BCD-NOMA enables two source nodes to communicate bidirectionally with their designated destination nodes, concurrently exchanging D2D messages via a relaying node. biosoluble film BCD-NOMA's architecture is optimized for improved outage probability (OP), high ergodic capacity (EC), and high energy efficiency. This architecture enables two data sources to share a single relay node for transmission to their respective destinations, and additionally supports bi-directional device-to-device (D2D) communication via downlink NOMA. Using analytical expressions and simulations of the OP, EC, and ergodic sum capacity (ESC) under perfect and imperfect successive interference cancellation (SIC), the benefit of BCD-NOMA over conventional schemes is illustrated.
The adoption of inertial devices in sports is experiencing a surge in popularity. This research project aimed to assess the degree to which various jump height measurement devices in volleyball were both valid and reliable. Employing keywords and Boolean operators, the search encompassed four databases: PubMed, Scopus, Web of Science, and SPORTDiscus. A total of twenty-one studies, complying with the specified selection criteria, were identified. Investigations concentrated on establishing the authenticity and dependability of IMUs (5238%), overseeing and measuring external burdens (2857%), and characterizing contrasts amongst playing positions (1905%). Indoor volleyball stands out as the modality where IMU application has reached the highest level. Senior, adult, and elite athletes were the demographic most subjected to evaluation. In both training and competition, IMUs were employed to assess jump quantity, height, and specific biomechanical characteristics. Established criteria and robust validity values are available for jump counting. A discrepancy exists between the reliability of the devices and the supporting evidence. Utilizing vertical displacement data, volleyball IMUs assess and record player movements, then compare them to playing positions, training protocols, and calculated athlete external loads. The measure displays sound validity, yet improvements in the reliability of measurements taken at different times are warranted. To establish IMUs as effective measurement tools for analyzing jumping and athletic performance in players and teams, further research is warranted.
The optimization function for sensor management in target identification is usually based on information-theoretic indicators, including information gain, discrimination, discrimination gain, and quadratic entropy. These metrics aim to reduce the overall uncertainty surrounding all targets, yet they don't consider the rate of target confirmation. Hence, guided by the maximum posterior criterion for target identification and the confirmation process for target identification, we study a sensor management approach preferentially allocating resources to targets that can be identified. An improved probability prediction method, rooted in Bayesian theory, is presented for distributed target identification. This approach leverages global identification results, providing feedback to local classifiers to boost the accuracy of identification probability prediction. A sensor management approach, utilizing information entropy and anticipated confidence values, is introduced to optimize the inherent ambiguity in target identification rather than its variations, thereby increasing the priority of targets achieving the desired confidence level. The sensor management strategy for identifying targets is ultimately modeled as a sensor allocation problem. An optimization function, based on an effectiveness metric, is then formulated, thereby improving the speed of target identification. Evaluation of experimental results shows a similar correct identification rate for the proposed method compared to information gain, discrimination, discrimination gain, and quadratic entropy methods; however, the average time needed to confirm the identification is the shortest.
Engagement is augmented by the capacity to reach a state of flow, which defines full immersion in the task. Two empirical studies demonstrate the efficacy of using physiological data captured from a wearable sensor to automate the prediction process of flow. Study 1's design utilized a two-level block structure, wherein activities were integrated within the participants themselves. Five participants, to whom the Empatica E4 sensor was attached, were given the challenge of completing 12 tasks that were directly relevant to their personal interests. Summing up the tasks from each of the five participants, a total of 60 was found. GC7 A subsequent study examined the device's practical, everyday use through having a participant wear it during ten different, impromptu activities spanning two weeks. The first study's derived features were examined for their effectiveness when applied to the provided data. The initial study's two-level fixed effects stepwise logistic regression analysis revealed five features to be significant predictors of flow. Two analyses concerning skin temperature were undertaken: the median change relative to baseline and the skewness of the temperature distribution. Three analyses concerning acceleration included the skewness of acceleration in the x and y dimensions, and the kurtosis of acceleration in the y-axis. Classification results for logistic regression and naive Bayes models were excellent, exceeding an AUC of 0.70 in a between-participant cross-validation scheme. Further investigation with the same features produced a satisfactory flow prediction for the new participant wearing the device in a random daily-use setting (AUC greater than 0.7, with leave-one-out cross-validation). The acceleration and skin temperature features seem to effectively track flow in everyday use.
