Machine learning (ML)-based intrusion detection systems (IDSs) are crucial in safeguarding computer networks against malicious activities. However, building an optimal (accurate and high-performance) ML-based IDS, a combination of data sets, feature selectors, and classifiers, is challenging. This article presents a comprehensive empirical analysis to enhance the effectiveness of IDSs by delving into these three critical components: 1) data sets; 2) feature selection; and 3) classification techniques based on regression models and linear support vector machines (LSVMs), respectively. We begin by evaluating six different data sets commonly used in IDS research, identifying their strengths, limitations, and suitability for real-world scenarios. Next, we explore regression-based feature selectors to identify the most relevant features for intrusion detection, enhancing the accuracy and efficiency of the IDSs. Then, we examine various LSVM classifiers, comparing their performance and highlighting their strengths and weaknesses. By combining these components, this study aims to provide a holistic understanding of the intricate relationship between data sets, regression-based feature selectors, and SVM-based linear classifiers, thus aiding researchers and practitioners in designing more effective and robust IDSs. The empirical analysis conducted in this study employs rigorous evaluation metrics and a comprehensive experimental setup to ensure reliable and unbiased results. The insights gained from our investigation can help guide future research and development efforts toward more efficient and reliable ML-based IDSs.

Federated Learning (FL) trains machine learning (ML) models with privacy protection. However, current FL algorithms use the same hyper-parameters for all clients regardless of their statistical or system heterogeneity, leading to slower convergence. Convergence time and communication rounds may be reduced by using appropriate values for hyper-parameters like learning rate and epochs. We present an adaptive client-side hyper-parameter optimization algorithm, FedAdap, that uses metrics gathered during model training to optimize hyper-parameters on each client and can be used in conjunction with any other FL algorithm. Preliminary results show that FedAdap enhances the performance of existing FL algorithms by decreasing convergence time by up to 34 % and reducing communication rounds by up to 37 % in the case of lID data. Moreover, in non-lID data settings, the convergence time is reduced by up to 82.5 % and the number of communication rounds is reduced by up to 77 %.

To fix network congestion resulting from the increase in high volume traffic in data-intensive science and the increase in internet traffic due to COVID-19, there has been a necessity of traffic engineering through traffic prediction. For this, there have been various attempts from a statistical method such as ARIMA to machine learning including LSTM and GRU. This study aimed to collect and learn KREOENT backbone and subscribers’ traffic volume through diverse machine learning techniques (e.g., SVR, LSTM, GRU, etc.) and predict maximum traffic on the following day.

Slow convergence is a major drawback of average-based consensus time synchronization protocols, particularly in large or sparse wireless sensor networks. The convergence speed can be increased by adding more nodes or increasing the transmission range of nodes, because the network becomes strongly connected. However, these solutions are not always feasible owing to hardware constraints. In this article, a virtual topology-based time synchronization protocol (VTSP) is proposed to address the aforementioned drawback of consensus-based protocols. Notably, VTSP performs the consensus process on a virtual topology that has a stronger algebraic connectivity than a physical one. Therefore, VTSP can significantly accelerate convergence speed without modifying the physical structure of the network. The virtual topology is formed by creating virtual links between each node and its two-hop neighbors. Moreover, two optimization techniques are suggested for reducing data redundancy, which is caused during the creation of the virtual links, in timing messages and for speeding up the convergence by excluding edge nodes from the consensus process. Simulation results demonstrate that VTSP can achieve convergence three times faster than the gradient time synchronization protocol, a well-known consensus-based time synchronization protocol, in various topologies while preserving the same level of accuracy.

