With the rise of serverless computing, Function-as-a-Service (FaaS) has become an important paradigm for elastic, on-demand execution. However, cold start latency remains a critical issue that impacts the performance of FaaS platforms. We propose a scheduling strategy called CH-AL, which preserves function locality while ensuring global load balancing. By colocating functions with shared container image layers, CH-AL enhances OS page cache reuse and reduces cold start overhead. To handle heterogeneous workloads, CH-AL employs a dynamic forwarding policy that redirects invocations based on real-time node load, reducing inter-node load imbalance. We implemented CH-AL on OpenWhisk and evaluated its effectiveness through a series of experiments. The results show that CH-AL improves function throughput by 26% compared to OpenWhisk's default load balancer, while reducing average latency and inter-node load variance. Unlike prior methods that depend on runtime feedback from worker nodes, CH-AL leverages historical load statistics maintained solely by the load balancer, offering better scalability and stability.

In this article, we investigate the cluster consensus problem of multiagent systems (MASs) with general linear dynamics and weighted antagonistic interactions under aperiodic denial-of-service (DoS) attacks. First, an event-triggered communication mechanism is designed to efficiently reduce unnecessary message transmission over unreliable networks, where the event-triggered function with dynamically varying thresholds is designed based on the state estimators instead of the real-time states so that continuous communication can be avoided both in controller updates and triggering threshold detection. Then, a novel event-triggered cluster consensus protocol is designed for MASs with DoS attacks and structurally balanced signed digraphs, where Zeno behavior can be strictly excluded by the proposed triggering mechanism. Furthermore, to cope with the structurally unbalanced signed digraph, an improved event-triggered resilient consensus protocol is developed by introducing a pinning control strategy. By utilizing the piecewise Lyapunov functional approach, some sufficient conditions are derived for the cluster consensus under structurally balanced and unbalanced signed digraphs, while the selection principles of event-triggered parameters and resilient controller gains are obtained. Finally, the validity of the theoretical results is verified by practical simulation examples.

Identity-based remote data auditing schemes can verify data integrity and provide a simple identity authentication and management for multiple users. However, prior works on identity-based remote data auditing lack the support of dynamic operations. In these schemes, tag generation is linked to the index of data block, which is related to update operations such as modification, insertion and deletion. If users perform dynamic operations on a data block, the tags of all subsequent blocks need to be modified. It means that if users want to update data on a big data platform, they have to download the whole file, update the file and send the updated file to the big data platform. Such pattern will bring huge communication overhead. In this paper, we propose an identity-based dynamic data auditing scheme which supports dynamic data operations, including modification, insertion and deletion. As far as we know, there is still no other identity-based data auditing scheme that supports dynamic operations. In particular, to achieve efficient dynamic operations, we use the data structure of Merkle hash tree for block tag authentication, which helps update data with integrity assurance. Analyses of security and performance show that the proposed scheme is efficient and secure.

With the development of location-acquisition technologies, there are a huge number of mobile trajectories generated and accumulated in a variety of domains. However, due to the constraints of device and environment, many trajectories are recorded at low sampling rate, which increases the uncertainty between two consecutive sampled points in the trajectories. Our task is to recover a high-sampled trajectory based on the irregular low-sampled trajectory in free space, i.e., without road network information. There are two major problems with traditional solutions. First, many of these methods rely on heuristic search algorithms or simple probabilistic models. They cannot well capture complex sequential dependencies or global data correlations. Second, for reducing the predictive complexity of the unconstrained numerical coordinates, most of the previous studies have adopted a common preprocessing strategy by mapping the space into discrete units. As a side effect, using discrete units is likely to bring noise or inaccurate information. Hence, a principled post-calibration step is required to produce accurate results, which has been seldom studied by existing methods. To address the above difficulties, we propose a novel Deep Hybrid Trajectory Recovery model, named DHTR. Our recovery model extends the classic sequence-to-sequence generation framework by implementing a subsequence-to-sequence recovery model tailored for the current task, named subseq2seq. In order to effectively capture spatiotemporal correlations, we adopt both spatial and temporal attentions for enhancing the model performance. With the attention mechanisms, our model is able to characterize long-range correlations among trajectory points. Furthermore, we integrate the subseq2seq with a calibration component of Kalman filter (KF) for reducing the predictive uncertainty. At each timestep, the noisy predictions from the subseq2seq component will be fed into the KF component for calibration, and then the refined predictions will be forwarded to the subseq2seq component for the computation of the next timestep. Extensive results on real-world datasets have shown the superiority of the proposed model in both performance and interpretability.