This work presents a proactive computer vision-based ADAS system designed to enhance VRU safety in risk zones at the rear of the vehicle and filter lanes. The system combines object detection, multi-object tracking, and risk assessment to generate real-time proximity alerts. Compared to a detection-only approach, the proposed method shows a 19% improvement in precision and a 10% increase in F1-score, while introducing a latency overhead of less than 1 ms. Evaluated on edge devices, the system maintains efficient performance across resource-constrained platforms. To support research in this domain, we introduce FilterLane-VRU, a new dataset of rearview urban traffic scenarios with temporal proximity annotations. Results validate the feasibility of deploying the system in real-world ADAS applications, offering a reliable and cost-effective solution for VRU safety.

Deep Neural Networks (DNNs) have demonstrated exceptional performance in diverse tasks. However, deploying DNNs on resource-constrained devices presents challenges due to energy consumption and delay overheads. To mitigate these issues, early-exit DNNs (EE-DNNs) incorporate exit branches within intermediate layers to enable early inferences. These branches estimate prediction confidence and employ a fixed threshold to determine early termination. Nonetheless, fixed thresholds yield suboptimal performance in dynamic contexts, where context refers to distortions caused by environmental conditions, in image classification, or variations in input distribution due to concept drift, in NLP. In this article, we introduce Upper Confidence Bound in EE-DNNs (UCBEE), an online algorithm that dynamically adjusts early exit thresholds based on context. UCBEE leverages confidence levels at intermediate layers and learns without the need for true labels. Through extensive experiments in image classification and NLP, we demonstrate that UCBEE achieves logarithmic regret, converging after just a few thousand observations across multiple contexts. We evaluate UCBEE for image classification and text mining. In the latter, we show that UCBEE can reduce cumulative regret and lower latency by approximately 10%–20% without compromising accuracy when compared to fixed threshold alternatives. Our findings highlight UCBEE as an effective method for enhancing EE-DNN efficiency.

Payment Channel Networks (PCNs) have successfully replaced slow global consensus mechanisms with local cryptographic agreements between nodes. As PCN payments heavily depend on network topology for payment routing, strategic node positioning is critical to building cost-effective channels for users and enhancing network robustness against topological attacks. Nevertheless, existing node attachment strategies in the Lightning Network (LN), the most popular PCN, ignore crucial topology issues, such as network centralization and the scarcity of cycles for cheap off-chain rebalancing. In this paper, we first investigate the current state of the LN topology and show that the availability of topology cycles is highly unequal in the network, which exposes the network to several vulnerabilities. Then, we design ProfitPilot, a node positioning strategy that encourages cycle creation in PCNs to reverse the trend in centralization and enable cheap off-chain rebalancing. We compare our proposed algorithm with heuristics available in the Lightning Network and verify that even by focusing on creating cycles, ProfitPilot successfully increases the user’s probability of collecting fees by over 2× while reducing average paying fees. Furthermore, out of all the evaluated heuristics, ProfitPilot presents the fastest increase in network transitivity and mitigates the impact of targeted topological attacks by over 17% compared with the regular Lightning Network operation.

Deep Neural Networks (DNNs) are widely used to solve a growing number of tasks, such as image classification. However, their deployment at resource-constrained devices still poses challenges related to energy consumption and delay over-heads. Early-Exit DNNs (EE-DNNs) address the challenges by adding side branches through their architecture. Under an edge-cloud co-inference, if the confidence at a side branch is larger than a fixed confidence threshold, the inference is performed completely at the edge device, saving computation for more difficult observations. Otherwise, the edge device offloads the inference task to the cloud, incurring overhead. Despite its success, EE-DNNs for image classification have to cope with distorted images. The baseline distortion level depends on the environmental context, e.g., time of the day, lighting, and weather conditions. To cope with varying distortion, we propose Adaptive Early-Exit in Deep Neural Networks (AdaEE), a novel algorithm to dynamically adjust the confidence threshold based on context, leveraging the Upper Confidence Bound (UCB) for that matter. AdaEE provably achieves logarithmic regret under mild conditions. We experimentally verify that 1) convergence occurs after collecting a few thousand observations for images with different distortion levels and overhead values, and 2) AdaEE obtains a lower cumulative regret when compared against alternatives using the Caltech-256 dataset subject to varying distortion.

Flow classification employs machine learning techniques to identify attacks on computer networks. This classification relies on quantitative features that synthesize the information of packets from the same flow. Conventional features, however, such as packet size and the number of bytes, generate redundancies and do not capture the temporal correlations between the packets in a flow. Automated network attacks generate periodic patterns observable through spectral decomposition, which facilitates classification. This paper proposes FENED (Feature Extraction by Network spEctrum Decomposition), a method to extract features from network data. We consider the packet-arrived order within the same flow using the fast Fourier transform for binary classification. The proposed feature vector contains the module of the spectral components of the flow. Experimental results show that FENED outperforms conventional proposals because it extracts features that consider intra-flow packet-arrival order.

