Biometrics are intended to add an extra layer of security to various applications by authenticating a user based on a biometric sample taken from them. At the same time advantage, from the users' perspective, is that they are no longer required to use biometric memorizing credentials. Fingerprinting has been in use since the 19th century in determining identity. Because users' biometric templates are stored within any system, their security becomes extremely sensitive, as theft or simple copying of the biometric can compromise the identity of all users enrolled in the system. This paper presents a quantum-based biometric fingerprint encryption method. The method uses an adapted Novel Enhanced Quantum (NEQR) model to store biometric fingerprint in quantum systems and proposes quantum techniques of Confusion and Feistel algorithm to encryption it. The quantum encryption circuit was implemented and tested both in the IBM Quantum cloud and locally on the QASM quantum simulator.
As the demand for wireless communications continues to grow, securing limited spectrum resources has become an issue. Under such circumstances, Dynamic Spectrum Allocation (DSA), in which existing frequency bands are jointly used and allocated to carriers, has been attracting attention. However, the problem is that a large number of base stations and interferers causes a combinatorial explosion, which requires an enormous amount of computation time. In a previous study, this problem was viewed as a combinatorial optimization problem and formulated to be solved by quantum annealing for the time slot-specific requirements of each base station. However, each base station uses the frequency band differently, and the timeslot length requirement must also be considered. In this paper, we focus on the DSA problem of maximizing the satisfaction of the time-based requirements of each base station and describe a formulation for solving the problem using quantum annealing. These time-based requirements can accommodate the demand for bursty traffic, which was not the case in previous papers. We also present implementation results using Fixstars Amplify and discuss future prospects.
With the development of machine learning technology, as a typical application of the new retail industry, unmanned retail store has developed rapidly. Using deep learning models for retail commodity recognition provides a new way to bring efficient checkout services. In this paper, we use YOLOv5 model for building an efficient retail object recognition system. First, in order to improve the computational efficience and lower the hardware cost, we propose to use an image quality detection module for pre-processing. Blurry images are discarded for improved recognition efficiency. Second, we adopt the convolutional neural network as the back-bone network for feature extraction and object recognition. Third, we implement a real-time recognition system in the lab environment for verification of our proposed system. Experiments on RPC dataset show that the proposed system can detect the location and category of retail goods in real-time. The model performance is promising for unmanned retail applications.
The rapid proliferation of the Internet of Things (IoT) requires the development of effective methods to identify devices over a network. Although traditional machine learning and deep learning approaches have been successful in this task, they require continuous retraining when new devices join the IoT network. To address this challenge, Siamese Neural Networks (SNNs) have been proposed, demonstrating strong performance without the need for model retraining when a new device is added to the network. In this work, we further assess the reliability of SNNs in reducing the necessity for model retraining when multiple devices join the network, as well as their generalization capabilities. Our findings indicate that while performance slightly declines with an increasing number of newly added devices, it remains satisfactory overall. Furthermore, we demonstrate the generalization of SNNs by training the model on one dataset and evaluating its performance when incorporating devices from a different dataset without retraining. The results highlight the robustness of SNNs and their ability to recognize devices on a different IoT network than the one they were initially trained on, and most importantly, without retraining. Finally, we show how the quality of the initial training set can affect the generalization capabilities of SNNs.
Nowadays, Unmanned Aerial Vehicles (UAVs) tend to become an attractable platform for smart city applications, especially for environmental sensing tasks. UAV downwards view image segmentation can extract meaningful information, but the challenge lies in segmenting small-size objects. This paper proposed a reasonable method where edge features were aggregated to enhance the performance of segmentation results, together with a suitable re-weighting method to alleviate the imbalance of interested object class distribution. The effectiveness of the designed network architecture was proven by experimental results that around a 2–4% increase of IoU for aimed small-size classes was achieved. Additionally, the proposed model remains in a lightweight structure.
