequencing (EDGES) is proposed. First of all, a batch of RSS sequences is sporadically collected in target area. Second, the spectral clustering is conducted on RSS sequences to construct the cluster graphs, and then the EDGES approach is applied to assemble the cluster graphs into a logic graph. Finally, the mapping from the logic graph into ground-truth graph is established to realize indoor WLAN SLAM. The extensive experimental results prove that the proposed approach can achieve satisfying localization accuracy without site survey of location fingerprinting or motion sensing.

In the area of Wireless Local Area Network (WLAN) based indoor localization, the Received Signal Strength (RSS) fingerprinting based localization technique has been studied extensively. Site survey phase in RSS fingerprinting is always considered to be time-consuming and labor intensive. To solve this problem, we propose a novel Indoor Mapping and Localization Using RSS Solely (IMLours) approach, which utilizes the spectral clustered time-stamped WLAN RSS data to characterize environmental layout, as well as conduct target localization. First of all, we use the off-the-shelf smartphones to sporadically record a batch of WLAN RSS data in indoor environment. Second, spectral clustering is applied to classify the RSS data in each sequence into different clusters. The clusters are then used to construct the logic graphs. Third, we do the mapping from logic graphs into ground-truth graph. Finally, based on the extensive experiments conducted in a real WLAN indoor environment, our proposed IMLours approach is proved to achieve satisfactory localization accuracy.

In the area of Wireless Local Area Network (WLAN) based indoor localization, the Received Signal Strength (RSS) fingerprinting based localization technique has been s-tudied extensively. Site survey phase in RSS fingerprinting is always considered to be time-consuming and labor intensive. To solve this problem, we propose a novel Indoor Mapping and Localization Using RSS Solely (IMLours) approach, which utilizes the spectral clustered time-stamped WLAN RSS data to characterize environmental layout, as well as conduct target localization. First of all, we use the off-the-shelf smartphones to sporadically record a batch of WLAN RSS data in indoor environment Second, spectral clustering is applied to classify the RSS data in each sequence into different clusters. The clusters are then used to construct the logic graphs. Third, we do the mapping from logic graphs into ground-truth graph. Finally, based on the extensive experiments conducted in a real WLAN indoor environment, our proposed IMLours approach is proved to achieve satisfactory localization accuracy.

The automatic identification system (AIS) is a wireless synchronous narrowband communication system for exchanging navigational data through short messages between ships and base stations in the maritime VHF frequency band. The service coverage of each communication cell is limited within the horizon by the line of sight propagation of VHF signals. AIS receivers are put on high buildings besides harbors, lighthouses along coastlines and even on LEO satellites in space to detect signals from more cells. The toughest problem that these receivers are facing is the partial overlapping of messages from multiple cells. Antenna arrays are used in AIS receivers to suppress the interference. A good beamforming algorithm is needed for computing beamformers. Previously, we proposed blind beamforming techniques based on subspace intersection to solve this problem. In this paper, we propose more simple and efficient blind beamforming techniques using the generalized singular value decomposition (GSVD) and its tracking version for the same problem. The proposed algorithms are simulated in an exact dynamic AIS software model, and are tested on real-world AIS signals collected from our experimental hardware. Simulation results show that the proposed algorithms achieve satisfactory performance. Experimental results confirm the effectiveness of the proposed algorithms.

Brain tumor heterogeneity is well recognized in clinical MRI imaging and it is a challenging problem to quantitatively explore the underlying variations. It is known that brain tumors in different patients can have remarkably diverse visual appearances. In this paper, we propose a novel concept to categorize brain tumors with emphasis on spatial "habitats": a tumor can be quantified into distinctive sub-regions where the potential dynamics of tumor evolution may be evident. Our work is aimed at discovering spatially distinctive habitats within the tumor region, which may be useful in clinical practice for image-guided therapy. In particular, the heterogeneity can be well captured by two main steps: (a) intra-tumor segmentation, (b) spatial mapping scheme from a multi-modality MRI imaging dataset (Tl-weighted, FLAIR and T2-weighted MRI slices). A tumor region is initially segmented into high and low signal groups and then a joint mapping scheme is used to consider the correlation between different input modalities. In addition, focusing on signal contrast, we propose a set of quantitative features to measure differences between sub-regions. We further examined the application of survival time prediction for patients with malignant Glioblastoma multiforme (GBM). Experimental results showed that these features enabled classifiers to predict survival groups.

