Search

A scenario robust fall detection system based on few-shot learning (FDFL) in wireless environment was designed.The performance of existing fall detection methods based on Wi-Fi channel state information (CSI) degrades significantly across scenarios, which requires collecting and marking a large number of CSI samples in each application scenario, resulting in high cost for large-scale deployment.Therefore, the method of few-shot learning was introduced, which can maintain the performance of fall detection with high accuracy when the number of annotated samples in unfa-miliar scenes is insufficient.The proposed FDFL was mainly divided into two stages, source domain meta-training and target domain meta-learning.The meta training stage of the source domain consists of two parts: data preprocessing and classification training.In the data preprocessing stage, the collected original CSI amplitude and phase data were denoised and segmented.In the classification training stage, a large number of processed source domain data samples were used to train a CSI feature extractor based on convolutional neural network.In the meta-learning stage of the target domain, the limited labeled data sampled in the target domain was effectively extracted based on the feature extractor trained in the meta-training module, and then a lightweight machine learning classifier was trained to detect the fall behavior under the cross-scene.Through several experiments in different scenarios, FDFL can achieve an average accuracy of 95.52% for the four classification tasks of falling, sitting, walking and sit down with only a small number of samples in the target domain, and maintain robust detection accuracy for changes in test environment, personnel target and equipment location.

In order to deal with the increasingly scarce computing resources, a cooperative edge computing scheme was proposed, which makes full use of the idle resources among users to improve the overall data processing performance.To maximize the user utility, the target problem was formulated as an MINLP (mixed integer non-linear programming), and a learning-optimization-integrated method was proposed to jointly optimize the resource allocation and user offloading decisions.Simulation results show that the proposed scheme can produce a near-optimal solution in sub-second and effectively improve the system utility at least 85.4% compared to the considered benchmark methods.

In this paper, we consider a wireless powered communication network (WPCN) consisting of a multi-antenna hybrid access point (HAP) that transfers wireless energy to and receives sensing data from a cluster of low-power wireless devices (WDs). To enhance the throughput performance of some faraway WDs, we allow one of the WDs to act as the cluster head (CH) that helps forward the messages of the other cluster members. However, the performance of the proposed cluster-based cooperation is fundamentally limited by the high energy consumption of the CH, who needs to transmit all the WDs' messages including its own. To tackle this issue, we exploit the capability of multi-antenna energy beamforming (EB) at the HAP, which can focus more transferred power to the CH to balance its energy consumption in assisting the other WDs. Specifically, we first derive the throughput performance of each individual WD under the proposed scheme. Then, we jointly optimize the EB design, the transmit time allocation among the HAP and the WDs, and the transmit power allocation of the CH to maximize the minimum data rate achievable among all the WDs (the max-min throughput) for improved throughput fairness among the WDs. An efficient optimal algorithm is proposed to solve the joint optimization problem. Moreover, we simulate under practical network setups and show that the proposed multi-antenna enabled cluster-based cooperation can effectively improve the throughput fairness of WPCN.