Object detection in autonomous driving benefits from federated learning (FL) for privacy, yet real deployments suffer from client heterogeneity and feature-dependent label noise, which hinder convergence and accuracy. We present FedPLC, a noise-robust FL framework with two synergistic components: (i) a candidate-box-aware progressive label correction (PLC) that performs confidence-gated, progressive pseudo-label updates at region level, and (ii) a performance-aware adaptive aggregation that re-weights client updates by validation performance, suppressing noisy clients. On UA-DETRAC with uniform, asymmetric, and feature-dependent noise at various ratios, FedPLC achieves higher mAP and more stable convergence than FedAvg and FL-OD, especially under high and feature-dependent noise.

Vehicular edge computing (VEC) leverages roadside units (RSUs) to provide low-latency and energy-efficient computation for vehicular networks. However, existing offloading schemes still face challenges under high vehicle mobility, dynamic network conditions, uneven RSUs loads, and different Quality of Service (QoS) requirements. To address this, we propose a mobility-aware partial task offloading (MAPRO) framework that partitions tasks for parallel processing across multiple RSUs while considering mobility, network dynamics, and QoS demands. A hybrid framework combining Proximal Policy Optimization (PPO) and numerical optimization efficiently solves offloading, resource allocation, and power control. Experiments on real trajectories show that MAPRO outperforms state-of-theart methods.

Real-time edge video analytics is essential for enabling effective decision support by offloading computation to edge servers. However, traditional methods suffer from limitations: transmission latency faces challenges under constrained network bandwidth, and adaptive configurations often inadequately consider the dynamic variations in video content. To this end, we propose a deep reinforcement learning based scenetarget fusion for content-aware adaptive configuration method (FCAC). Independent of preconceived assumptions about the environment, FCAC makes optimal choices based on experience. First, it extracts dynamic keyframes from original video frames. Then, it performs scene-target fusion to measure content features, obtaining factors affecting resolution, frame rate, and quantization. Finally, we implement and thoroughly evaluate the performance of FCAC in real-world network traces, the results on three real-world scenarios show the advantages of FCAC over state-of-the-art solutions.

IoT streaming applications serve as critical drivers for the artificial intelligence of things (AIoT), characterized by their sensitivity to delays and resource requirements. Placing these application tasks on cooperative edge systems efficiently utilizes edge computing and network resources, enhancing data processing efficiency. However, existing placement methods often oversimplify the network environments or overlook dependencies of tasks, leading to suboptimal performance in real-world scenarios. In addition, the varying Quality of Service (QoS) requirements for IoT streaming applications emphasize the need to make different optimization decisions for applications with different QoS requirements. To address these challenges, we propose DAPNet, a neural adaptive dependent task placement method for IoT streaming applications. We model dependent tasks in an IoT streaming application as a directed acyclic graph (DAG) and then formalize the dependent task placement problem as a multi-objective optimization problem to maximize the IoT streaming application’s QoS. DAPNet adapts resource allocation to meet varying QoS requirements. It uses a deep reinforcement learning algorithm based on Proximal Policy Optimization (PPO) to optimize task placement and resource allocation for IoT streaming applications in dynamic network environments. Simulations with real-world datasets were conducted, comparing our approach with state-of-the-art methods across two network environments, evaluating completion time, energy consumption, and QoS. Results show that our method outperforms existing approaches in all three metrics.