In breast cancer, tumor heterogeneity is a reflection of differing tumor subtypes, which may display markedly different genotypes and clinical phenotypes. Although pathological and qualitative (based on contrast enhancement patterns) studies suggest the presence of clinical and molecular predictive tumor subregions, this has not been fully investigated. Our goal is to develop a novel algorithm to utilize the potential information available in different tumor subregions (periphery and core) by extracting textural kinetic features, for the purpose of estrogen receptor (ER) classification. We show that features from different tumor subregions, at appropriate scales and quantization levels, can be used to better classify ER subtypes than features averaged from the whole tumor. We analyzed representative two dimensional (2D) slices from twenty breast tumors with volumetric dynamic contrast enhanced magnetic resonance imaging (DCE-MRI) available by extracting multi-parametric textural kinetic features from the periphery, core and whole tumor. The utility of the features from different subregions are evaluated using six meta-classifiers (feature selector and classifier pairs), formed from two feature selectors and three classifiers. Classification accuracy approached 94%.

Automated prediction of patient-specific disease progression can significantly contribute to clinical treatment. This paper presents a computer-assisted framework to tackle the survival time prediction problem. Inspired by the assumption that niche tumor regions may play a significant role in cancer diagnosis, we explore local visual variations from multiple MRI sequences. The research consists of three parts: 1) the extraction of multi-scale Local Binary Patterns (LBP) to describe the visual variations, 2) a supervised forward feature selection approach, called the Feature Ranking Model (FRM) which captures single feature predictive ability efficiently, and combines the top features to form a feature subset, 3) We cast the clinical survival time prediction task as a binary category classification problem. We tested the framework using a dataset of 32 cases collected from The Cancer Genome Atlas (TCGA). We obtained a 93.75% accuracy rate for the prediction of survival time.

In the area of performance analysis of wireless networks, one critical issue is the hidden terminal problem, which is considered as one of the severest reasons for the degradation of network performance. In this paper, we incorporate reactive jamming scheme with distributed coordination function (DCF) in IEEE 802.11 based wireless local area networks (WLANs) to improve network throughput in the presence of hidden terminals. In the proposed protocol, we schedule access point (AP) to broadcast jamming signal reactively to hinder the simultaneous transmission of hidden terminals. Both analytical and numerical results show that our reactive jamming based protocol can constantly improve WLAN throughput for a wide range of conditions, compared with the traditional RTS/CTS.

In this paper, we present a fingerprint-based particle filtering (PF) approach for the mobility tracking in cellular networks. With the popularity of location-based services (LBSs) in a recent decade, the mobility tracking has now become one of the indispensable techniques to fulfill the pervasive and location-aware computing. However, the tracking results from the conventional K nearest neighbors (KNN) and Kalman filtering (KF) by the cellular data appear to be coarse due to the reflection, refraction and diffraction of received signal code power (RSCP). Therefore, we propose the KNN/PF as an effective way to improve the stability and accuracy of mobility tracking in cellular networks. Furthermore, the experiments conducted in Pudong New District, Shanghai, China, show that the KNN/PF approach can achieve better cumulative density function (CDF) of tracking errors compared with the KNN and the mixed KNN and KF (KNN/KF) approaches.

P2P live streaming systems have been widely adopted nowadays. In such systems, flash crowds still remains a big challenge, which often occur when an enormous number of users suddenly arrive to view a newly released program. In a flash crowd scenario, users often suffer from a long startup delay and a high failure rate. In this paper, we propose a topology-construction-based algorithm to alleviate the flash crowd. Specifically, first the tracker server constructs a multiple tree topology with total new peers. Then according to the topology, all new peers join the current P2P system in form of multiple trees. When constructing the topology, the tracker server puts new peers with higher bandwidth and longer waiting time more closer to the root in each tree, in order to reduce the average waiting time of new peers. Moreover, a new analytical model is also devised to evaluate our algorithm. Model analysis and simulation indicate that our method can enhance the joining process of new peers and improve their startup delay and failure rate.

