Yazar "Bidollahkhani, Michael" seçeneğine göre listele

Listeleniyor 1 - 3 / 3
  • Küçük Resim
    Öğe
    IMPLEMENTATION OF A LIQUID NEURAL NETWORK CONTROL SYSTEM FOR MULTI-JOINT CYBER PHYSICAL ARM
    (2023-06) Bidollahkhani, Michael
    Technological solutions are being produced to meet people's needs and fulfill their desires in a comfortable way. As technology becomes cheaper, more widespread, smaller in size, and able to operate independently from the power grid, the communication of devices with each other (Internet of Things) and the ability of devices to make their own decisions increase the effectiveness of solutions. In particular, the reduction in device size can be achieved by requiring less system resources and battery capacity. Therefore, existing methods need to be customized to work effectively in embedded systems. In this thesis a novel approach called LTC-SE, which enhances the Liquid Time-Constant Neural Network (LTC) technique for embedded environments with limited processing capabilities and strict performance requirements is presented. LTC-SE combines various neural network paradigms, including Leaky-Integrate-and-Fire (LIF) spiking neural networks, Continuous-Time Recurrent Neural Networks (CTRNNs), Neural Ordinary Differential Equations (NODEs), and customized Gated Recurrent Units (GRUs), resulting in improved adaptability, interoperability, and structural organization. In the thesis, a unified class library, is developed, called LTCCell that offers extensive configurability CTRNN, NODE, and CTGRU elements. The proposed method is evaluated by developing a control system for a multi-joint cyber-physical arm, demonstrating its effectiveness in achieving designated objectives and manipulating objects securely. The system's performance is presented through a decision support framework and multi-variable benchmarking, emphasizing the benefits of our refinements in terms of user interaction, functional coherence, and code clarity. Furthermore, the LTC-SE technique expands the scope of liquid neural networks, finding applications in diverse machine learning domains such as robotics, causality assessment, and time-series forecasting. This thesis presents innovative contributions to the field based on the pioneering work of LTC neural network.
  • Küçük Resim Yok
    Öğe
    LoRaline: A Critical Message Passing Line of Communication for Anomaly Mapping in IoV Systems
    (Ieee-Inst Electrical Electronics Engineers Inc, 2023) Bidollahkhani, Michael; Dakkak, Omar; Alajeeli, Adnan Saher Mohammad; Kim, Byung-Seo
    The importance of road safety is felt nowadays more than ever, where various technologies, including self-driving cars, have become abundant. Nowadays, it has more demand to build autonomous and electrical vehicles with information retrieval systems within the received sensory data not only from the local sensors but also the online and live streaming data over networks. To increase road safety dissemination of critical information, including the possibility of an obstacle or danger being in the middle of the road, automotive navigation and control systems are required. A novel method is proposed to make this critical communication possible over a specially designed vehicular ad-hoc network, where natural or urban barriers can prevent signal propagation. The network is implemented using the LoRaWAN interface and SX127x LoRa Radio module. The SX1272MB2xAS is fitted with the SX1272 transceiver, which added to a high-performance FSK/OOK RF transceiver modem. Additionally, LoRa long-range modem provides highly power-efficient communication. For this aim, two new mechanisms have been proposed. The first mechanism enables the nodes to receive data from a suggested communication link. While the second mechanism is designed to extract vital information such as establishing the connection, closing the connection, successful data transmission, errors, etc. The findings demonstrate that the proposed mechanisms have successfully enabled LoRaWAN to operate in IoV environment. The evaluation reveals that metrics such as battery consumption and covering range outperform similar technologies. Finally, this paper proposes a message-passing strategy based on Belief Propagation (BP) which provides more accurate marginal probabilities to overcome the low data rate as a foundation for our future work.
  • Küçük Resim Yok
    Öğe
    Real-Time Building Management System Visual Anomaly Detection Using Heat Points Motion Analysis Machine Learning Algorithm
    (Univ Osijek, Tech Fac, 2023) Avci, Isa; Bidollahkhani, Michael
    The multiplicity of design, construction, and use of IoT devices in homes has made it crucial to provide secure and manageable building management systems and platforms. Increasing security requires increasing the complexity of the user interface and the access verification steps in the system. Today, multi-step verification methods are used via SMS, call, or e-mail to do this. Another topic mentioned here is physical home security and energy management. Artificial intelligence and machine learning-based tools and algorithms are used to analyze images and data from sensors and security cameras. However, these tools are not always available due to the increase in data volume over time and the need for large processing resources. In this study, a new method is proposed to reduce the usage of process resources and the percentage of system error in anomaly detection by reducing visual data to critical points by using thermal cameras. This method can also be used in energy management using home and ambient temperature and user activity measurements. The statistical results of the visual comparison between the proposed method and the legacy CCTV -based visual and sensory surveillance shown in the results section demonstrate its reliability and accuracy.