Aerial images from drones have been used to search individuals in the crowd. However, using a single drone for human searching faces challenges including low accuracy and long latency, due to poor visibility and limited on-board computing resources. In this paper, we propose SkyNet, a multi-drone cooperation system for real-time human searching, including locating and identifying. To locate a person, SkyNet uses Multi-View Cross Search with only 2D images. To achieve accurate identification, SkyNet processes faces in images from multi-view drones in three steps. First, a Multi-Modal Face Correction is designed to transform less useful face into desired target face, guided by text instructions. Second, an Angle Masking Network is developed to minimize invalid data of a single profile face. Third, the multiple face from drones are fused by a Fusion Weight Network. Moreover, by predicting the estimated finishing time of tasks, SkyNet schedules and balances workloads among edge devices and the cloud server to minimize processing latency. We implement SkyNet in real life, and evaluate the performance with 20 human participants. The results show that SkyNet can locate people within 0.18m error. The identification accuracy reaches 95.87%, and the system process is completed within 0.84s.

To reduce costs and improve performance, video Content Delivery Networks (CDNs) have started to incorporate lightweight edge nodes, e.g., WiFi access points. Because of this, it is necessary for CDNs to intelligently select which video files should be placed at their core data centers vs. these edge nodes. This is more complex than traditional CDN management, as lightweight edge nodes are much more numerous and unstable than data centers. With this in mind, we present SDPush —- a system for managing content placement in edge CDNs. SDPush tackles two problems. First, it is necessary for SDPush to select which files to proactive push. To address this, we build a file popularity prediction model that effectively identifies video files that will receive many views. Second, SDPush should determine how many replicas of each file to push. To address this, we design a model to predict the benefits of pushing particular files (regarding traffic savings) and then formulate the replica decision problem as a lightweight problem, which is solvable within seconds, even for platforms that accommodate millions of daily active users. Through a trace-driven evaluation and a live deployment on a real video platform, we validate SDPush’s effectiveness, offloading peak-period traffic by 12.1% to 23.9% from the data center to edge nodes, thereby reducing the CDN costs.

Drones connected via the web are increasingly being used for crowd anomaly detection (CAD). Existing solutions, however, face many challenges, such as low accuracy and high latency due to drones' dynamic shooting distances and angles as well as limited computing and networking capabilities. In this paper, we propose Air-CAD, an edge-assisted multi-drone network that uses air-ground cooperation to achieve fast and accurate CAD. Air-CAD consists of two stages: person detection and multi-feature analysis. To improve CAD accuracy, Air-CAD dynamically adjusts the inference of person detection model based on drones' shooting distances and assigns appropriate feature analysis tasks to drones shooting at variable angles. To achieve fast CAD, edge devices connected to drones are deployed to offload assigned feature analysis tasks from drones. Air-CAD schedules the connection between each drone and edge to accelerate processing based on drone's assigned task and the computing/network resources of the edge device. To validate the performance of Air-CAD, we generate a new simulated human stampede dataset captured from various drone-view recordings. We deploy and evaluate Air-CAD in both simulation and real-world testbed. Experimental results show that Air-CAD achieves 95.33% AUROC and real-time inference latency within 0.47 seconds.

Recently, the rapid growth of video content and the increasing demand for high Quality of Experience (QoE) have significantly strained the backbone network. Edge caching is a promising approach to alleviate the strain by caching content closer to users. However, it confronts challenges stemming from the low capability of individual edge nodes and the high density of their distribution, resulting in low hit ratios and unbalanced workloads. In this paper, we conduct in-depth analyses of these challenges and formulate a typical cooperative edge caching problem. Based on the insights, we introduce MagNet, a cooperative edge caching system featuring two key mechanisms: Automatic Content Congregating (ACC) and Mutual Assistance Group (MAG). ACC improves hit ratios by intelligently guiding requests to their optimal edges, thereby facilitating content aggregation. Complementing this, Quick Cache is implemented to accelerate this congregation process by prefetching content and optimizing cache space, effectively boosting hit ratios. MAG, on the other hand, achieves workload balance by dynamically forming groups to augment edge capabilities and redistribute requests on overloaded edges. To elucidate the design principles of MagNet, we conduct detailed component-level comparisons and quantitative analyses. To validate the overall performance, we compare it with various caching solutions using real-world datasets, demonstrating significant performance improvements.

Edge-cloud collaborative video analytics is transforming the way data is being handled, processed, and transmitted from the ever-growing number of surveillance cameras around the world. To avoid wasting limited bandwidth on unrelated content transmission, existing video analytics solutions usually perform temporal or spatial filtering to realize aggressive compression of irrelevant pixels. However, most of them work in a context-agnostic way while being oblivious to the circumstances where the video content is happening and the context-dependent characteristics under the hood. In this work, we propose VaBUS, a real-time video analytics system that leverages the rich contextual information of surveillance cameras to reduce bandwidth consumption for semantic compression. As a task-oriented communication system, VaBUS dynamically maintains the background image of the video on the edge with minimal system overhead and sends only highly confident Region of Interests (RoIs) to the cloud through adaptive weighting and encoding. With a lightweight experience-driven learning module, VaBUS is able to achieve high offline inference accuracy even when network congestion occurs. Experimental results show that VaBUS reduces bandwidth consumption by 25.0%-76.9% while achieving 90.7% accuracy for both the object detection and human keypoint detection tasks.