The coverage of the terrestrial automatic identification system (AIS) is limited to close areas off the coast. Low earth orbit (LEO) satellites can expand the service of AIS to a global range but it brings large variance in Doppler shift, path loss and propagation delay. The communication between ships and LEO satellites becomes asynchronous. The collision of AIS messages from thousands of ground cells results in loss of all collided messages. Previous papers discussed the use of a single user receiver to detect AIS messages under heavy co-channel interference but the problem was never well solved. In this paper, we present a multi-user receiver equipped with an antenna array on LEO satellites, which explores the spatial multiplexing in space detection of AIS messages and significantly improves the detection performance. The proposed receiver performs rank tracking and subspace intersection based on the signed URV decomposition ahead of blind source separation to provide robust separation of user data for single user receivers. The proposed receiver is tested in an exact dynamic AIS model.

As the most competitive and cost-efficient localization technology, the map-aided fingerprinting-assisted positioning has drawn a large body of attentions during the past decade. There are normally two phases, the off-line and on-line phases, and the construction of radio map in the off-line phase will significantly influence the location accuracy in the on-line phase. However, the radio map which describes the mapping relationship between the physical positions of the reference points (RPs) and the recorded radio signal strength (RSS) or signal to noise ratio (SNR) always involves the cumbersome collection work. Further, thousands of Wi-Fi access points (APs) remaining idle will also bring serious concerns of the power consumption in the future. Therefore, in response to these compelling problems, we present a preliminary analysis towards the probabilistic location methods, develop the green global-greedy position estimation (3GPE) and introduce entropy deduction as a new metric for performance evaluation. Finally, our analytical expressions and simulation results with respect to the simple circle model and ideal indoor Wi-Fi regular environment demonstrate these issues and challenges.

This paper presents a novel framework for recognition of facial action unit (AU) combinations by viewing the classification as a sparse representation problem. Based on this framework, we represent a facial image exhibiting the combination of AUs as a sparse linear combination of basis constituting an overcomplete dictionary. We build an overcomplete dictionary whose main elements are mean Gabor features of AU combinations under examination. The other elements of the dictionary are randomly sampled from a distribution (e.g., Gaussian distribution) that guarantees sparse signal recovery. Afterwards, by solving L1-norm minimization, a facial image is represented as a sparse vector which is used to distinguish various AU patterns. After calculating the sparse representation, the classification problem is simply viewed as a rank maximal problem. The index of the maximal value of the sparse vector is regarded as the class label of the facial image under test. Extensive experiments on the Cohn-Kanade facial expressions database demonstrate that this sparse learning framework is promising for recognition of AU combinations.

The laborious collection of location fingerprints, that could also potentially change with time, remains a hurdle towards the widespread deployment of indoor and campus area positioning using WiFi. In this paper, we present a preliminary analysis of the complicated relationship between distance in signal space, the physical distance and location errors towards better guidelines for fingerprint collection. We introduce the idea of entropy of location fingerprints and investigate the relationships between physical and signal distances, entropy, and expected errors with positioning using location fingerprinting. We present results with no access point and one access point and consider variations in the density of reference points and the standard deviation of signal strength to illustrate the issues.

The class of Schur subspace estimators provides a parametrization of all minimal-rank matrix approximants that lie within a specified distance of a given matrix, and in particular gives expressions for the column spans of these approximants. Unlike previous numerically unstable algorithms, this paper presents a signed URV decomposition (SURV) that efficiently and stably computes the Schur subspace estimator. Given a threshold on the singular values of the data matrix, SURV tracks the orthonormal basis of the principal/minor subspace and the rank of the subspace at the same time exactly with respect to the threshold at a computational complexity of O(m2) per vector update or downdate. SURV is not an iterative method.

This paper proposes the optimal K nearest neighbors (KNN) positioning algorithm via theoretical accuracy criterion (TAC) in wireless LAN (WLAN) indoor environment. As far as we know, although the KNN algorithm is widely utilized as one of the typical distance dependent positioning algorithms, the optimal selection of neighboring reference points (RPs) involved in KNN has not been significantly analyzed. Therefore, in order to fill this gap, the optimal KNN positioning algorithm based on the best TAC is introduced. And this algorithm is beneficial to construct the reliable WLAN indoor positioning system and provide the efficient location based services (LBSs). The relationship among theoretical expectation accuracy, unit interval of neighboring RPs and dimensions of target location region is also revealed. Furthermore, the feasibility and effectiveness of optimal KNN positioning algorithm are verified based on the experimental comparisons respectively in the regular office room, straight corridors, static positioning and dynamic tracking situations.