As one of the most critical aspects of mobile ad hoc networks, vehicular ad hoc networks (VANETs) have attracted increasing attention with the growing demand for safety in transportation systems. Trust assessment is crucial in VANETs, as it can effectively identify and mitigate the impact of malicious vehicles, thereby ensuring the reliability and security of network communication. However, the models in most existing trust schemes cannot be dynamically updated over time, failing to adapt to the high dynamics of VANETs. Moreover, they lack protection for data privacy, increasing the risk of information tampering and leakage. To solve the above problems, this paper proposes a new Dynamic Model Update-based Trust Assessment Strategy for VANETs Considering Blockchain (DMTAS-VB). Specifically, DMTAS-VB addresses two primary aspects. Firstly, with respect to the dynamic model, the trust framework proposed in this study incorporates both direct trust and recommended trust models that are dynamically updated as interactions progress. Secondly, regarding privacy protection, blockchain is introduced into the trust model of this scheme. The innovative three-chain structure (MainBC, MesBC, and RepBC) is adopted to achieve functional separation, and core identity authentication, high-frequency message exchange, and reputation management are handled independently. The system can more flexibly optimize the performance parameters of each chain while enhancing data security and privacy protection. Simulation results demonstrate that the proposed scheme outperforms comparative approaches in detection performance under various conditions (vehicle number, vehicle speed, proportion of malicious vehicles) and different attack modes (SA, ZA, BMA, BA, and CA).

3D point cloud are widely used to represent 3D object in many security-crucial domains, such as self-driving and 3D face recognition. Due to the black-box characteristic of 3D point cloud deep neural network (3D DNN), many security concerns are raised. Backdoor attack mainly aims to destroy victim 3D DNN in the training stage by injecting backdoored training data. Recently, a few backdoor attacks for 3D point cloud are designed. Though they achieve promising attack success rate, the caused deformations are severe. We argue that this is because the existing backdoor attacks are too aggressive in order to achieve a higher attack success rate. Besides, another reason is that the methods to implant backdoor triggers are too limited, which causes deformation concentration. By addressing the above issues, we regard 3D transformations as the implanted backdoor trigger which include rotation, scaling, shearing and symmetry at the same time. Furthermore, in order to decide the optimal 3D transformation, we model the backdoor trigger selection process as an optimization problem which pursues high attack success rate and high stealthiness at the same time. Finally, a genetic algorithm is utilized to solve the modeled optimization problem. Extensive experiments suggest that the proposed backdoor attack achieves a competitive attack success rate and the best stealthiness.

Connected and autonomous vehicles produce a substantial amount of data that is essential for implementing advanced and intelligent features. Given the importance and the volume of in-vehicle data, storing it in the cloud for later extraction as critical evidence for vehicle digital forensics is a logical choice. However, ensuring the security of forensic data against tampering and forgery attacks throughout the process is a significant challenge. Existing solutions typically assume that vehicles will generate and upload the in-vehicle data to the cloud honestly. In reality, it may be necessary to prove whether the vehicle has uploaded authentic driving-related data in case of disputes about data authenticity. To address this issue, we propose an in-vehicle digital forensic scheme with public auditing, enabling anyone to perform a public auditing algorithm to check whether the data has been modified. The proposal is based on a process-oriented data integrity proof method that enables a vehicle to generate public verifiable integrity proof. Furthermore, we evaluated the practicality of our scheme by assessing its computational and communication overhead. In terms of computational cost, our proposed scheme demonstrates a power consumption of 0.0385 kWh per 100 km at a speed of 60 km/h. Regarding communication delay, our method exhibits a 50.1% decrease compared to similar approaches.

Scalable service-Oriented Middleware over IP (SOME/IP) is an Ethernet communication standard protocol in the Automotive Open System Architecture (AUTOSAR), promoting ECU-to-ECU communication over the IP stack. However, SOME/IP lacks a robust security architecture, making it susceptible to potential attacks. Besides, random hardware failure of ECU will disrupt SOME/IP communication. In this paper, we propose SISSA, a SOME/IP communication traffic-based approach for modeling and analyzing in-vehicle functional safety and cyber security. Specifically, SISSA models hardware failures with the Weibull distribution and addresses five potential attacks on SOME/IP communication, including Distributed Denial-of-Services, Man-in-the-Middle, and abnormal communication processes, assuming a malicious user accesses the in-vehicle network. Subsequently, SISSA designs a series of deep learning models with various backbones to extract features from SOME/IP sessions among ECUs. We adopt residual self-attention to accelerate the model’s convergence and enhance detection accuracy, determining whether an ECU is under attack, facing functional failure, or operating normally. Additionally, we have created and annotated a dataset encompassing various classes, including indicators of attack, functionality, and normalcy. This contribution is noteworthy due to the scarcity of publicly accessible datasets with such characteristics. Extensive experimental results show the effectiveness and efficiency of SISSA.