Service provisioning in next-generation networks, such as 5G and 6G, relies on virtualization to carry out multi-domain and multi-tenant connections. In these scenarios, virtual network functions (VNF) orchestration becomes susceptible to security threats once trust between peers cannot be assumed. This paper1proposes a blockchain-based system for an agile, secure, and distributed provisioning of virtual network functions in scenarios with multiple administrative domains. Our proposal employs smart contracts to deliver all stages of a service-level-agreement management life cycle automatically. We develop, implement, and evaluate a prototype of the proposed system using smart contracts running on Hyperledger Fabric. The performance evaluation results show that the system guarantees high-rate VNF provisioning, reaching hundreds of slice requests per second in a trustful way.

The multi-resource allocation problem arises in different scenarios. Different mechanisms have been proposed to fairly divide multiple resources, most notably, Dominant Resource Fairness (DRF). Even though DRF satisfies several desirable properties, it considers fairness only in the static setting. We propose Stateful DRF (SDRF), an extension of DRF that looks at past allocations and enforces fairness in the long run while keeping the fundamental properties of DRF. We prove that SDRF is strategyproof, since users cannot manipulate the system by misreporting their demands; incentivizes sharing, because no user is better off if resources are equally partitioned; and is efficient, as no allocation can be improved without decreasing another. In SDRF, users' priorities change over time. To avoid recalculating priorities at every task scheduling decision, we also propose Live Tree, a data structure that keeps elements with predictable time-varying priorities ordered. We implement SDRF on Mesos and run it in a real cluster. Moreover, we conduct large-scale simulations based on Google cluster traces of 30 million tasks over one month. Results show that SDRF reduces users' waiting time on average. This improves fairness, by increasing the number of completed tasks for users with lower demands, with negligible impact on high-demand users.

The Internet of Things (IoT) is based on data collection for future processing and decision making. In multihop Low-Power and Lossy Network (LLN) scenarios, efficient data forwarding in terms of generated traffic and energy consumption is fundamental. This paper revisits the concept of mobile agents to collect data along the agents’s itinerary. The idea is to avoid sending requests to the network when non-expired contents that were opportunistically collected are available in the cache of a central element. In the proposed mechanism, the itinerary is composed of devices of interest and intermediate devices in a closed loop at the origin. Knapsack optimization is used to add unsolicited data opportunistically. The reward is calculated according data popularity. Simulations show that it is possible to reduce network traffic and the energy consumed by devices when compared to the traditional mobile agent data gathering model.

The standard approach adopted by software middleboxes to use multiple cores has long been to direct packets to cores at flow granularity. This, however, has significant shortcomings. First, it is inefficient, since it cannot use all cores when there is a small number of concurrent flows---which happens frequently. Second, asymmetry in flow distribution causes unfairness even with a larger number of flows. Yet, the current trend of higher-speed links and core-richer CPUs only aggravates these problems. In this paper, we propose a natural alternative: that middleboxes should direct packets to cores at a finer granularity. Our system, Sprayer, solves the fundamental problems of per-flow solutions and addresses the new challenges of handling shared flow state that come with packet spraying. Sprayer builds on the observation that most middleboxes only update flow state when connections start or finish; ensuring that all control packets from the same TCP connection are processed in the same core. We show that, when compared to the per-flow alternative, Sprayer significantly improves fairness and seamlessly uses the entire capacity, even when there is a single flow.

Best-performing WLAN monitoring systems must capture as much wireless traffic as possible. To achieve this aim, several monitors are employed to capture wireless exchanges in a target area. Monitors potentially generate large traces that are all merged together to have a more complete, global view of the network behavior. Traces are often more equal than complementary, leading to the underutilization of monitors and to a higher system complexity. In this paper, we propose a methodology to make an efficient use of monitors in order to increase scalability. Such a methodology, based on trace similarity and community detection in graphs, ranks traces to reveal how many and which ones must be merged. Traces at the bottom of the rank, which belong to under-used monitors, are candidates to be relocated somewhere else to extend the target area. We evaluate the proposed methodology in two real-case scenarios. Results show that we can remove up to half of the monitors in our scenarios and still keep the same level of coverage.

A hot topic in data center design is to envision geo-distributed architectures spanning a few sites across wide area networks, allowing more proximity to the end users and higher survivability, defined as the capacity of a system to operate after failures. As a shortcoming, this approach is subject to an increase of latency between servers, caused by their geographic distances. In this paper, we address the trade-off between latency and survivability in geo-distributed data centers, through the formulation of an optimization problem. Simulations considering realistic scenarios show that the latency increase is significant only in the case of very strong survivability requirements, whereas it is negligible for moderate survivability requirements. For instance, the worst-case latency is less than 4 ms when guaranteeing that 80% of the servers are available after a failure, in a network where the latency could be up to 33 ms.

