With the development of smart agriculture, the Agricultural Artificial Intelligence Internet of Things (Agri-AIoT) has shown great potential in fields such as farmland monitoring and precision pesticide application. However, the computationally intensive tasks generated by massive heterogeneous agricultural sensing devices pose serious challenges to real-time and sustainability. To address this issue, this paper proposes a Cloud-Fog-Edge based collaborative computing framework, decoupling the problem into unmanned aerial vehicles (UAVs) deployment and task scheduling two sequential problems. Specifically, a hierarchical optimization framework is proposed with the goal of optimizing system latency and energy consumption. In network deployment stage, a semi-supervised K-Means based UAV deployment algorithm (SKm-UD) is designed, while an improved multi-agent deterministic policy gradient (MTD 3 PG) strategy is proposed in the distributed task offloading phase, which integrates dual delay network, a lightweight local policy update mechanism, and an adaptive learning rate adjustment strategy based on zebra optimization (ZOA-L) to support efficient computation offloading decisions in partially observable environments. Simulation results show that our proposed method can quickly converge to the optimal strategy, with network energy consumption decreases by at most 30.1%. In addition, the scalability of the mechanism in large-scale Agri-AIoT scenarios is also validated.

Including TOR, most of the anonymous communication systems adopt source routing, that the source has to share the globally consistent view of all relays and maintain the up-to-date information. To increase the scalability of TOR, researchers mainly utilize hop-by-hop routing during circuit extension. However, hop-by-hop routing has not been widely deployed since it suffers from route capture attacks, and most of the countermeasures require the source participate in the route extension indirectly, help verify the selection of next hop by intermediate nodes, thus introduces communication overhead. In this paper, we introduce a novel routing scheme called Jump Routing. In jump routing, the route extension follows the jumping way, that each relay chooses the successor of the next hop rather than the next hop itself. In particular, to the best of our knowledge, we are the first to route in the jumping way. In addition, to defend route capture attacks, enhance data privacy, and defend collusion attacks, we propose multiple schemes including jump verification, jump encryption, and corporative jump verification. Different from previous measures on route capture attacks, jump routingdoes not need the participation of the source, but deals with the attack by intermediate nodes only. We manage to realize the full jump routing prototype, and the evaluation results show that our jump routing is scalable, lightweight, and resilient.

Multipath packet scheduler is responsible for deliv-ering each packet to an appropriate path. However, rules-based schedulers struggle to adapt to varying network conditions and diverse Quality of Service (QoS) requirements. Despite learning- based scheduler can adapt to various network conditions, reacting quickly to network changes is still challenging. Moreover, for diverse QoS requirements, learning-based schedulers often neces-sitates training from scratch. To solve the above challenges, we propose a multi-head mapping model that selects an optimal sub- scheduler based on the current state. It works with a shared state encoder, a multi-head Q-value decoder, and QoS-oriented reward decomposition. Furthermore, we propose Aquilas, an adaptive multi path packet scheduler with hierarchical intelligence for the Quick UDP Internet Connection (QUIC) protocol. Aquilas adopts a learning-based scheduler selector, thereby enabling selection of an optimal policy from a pool of sub-schedulers. This sub- scheduler pool encompasses a wide range of knowledge for handling various network conditions. In coarse time intervals, the learning-based scheduler selector operates, while during fine- grained time intervals, the selected sub-scheduler delivers each packet to the corresponding path. Aquilas has been evaluated in both controlled emulation and real-world networks. Compared with the state-of-the-art schedulers, Aquilas improves transmis-sion performance in various network conditions and diverse traffic types.

The popularization of video streaming brings challenges in satisfying diverse Quality of Service (QoS) requirements. The multipath extension of the Quick UDP Internet Connection (QUIC) protocol, also called MPQUIC, has the potential to improve video streaming performance with multiple simultaneously transmitting paths. The multipath scheduler of MPQUIC determines how to distribute the packets onto different paths. However, while applying current multipath schedulers into MPQUIC, our experimental results show that they fail to adapt to various receive buffer sizes of different devices and comprehensive QoS requirements of video streaming. These problems are especially severe under heterogeneous and dynamic network environments. To tackle these problems, we propose MARS, a Multi-agent deep Reinforcement learning (MADRL)-based Multipath QUIC Scheduler, which is able to promptly adapt to dynamic network environments. It exploits the MADRL method to learn a neural network for each path and generate scheduling policy. Besides, it introduces a novel multi-objective reward function that takes out-of-order queue size and different QoS metrics into consideration to realize adaptive scheduling optimization. We implement MARS in an MPQUIC prototype and deploy in Dynamic Adaptive Streaming over HTTP system. Then, we compare it with the state-of-the-art multipath schedulers in both emulated and real-world networks. Experimental results show that MARS outperforms the other schedulers with better adaptive capability regarding the receive buffer sizes and QoS.

Recent research has sought to improve fuzzing performance via parallel computing. However, researchers focus on improving efficiency while ignoring the increasing cost of testing resources. Parallel fuzzing in the distributed environment amplifies the resource-wasting problem caused by the random nature of fuzzing. In the parallel mode, owing to the lack of an appropriate task dispatching scheme and timely fuzzing status synchronization among different fuzzing instances, task conflicts and workload imbalance occur, making the resource-wasting problem severe. In this paper, we design UltraFuzz, a fuzzer for resource-saving in distributed fuzzing. Based on centralized dynamic scheduling, UltraFuzz can dispatch tasks and schedule power globally and reasonably to avoid resource-wasting. Besides, UltraFuzz can elastically allocate computing power for fuzzing and seed evaluation, thereby avoiding the potential bottleneck of seed evaluation that blocks the fuzzing process. UltraFuzz was evaluated using real-world programs, and the results show that with the same testing resource, UltraFuzz outperforms state-of-the-art tools, such as AFL, AFL-P, PAFL, and EnFuzz. Most importantly, the experiment reveals certain results that seem counter-intuitive, namely that parallel fuzzing can achieve “super-linear acceleration” when compared with single-core fuzzing. We conduct additional experiments to reveal the deep reasons behind this phenomenon and dig deep into the inherent advantages of parallel fuzzing over serial fuzzing, including the global optimization of seed energy scheduling and the escape of local optimal seed. Additionally, 24 real-world vulnerabilities were discovered using UltraFuzz.

Network technology is the basis for large-scale high-efficiency network computing, such as supercomputing, cloud computing, big data processing, and artificial intelligence computing. The network technologies of network computing systems in different fields not only learn from each other but also have targeted design and optimization. Considering it comprehensively, three development trends, i.e., integration, differentiation, and optimization, are summarized in this paper for network technologies in different fields. Integration reflects that there are no clear boundaries for network technologies in different fields, differentiation reflects that there are some unique solutions in different application fields or innovative solutions under new application requirements, and optimization reflects that there are some optimizations for specific scenarios. This paper can help academic researchers consider what should be done in the future and industry personnel consider how to build efficient practical network systems.