As an essential part of the website, JavaScript greatly enriches its functions. At the same time, JavaScript has become the most common attack payload on malicious websites. Although researchers are constantly proposing methods to detect malicious JavaScript, the emergence of obfuscation technology makes it difficult for previous approaches to detect disguised malicious JavaScript effectively. To solve this problem, we find that there are fixed templates for generating obfuscated code, which makes the original and obfuscated script have a mapping relationship in their structure. The structure information of the code is critical for malicious detection. Therefore, this paper proposes TransAST, a novel static detection method for obfuscated malicious JavaScript. Our approach's key is restoring the obfuscated JavaScript structure information by training the machine translation model. The experiment shows it can achieve 91.35% accuracy and 94.57% recall in the public dataset, which is 5.5% and 10.94% higher than the existing optimal method.

For multi-user satellite Internet of Things (IoT) systems operating at lower signal-to-noise ratio, spread spectrum techniques are usually used to combat narrowband interference. In addition, the communication performance in the spread spectrum system depends on the anti-jamming ability of the spreading codes (SCs). Therefore, how to design the SCs scheduling strategies under users' requirements and resource constraints has become a crucial problem for satellite IoT systems. In this paper, communication rewards and scheduling delays are introduced as gauges to measure the scheduling performance of the satellite gateway station control center (SGSCC). Specifically, SGSCC must efficiently and effectively allocate limited available SCs over terminal gateways under request at each transmission time slot. The SCs scheduling problem is formulated as a Markov Decision Process (MDP) along with the observed environments composed of resource status and user request status. Then a deep reinforcement learning scheduling algorithm is devised by embedding the idea of Long Short-Term Memory (LSTM) in the standard Double Deep Q-learning (DDQN). Simulation results show that the proposed algorithm can achieve much better performance than traditional algorithms in terms of communication rewards and scheduling delays. Finally, we draw some conclusions.