Traffic monitoring and control are becoming more and more important as the number of vehicles and traffic jams grow. Nevertheless, these tasks are still predominantly performed by visual means using strategically placed video cameras. For more effectiveness, proposals to improve traffic monitoring and control should consider automated systems. In this paper, we propose the Collaborative and Opportunistic Traffic Monitoring System (COTraMS), which is a system that monitors traffic using available IEEE 802.11 networks. COTraMS is collaborative because user participation is essential in defining the vehicle movement and opportunistic because it uses existing information. To evaluate the performance of COTraMS, a prototype is implemented using an IEEE 802.11 b/g network. Measurements from a real public wireless network in Rio de Janeiro, Brazil, demonstrate the possibility of obtaining traffic conditions with our proposed monitoring system. In addition, we analyze COTraMS via simulation to evaluate its performance in scenarios with a larger number of vehicles. The comparison of the obtained results with data obtained from Global Positioning System shows high accuracy in detecting both the position of the vehicle and the estimation of the road condition, using a simple architecture and a small amount of network bandwidth.

The network infrastructure plays an important role for datacenter applications. Therefore, datacenter network architectures are designed with three main goals: bandwidth, latency and reliability. This work focuses on the last goal and provides a comparative analysis of the topologies of prevalent datacenter architectures. Those architectures use a network based only on switches or a hybrid scheme of servers and switches to perform packet forwarding. We analyze failures of the main networking elements (link, server, and switch) to evaluate the tradeoffs of the different datacenter topologies. Considering only the network topology, our analysis provides a baseline study to the choice or design of a datacenter network with regard to reliability. Our results show that, as the number of failures increases, the considered hybrid topologies can substantially increase the path length, whereas servers on the switch-only topology tend to disconnect more quickly from the main network.

Traffic monitoring and control is getting more and more important as the number of vehicles and traffic jams steadily grow. Nevertheless, traffic control is still predominantly done by visual means using strategically placed video cameras. To be more effective, proposals to improve the traffic conditions should consider automated monitoring systems. This work proposes an opportunistic system for collaborative traffic monitoring using available IEEE 802.11 networks. Based on the information received by 802.11 beacon frames, vehicles provide the data needed by a central entity to handle and disseminate information about traffic conditions on urban roads, exploiting readily available network resources. Experiments performed with data from a real public wireless network, in Rio de Janeiro, demonstrate the possibility of obtaining traffic conditions with our proposed traffic monitoring method. The system results are close to GPS traces. Moreover, the proposed system facilitates large-scale adoption since it does not require specific hardware.

Xen is a tool for hardware virtualization often used to build virtual routers. Xen, however, does not assure the fundamental requirement of network isolation among these routers. This work proposes XTC (Xen Throughput Control) to fill this gap, and therefore, to guarantee multiple network coexistence without interference. XTC sets the amount of CPU allocated to each virtual router according to the maximum throughput allowed. Xen behavior is modeled by using experimental data, and based on these data, XTC is designed using feedback control. Results obtained in a testbed demonstrate the XTC ability to isolate virtual network capacities and to adapt to system changes.

As of today, many routing protocols for wireless mesh networks have been proposed. Nevertheless, quite a few take the high loss rate of control packets into account. This work analyzes the problem of consistent routing information among wireless network nodes. To accomplish this, we propose a metric to evaluate the level of inconsistency among routing tables. Our experimental analysis demonstrates that the high loss rates seen in indoor environments negatively influence route computation. In addition, we demonstrate that the high network dynamics leads to severe instability in next hop selection. Results show that the effect of loss is significant and that the simple manipulation of routing protocol configuration parameters may be not enough to cope with the problem.

This work proposes the wireless-mesh-network proactive routing (WPR) protocol for wireless mesh networks. Unlike current routing protocols, such as the optimized link- state routing (OLSR), WPR uses a controlled-flooding algorithm tailored to the typical wireless-mesh-network traffic matrix, which concentrates traffic on links close to the gateway. The goal is to improve efficiency by saving network resources and avoiding network bottlenecks. WPR also avoids redundant messages using the AMPR (adapted multipoint relay) set. In this paper, we provide a complexity analysis of the algorithms used by WPR and OLSR. Besides, simulation results show that WPR outperforms OLSR in throughput and packet delivery rate.

A home network is a communication system that aims to interconnect household appliances and share the access to the Internet. This work proposes a novel mechanism which is able to improve the multiple access method of home networks. The Contention window Proactive Increase (CPI) mechanism avoids collisions by increasing the number of times the backoff procedure is called. We applied the CPI mechanism to the IEEE 802.11 and HomePlug standards given their similar access methods. We show the efficiency of the proposed mechanism evaluating through simulations the network throughput gains compared to the original standards. 1.

Powerline communications (PLC) are receiving special attention since they use an already available and ubiquitous infrastructure. The main standard for PLC home networks is HomePlug. This work improves the throughput of HomePlug by modifying the medium access control sub-layer. The key idea is to define a fast collision avoidance mechanism where every station that wants to access the medium increments its contention window after sensing another ongoing transmission. The proposal reduces the number of collisions in the network improving the achievable throughput. We compared our mechanism to the original HomePlug standard through simulation and mathematical analysis. We verified that the improvement is independent from the packet size, the transmission rate and the number of nodes when the network is high loaded