The security and privacy challenges faced by Vehicular Ad hoc Networks (VANETs) have led to the development of conditional privacy-preserving authentication (CPPA) schemes. Hardware security modules (HSMs) are seen as a promising solution for implementing these schemes while minimizing the burden on certificate storage. However, existing HSM-based CPPA schemes still have high computation overhead and do not meet the forward security requirements for system secret key (SSK) updates. To address these challenges, we propose an HSM-based lightweight CPPA scheme for VANETs that enjoy low computation costs. Most operations could be performed within the HSM before the message is ready to be signed, reducing real-time computation delay. The scheme also supports SSK updating using an identity-based batch multi-signature algorithm, which helps to provide forward security and vehicle revocation. Especially, the proposed SSK update scheme does not rely on any single trusted authority. Formal proof demonstrates that the proposed scheme satisfies the desired security notions. Our analysis shows that this scheme surpasses other similar ones in terms of efficiency when it comes to generating signatures.

WiFi-based human gesture recognition has recently enjoyed increasing popularity in the Internet of Things (IoT) scenarios. Simultaneously recognizing user identities and user gestures is of great importance for enhancing the system security and user quality of experience (QoE). State-of-the-art approaches that perform dual tasks suffer from increased latency or degraded accuracy in cross-domain scenarios. In this paper, we present WiDual, a dual-task system that achieves accurate cross-domain gesture recognition and user identification based on WiFi in a real-time manner. The basic idea of WiDual is to use the attention mechanism to adaptively explore cross-domain features worthy of attention for dual tasks. WiDual employs a CSI (Channel Statement Information) visualization method that transfers WiFi signals to images for further feature extraction and model training. In this way, WiDual mitigates the possible loss of useful information and excessive delays caused by extracting handcrafted features directly from the WiFi signal. Furthermore, WiDual utilizes a collaboration module to combine gesture features and user identity features to enhance the performance of dual-task recognition. We implement WiDual and evaluate its performance extensively on a public dataset including 6 gestures and 6 users performed across domains. Results show that WiDual outperforms state-of-the-art approaches, with 26% and 8% improvements on the accuracy of cross-domain user identification and gesture recognition respectively.

Mobile video analytics applications, such as smart driving, VR/AR, and video surveillance, have become increasingly popular due to the proliferation of mobile devices. These applications typically use compute-intensive Deep Neural Networks (DNNs) inference in real-time and require high accuracy. Recent studies have shown that edge computing can significantly improve the performance of these applications by offloading the computation, particularly neural network inference, from mobile devices to nearby edge servers. However, offloading continuous video streams to edge servers still faces the challenge of scarce and variable network bandwidth, resulting in high latency for mobile deep vision applications. Existing works often assume sufficient networks and powerful servers to offload all streaming computation to the edge, resulting in unsatisfactory performance in practical scenarios. In this paper, we propose EAVA, an adaptive Edge-Assisted framework on mobile devices designed for Video Analytics that considers a more practical edge situation with an unstable network environment and multiple DNN model choices. EAVA initially partitions video frame and combines mobile devices with powerful edge servers, allowing these frame partitions to parallel perform video analytics algorithms on local devices or edge servers. To handle the intricate network and inference model dynamics, EAVA trains a deep reinforcement learning model to optimize the Quality of Experience (QoE) for mobile deep vision applications, making adaptive configuration choices. Without relying on preconceived assumptions about the environment, EAVA makes optimal choices based on experiences. Finally, we implement and thoroughly evaluate the performance of EAVA using diverse real-world network traces, demonstrating its superior advantages over existing state-of-the-art solutions.

