Earth observation systems play important roles in many critical applications. The rapid increase of the number of satellites and their sensing capability, however, makes it challenging to send the massive amount of observed data back to the Earth. One promising direction to enhance the earth observation throughput is to use inter-satellite communication. Towards this, we identify two key design factors: 1) the capability to support on-demand scheduling of inter-satellite communication; and 2) the capability to co-optimize the scheduling of observation and transmission missions. For both, rigorous study is needed to determine whether they provide sufficient throughput gain to justify their additional complexity. Our work formulates a generic earth observation and transmission problem to study the maximum network throughput under different settings. By succinctly modeling the different constraints using a generalized time-varying graph representation, we can efficiently find the optimal scheduling solutions. We conduct an extensive study, which shows that using 40 relay satellites from the “starlink” constellation can increase the throughput of 10 sensing satellites from the “Gaofen” constellation by more than 400%. In particular, on-demand scheduling under heavy load and co-optimization of observation/transmission under light but time-critical load can improve the throughput by more than 180% and 100%, respectively.

Traffic classification occupies a significant role in cybersecurity and network management. The widespread of encryption transmission protocols such as SSL/TLS has led to the dominance of deep learning based approaches. In cybersecurity, strong adversaries often complicate their strategies by constantly developing emerging attacks. Meanwhile, security practitioners desire to grasp the reasons for inference results. However, existing deep learning approaches lack efficient adaptation for incremental traffic types and often have less interpretability. In this paper, we propose I $^{2}$ RNN, an Incremental and Interpretable Recurrent Neural Network for encrypted traffic classification. The I $^{2}$ RNN proposes a novel propagation process to extract the sequence fingerprints from sessions with local robustness. Meanwhile, this proposal provides interpretability including time-series feature attribution and inter-class similarity portrait. Moreover, we design I $^{2}$ RNN in an incremental manner to adapt to emerging traffic types. The I $^{2}$ RNN only needs to train an additional set of parameters for the newly added traffic type rather than retraining the whole model with the entire dataset. Extensive experimental results show that our I $^{2}$ RNN can achieve remarkable performance in traffic classification, incremental learning, and model interpretability. Compared with other local interpretability methods, our I $^{2}$ RNN exhibits excellent stability, robustness, and effectiveness in the interpretation of network traffic data.

The group testing problem consists of determining a small set of defective items from a larger set of items based on tests on groups of items, and is relevant in applications such as medical testing, communication protocols, pattern matching, and many more. While rigorous group testing algorithms have long been known with runtime at least linear in the number of items, a recent line of works has sought to reduce the runtime to poly(k log n), where n is the number of items and k is the number of defectives. In this paper, we present such an algorithm for non-adaptive group testing termed bit mixing coding (BMC), which builds on techniques that encode item indices in the test matrix, while incorporating novel ideas based on erasure-correction coding. We show that BMC achieves asymptotically vanishing error probability with O(k log n) tests and O(k2 · log k · log n) runtime, in the limit as n → ∞ (with k having an arbitrary dependence on n). This closes an open problem of simultaneously achieving poly(k log n) decoding time using O(k log n) tests without any assumptions on k. In addition, we show that the same scaling laws can be attained in a commonly-considered noisy setting, in which each test outcome is flipped with constant probability.

Automatic train operation systems of high-speed trains are critical to guarantee operational safety, comfort, and parking accuracy. However, implementing optimal automatic operation control is challenging due to the train's uncertain dynamics and actuator saturation. To address this issue, this paper develops a data-driven Koopman model based predictive control method for automatic train operation systems. The proposed control scheme is designed within a data-driven framework. First, using operational data of trains and the Koopman operator, an explicit linear Koopman model is built to characterize the train dynamics. Then, a model predictive controller is designed based on the Koopman model under comfort and actuator constraints. Furthermore, an online update mechanism for the Koopman model is developed to cope with the changing dynamic characteristics of trains, which reduces the accumulation errors and improves control performance. Stability analysis of the closed-loop control system is provided. Comparative simulation results validate the effectiveness of the proposed control approach.