Given the limitations of a single, difficult-to-identify sample image for internal detection of DN100 buried gas pipeline microleaks, a novel method for recognizing microleakage images from internal pipeline detection robots is proposed. To increase the microleakage images of gas pipelines, a non-generative data augmentation approach is first implemented. Next, a generative data augmentation network, Deep Convolutional Wasserstein Generative Adversarial Networks (DCWGANs), is employed to generate microleakage images displaying various features to aid in detection within the gas pipeline system, thus ensuring a wide variety of microleakage image samples from gas pipelines. In the You Only Look Once (YOLOv5) model, a bi-directional feature pyramid network (BiFPN) is implemented to preserve deep feature information by adding cross-scale connections to the feature fusion structure; then, a compact target detection layer is designed within YOLOv5 to retain crucial shallow features for the recognition of small-scale leak points. According to the experimental results, this approach exhibits 95.04% precision for microleak identification, a recall rate of 94.86%, a mean average precision (mAP) score of 96.31%, and a minimum identifiable leak size of 1 mm.
Magnetic levitation (MagLev), a density-focused analytical technique, shows potential in numerous applications. Different MagLev structures with distinct levels of sensitivity and operating distances have been analyzed. The MagLev structures, though theoretically sound, often fail to simultaneously achieve high sensitivity, a wide measuring range, and convenient operation, limiting their practical applicability. This research produced a tunable magnetic levitation (MagLev) system. Through the combination of numerical simulation and experimental testing, the superior resolution of this system, achievable down to 10⁻⁷ g/cm³, is confirmed, exceeding the capabilities of existing systems. Timed Up and Go Likewise, the resolution and range settings of this tunable system can be modified in response to varying measurement needs. Significantly, this system boasts a remarkably simple and convenient operation. The particular traits of this tunable MagLev system suggest its adaptability to diverse density-based analyses on demand, thus significantly increasing the potential applications of MagLev technology.
Rapidly growing research is focused on wearable wireless biomedical sensors. Multiple body-mounted sensors, untethered by local wiring, are frequently required to capture a broad range of biomedical signals. While the creation of multi-site systems with low cost, low latency, and precise data time synchronization is desirable, it presents a currently unresolved issue. Custom wireless protocols and extra hardware are employed in current synchronization solutions, resulting in customized systems with high power consumption, which obstruct migration to different commercial microcontrollers. We pursued the development of a more advanced solution. A low-latency, Bluetooth Low Energy (BLE)-based data alignment method was successfully developed and implemented within the BLE application layer, ensuring its transferability across various manufacturer devices. A trial of the time synchronization method was conducted on two commercial BLE platforms; common sinusoidal input signals (at various frequencies) were input to evaluate the time alignment precision between two separate peripheral nodes. In our analysis of time synchronization and data alignment, we found absolute time differences of 69.71 seconds for the Texas Instruments (TI) platform and 477.49 seconds for the Nordic platform. In terms of 95th percentile absolute errors, their measurements each fell short of 18 milliseconds. Our method, compatible with commercial microcontrollers, is found to be sufficient for numerous biomedical applications.
An innovative indoor-fingerprint-positioning algorithm utilizing weighted k-nearest neighbors (WKNN) and extreme gradient boosting (XGBoost) was developed in this study to overcome the challenges of low accuracy and poor stability associated with traditional machine learning algorithms. An initial step to increase the reliability of the established fingerprint dataset involved the Gaussian filtering of outlier values.