Recent research has brought to light the vulnerability of deep neural networks (DNNs) to adversarial examples. While several methods have been proposed for generating physical adversarial examples, they often suffer from a critical flaw -conspicuous and easily detectable patterns by humans, limiting their real-world effectiveness. To overcome this limitation, we introduce an innovative approach termed "dual adversarial camouflage" (DAC) that generates natural adversarial camouflage in the physical world. Our DAC method leverages natural styles to hide attacks effectively. The process involves a two-stage training process. In the first stage, we learn the style features from style images. Building on this, the second stage optimizes the camouflage obtained in the first stage by minimizing the target detection score, thus significantly enhancing the attack performance. Experiment results show that the adversarial camouflage generated by our method has high naturalness and can effectively deceive object detectors. In practical tests, the attack success rate of our adversarial camouflage in both the digital and physical worlds is impressive, achieving 96.9% and 80% respectively. This showcases the real-world potential and robustness of our DAC method in evading detection.

Vehicular edge computing (VEC) has emerged in the Internet of Vehicles (IoV) as a new paradigm that offloads computation tasks to Road Side Units (RSU) aiming to reduce the processing delay as well as the resource consumption of vehicles. Ideal computation offloading policies for VEC are expected to achieve both low latency and low energy consumption. Although existing works have made great contributions, they rarely consider the coordination of multiple RSUs and the individual Quality of Service (QoS) requirements of different applications resulting in suboptimal offloading policies. In this paper, we present FEVEC, a Fast and Energy-efficient VEC framework with the objective of making the optimal offloading strategy that minimizes both delay and energy consumption. FEVEC coordinates multiple RSUs and considers the application-specific QoS requirement. We formalize the computation offloading problem as a multi-objective optimization problem by jointly optimizing offloading decision and resource allocation, which is a mixed-integer nonlinear programming (MINLP) problem and NP-hard. We propose MOV, a Multi-Objective computing offloading method for VEC, where an improved Non-dominated Sorting Genetic Algorithm-II (NSGA-II) is adopted to obtain the Pareto-optimal solutions with low complexity. Furthermore, the optimal offloading strategy is selected for QoS maximization. Extensive evaluation results based on realistic and simulated vehicle trajectories verify that our proposed algorithm has a better performance compared with the state-of-the-art VEC mechanism.

The Internet of Vehicles (IoV) was proposed as an approach to enable intelligent traffic management and enhance road safety. In order to achieve the intended objective of improving road safety, vehicles are required to constantly broadcast messages to the traffic management infrastructure as well as to other vehicles in the vicinity. Cybersecurity protection of the IoV system is critical as security attacks on IoV and safety-related messages could be life threatening. In this connection, it is essential to ensure the authenticity of IoV messages. Whereas, from the angle of privacy protection, it is undesirable to directly authenticate the identities of vehicles that send the IoV messages. To cope with these conflicting requirements, researchers proposed the notion of conditional anonymous authentication, which aims to authenticate message senders anonymously. When necessary, a trusted third party, named tracer, will be allowed to reveal the true identities of malicious vehicles who sent fake messages. However, existing security techniques, including pseudonyms and group signatures typically assume that the tracer is trusted. This assumption may not be desirable in situations when a curious tracer may reveal the identities of honest vehicles in the IoV system. To address this challenge, this article proposes a privacy-preserving authentication scheme with abuse-resistant tracing. Compared with existing conditional anonymous authentication schemes, our scheme prevents a single tracer from revealing the identity of vehicles. Besides, the tracing key is generated in a distributed manner, and hence no single authority in the system can reveal the true identity of a vehicle.

