Unmanned Aerial Vehicle (UAV) technology has been widely applied in both military and civilian applications. With the increasing complexity of application scenarios, the coordination of multiple UAVs has become a hot topic. However, the limited capability of UAVs make it hard to achieve stable and reliable control. Considering this practical problem, we propose a cloud-based UAV system. It extricates the computing and data storage from UAVs and utilizes the cloud to process the sensor data and to maintain the stable operation of multi-UAV systems. Firstly, we analyze the cloud-based system's on-demand service ability and its impact on UAVs' control procedure. Secondly, we propose a UAV cloud control system (CCS) which serves as a network control system. Moreover, the stable condition of the UAV cloud control system is derived. It reveals the relationship between the acquisition rate of sensor data and the stability of the cloud-based UAV system. Finally, simulations are conducted to verify the effectiveness of previous theoretical analysis.
Relying on mini wearable or implantable biosensors, the wireless body area network (WBAN) is capable of efficiently collecting as well as of analyzing human physiological information. It has shown great potential in terms of beneficially improving healthcare quality. However, due to stringent resource constraints of biosensors, traditional security schemes, i.e. the encryption and the authentication, may not do well in countering security threats. Moreover, they are not competent in protecting the network from inside attacks and deny of service (DoS) attacks. In this paper, we propose a sink node assisted lightweight intrusion detection mechanism for WBAN, where the sink node can periodically monitor the packet transmission and record the abnormality for further analysis. Our lightweight mechanism results in a very high true positive rate and an ultra-low false positive rate. Extensive analysis and simulations based on Castalia are conducted and verify the validity and efficiency of our proposed mechanism.
Card games are regarded as an idealized model for many real-world problems for their rich hidden information and strategic decision-making process. It provides a fertile environment for artificial intelligence (AI), especially reinforcement learning algorithms. With the boom of deep neural networks, increasing breakthroughs have been made in this challenging domain. Koi-Koi is a traditional two-player imperfect-information playing card game. However, due to its unique deck and complex rules, related researches are mostly based on handcrafted features and the custom network architecture. In this paper, we design a more general AI framework, relying a Transformer encoder as the network backbone with tokenized card state input, which is trained by Monte-Carlo reinforcement learning with phased round reward. Experimental results show that our AI achieves a winning rate of 53% and +2.02 average difference point versus experienced human players in multi-round Koi-Koi games. Moreover, with the aid of attention mechanism, we provide a novel view for analyzing the playing strategy. Such framework design can be applied to various card games. This project is available at https://github.com/guansanghai/KoiKoi-AI .