Video analytics systems conduct video preprocessing to filter out unnecessary frames and model inference using appropriately selected neural networks for high analytics speed. Video preprocessing is instruction-intensive computing (IIC) executed by CPU, and model inference is data-intensive computing (DIC) executed by GPU. In this paper, we show the analytics accuracy of existing systems can largely vary in fields, caused by thedynamicIIC and DIC workloads of differentcontentsin applications. Unfortunately, cameras havefixedCPU/GPU resources and cannot effectively adapt to workload dynamics. We develop Gemini, a new edge-side real-time video analytics system enhanced by a dual-image FPGA. We take the advantage of negligible image switching time of dual-image FPGAs, pre-configure one CPU image and one GPU image and elastically multiplex the dual CPU-GPU resources intimedimension. Gemini requires both hardware and software revisions. In hardware, we overcome challenges of hardware-dependent application development, low communication efficiency between the microprocessor and FPGA, and high programming complexity by hardware abstraction, asynchronous data transfer mechanism and stub-skeleton middleware. In software, we overcome the challenge of adapting to the dynamic workloads by a bandit learning approach. We implement Gemini and show that Gemini can improve the analytics accuracy to 90.35%.

Nowadays, the surge of Internet contents and the need for high Quality of Experience (QoE) put the backbone network under unprecedented pressure. The emerging edge caching solutions help ease the pressure by caching contents closer to users. However, these solutions suffer from two challenges: 1) a low hit ratio due to edges’ high density and small coverages. 2) unbalanced edges’ workloads caused by dynamic requests and heterogeneous edge capacities. In this paper, we formulate a typical cooperative edge caching problem and propose the MagNet, a decentralized and cooperative edge caching system to address these two challenges. The proposed MagNet system consists of two innovative mechanisms: 1) the Automatic Content Congregating (ACC), which utilizes a neural embedding algorithm to capture underlying patterns of historical traces to cluster contents into some types. The ACC then can guide requests to their optimal edges according to their types so that contents congregate automatically in different edges by type. This process forms a virtuous cycle between edges and requests, driving a high hit ratio. 2) the Mutual Assistance Group (MAG), which lets idle edges share overloaded edges’ workloads by forming temporary groups promptly. To evaluate the performance of MagNet, we conduct experiments to compare it with classical, Machine Learning (ML)-based and cooperative caching solutions using the real-world trace. The results show that the MagNet can improve the hit ratio from 40% and 60% to 75% for non-cooperative and cooperative solutions, respectively, and significantly improve the balance of edges’ workloads.

The task of the Intent Classification & Slot Filling serves as a key joint task in the voice assistant, which also plays the role of the pre-work in the construction of the medical consultation assistant system. How to distribute a doctor-patient conversation into a formatted electronic medical record to an accurate department (Intent Classification) to extract the key named entities or mentions (Slot Filling) through a specialized domain knowledge recognizer is one of the key steps of the entire system. In real cases, the medical vocabulary and clinical entities in different departments of the hospital often differ to some extent. Therefore, we propose a comprehensive model based on CMed-BERT, RCNN and BiGRU-CRF for a joint task of department identification and slot filling of the specific domain. Experimental results confirmed the competitiveness of our model.

Electronic Medical Records (EMR) carry a large number of diseases characteristics, history and other specific details of patients, which has great value for medical diagnosis. These data with diagnostic labels can help automated diagnostic assistant to predict disease diagnosis and provide a rapid diagnostic reference for doctors. In this study, we designed a BiGRU-Att-CapsNetwork model based on our proposed CMedBERT Chinese medical domain pre-trained language model to predict disease diagnosis in Chinese EMR. In the wide-ranging comparative experiments involving a real EMR dataset (SAHSU) and an academic evaluation task dataset (CCKS 2019), our model obtained competitive performance.

Writing Electronic Medical Records (EMR) as one of daily major tasks of doctors, consumes a lot of time and effort from doctors. This paper reports our efforts to generate electronic medical records using the language model. Through the training of massive real-world EMR data, the CMedGPT2 model provided by us can achieve the ideal Chinese electronic medical record generation. The experimental results prove that the generated electronic medical record text can be applied to the auxiliary medical record work to reduce the burden on the compose and provide a fast and accurate reference for composing work.

Introducing a large amount of external prior domain knowledge will effectively improve the performance of the word embedded language model in downstream NLP tasks. Based on this assumption, we collect and collate a medical corpus data with about 36M (Million) characters and use the data of CCKS2019 as the test set to carry out multiple classifications and named entity recognition (NER) tasks with the generated word and character vectors. Compared with the results of BERT, our models obtained the ideal performance and efficiency results.