The emerging IoT stream processing is a key enabling technology for the time-critical IoT applications, which often require high accuracy and low latency. Existing stream processing engines are insufficient to meet these requirements, since they could not integrate and respond timely to variable network conditions in the dynamic wireless environment. Recent efforts focusing on adaptive streaming support user-specified policies to adapt to the variable network conditions. However, those manual-policies can hardly achieve optimal performance across a broad set of network conditions and quality of experience (QoE) objectives. In this paper, we present a Reinforcement Learning-based Adaptive streaming system (RL-Adapt) that is capable of generating adaption policies using RL-strategy and providing declarative APIs for efficient development. RL-Adapt trains a neural network model that can automatically select the optimal policy based on the observed network conditions. RL-Adapt does not rely on pre-defined models or assumptions on the environment. Instead, it learns to make decisions solely through observations of the resulting performance of past decisions. We implemented RL-Adapt and evaluated its performance extensively in three representative real-world IoT applications. Our results show that RL-Adapt outperforms the state-of-the-art scheme, with 20% improvements on average QoE.

Router security defense technologies emerging in recent years could hardly detect and block the new booming threats with unknown signatures such as hardware Trojan, zero-day attacks, etc. We present a novel router defense technology, distributed router shadow, which builds a closed execution environment to deceive attacks entering into the router, thereby misleading the attackers into regarding it as the real attack target and executing the suspicious code to maximize the chances of detonating the system exploit; thus the original router is prevented from attacking and the suspicious code can be detected. Our experiment and analysis show that the router shadow can defend not only attacks with signature but also some new attacks without signature.

Protocol independent multicast-sparse mode can use either a shared tree or a shortest path tree to deliver IPv6 multicast packets, consequently the multicast IP lookup engine requires, in some cases, two searches to get a correct forwarding decision, and this will lead to a new requirement of doubling the lookup speed. The ordinary method to satisfy this requirement in TCAM-based (ternary content addressable memory) lookup engines is to exploit parallelism among multiple TCAMs, however, parallel methods always incur more resources and higher design difficulty. We propose in this paper a approach to solve this problem. By arranging multicast forwarding table in class sequence in TCAM together with using the intrinsic characteristic of the TCAM, our approach can use just one search and a single TCAM to get the right lookup result, while keeping the hardware of lookup engine unchanged. Experimental results have shown that the approach can make it possible for just one TCAM to satisfy forwarding IPv6 multicast packets at the full link rate of 20 Gb/s with the current TCAM chip level.

When transferred in a packet-switched network, the temporal structure of continuous media may be damaged by delay and delay jitter. Compensation buffering is a well-known method to absorb the delay jitter. However, added buffering increases the latency, which may degrade the interactivity between users. As delay and delay jitter are both perceived QoS parameters to users, changing compensation buffer size may result in completely opposite effect on user-level QoS. How to set the buffer size to provide both delay and delay jitter guarantee with preferable user-level QoS? To answer the question, we investigate the effect of buffer size on maintaining the temporal structure of media flows. By performing QoS mapping from network-level to user-level, we prove that there is an optimal buffer size to provide the optimal user-level QoS and obtain the optimal buffer size by differentiating approach. Experiment results validate our studies on the effect of the buffer size.

The ever increasing line speed and the continuous growing demands of various functions support(for example QoS, multicast and security) have interact- tively made it harder for forwarding engines to process packets at line speed, and this will increasingly make the forwarding engines call for additional buffers to accommodate the burst transmission and decrease the packet loss rate. In this paper, a high-speed pipeline designed for TCAM-based forwarding engines is presented, and its buffer analysis model is also given, then, the buffer requirement of the forwarding engine is analyzed under two conditions: the forwarding rate is not less than and less than the input rate. Our analysis results and experiments both show that, the proposed forwarding pipeline is of high performance, and just one pipeline can easily deal with the data transfer rate of 30 Gb/s or even higher; the pipelined forwarding engine only need to buffer a several packets, then the loss rate will be an acceptable value or even zero, further increasing the buffer size will have little effect on reducing the loss rate.

Centralized parallel packet switch algorithm and distributed parallel packet switch algorithm are two typical scheduling algorithms for parallel packet switch. This paper analyzes the two algorithms in detail, addresses several key problems in their implementation and finally presents several available methods and suggestions to make the parallel packet switch more practical