Average-based consensus time synchronization protocols have been widely used in wireless sensor networks owing to their robustness against a single point of failures. In these types of distributed protocols, each node is designed to have the same role as other nodes in the network. However, because of this design, these protocols are limited in that any unsynchronized node can negatively affect other nodes in the system. Since the presence of unsynchronized nodes, such as newly joined nodes or malicious nodes is unavoidable, average-based consensus time synchronization protocols are vulnerable in dynamic or hostile environments. In this article, we propose a new protocol named neighbor-aware time synchronization protocol (NTSP) that incorporates the neighbor-aware concept to overcome the aforementioned limitation and improve the robustness of the average-based consensus time synchronization protocol in both dynamic and hostile environments. In particular, by being aware of each neighbor’s status (e.g., synchronized, new, or unsynchronized neighbor), each node can decide whether to include a neighbor in the calculation of the consensus process. As a result, NTSP can prevent adverse effects from unsynchronized nodes. The simulation results demonstrate that NTSP has a relatively faster recovery time than gradient time synchronization protocol and Average TimeSynch when new nodes join and can protect the synchronization of the network from attacks by a malicious node.

Gradient Time Synchronization Protocol (GTSP) is a consensus-based time synchronization protocol in which each node adjusts its logical clock by averaging the relative clock rate of neighbor nodes in every synchronization beacon interval. This process is repeated until the network achieves convergence (synchronization). Therefore, a short beacon interval can reduce the convergence time. However, it causes high energy consumption because messages are sent frequently. On the other hand, a longer beacon interval can reduce energy consumption, but the convergence time can be longer. In this paper, we propose Adaptive Gradient Time Synchronization Protocol (AGTSP), which allows to adjust the beacon interval of each node dynamically in a fully distributed manner to reduce convergence time as well as save energy. A performance evaluation is conducted to prove the effectiveness of AGTSP compared with GTSP.

TDMA protocols can provide a reliable, collision-free data-transferring mechanism for wireless sensor networks. However, it requires an effective scheduling (time slot assignment)algorithm, which is a challenging issue, especially in wireless multi-hop networks due to random-based competition. In this paper, we propose DSTO, a distributed TDMA scheduling algorithm using Topological Ordering (TO). DSTO aims to reduce conflict of scheduling time among neighbor nodes by creating a Topological Order with local neighborhood size as the main priority factor. We implemented DSTO on OPNET Network Simulator and proved its effectiveness compared to DRAND in terms of running time and message overheads.

Air pollution has become a serious environmental problem causing severe consequences in our ecology, climate, health, and urban development. Effective and efficient monitoring and mitigation of air pollution require a comprehensive understanding of the air pollution process through a reliable database carrying important information about the spatiotemporal variations of air pollutant concentrations at various spatial and temporal scales. Traditional analysis suffers from the severe insufficiency of data collected by only a few stations. In this study, we propose a rigorous framework for the integration of air pollutant concentration data coming from the ground-based stations, which are spatially sparse but temporally dense, and mobile sensors, which are spatially dense but temporally sparse. Based on the integrated database which is relatively dense in space and time, we then estimate air pollutant concentrations for given location and time by applying a two-step local regression model to the data. This study advances the frontier of basic research in air pollution monitoring via the integration of station and mobile sensors and sets up the stage for further research on other spatiotemporal problems involving multi-source and multi-scale information.

Software defined networking controller platforms support clustering architectures to provide scalability and availability to large-scale carrier-grade networks. OpenDaylight, a well-known open source project, provides a clustering and distributed datastore architecture. The datastore is distributed into shards such that a subset of the shards can be located in any cluster member. To guarantee strong consistency in the datastore so all shard replicas have the same value, only the shard leader is responsible for accepting data updates. Even though OpenDaylight supports a distributed shard leader election algorithm, elected leaders are not distributed over the cluster members due to a lack of centralized control. Since update requests for the datastore are inherently concentrated in the leader, this results in unbalanced CPU usage of the controller cluster, as well as degradation of throughput in the distributed datastore. In this paper, we propose a shard leader distribution algorithm that maximizes the throughput of the distributed datastore by distributing the shard leaders to cluster members as evenly as possible. The shard leader distribution algorithm restricts the maximum number of leaders that a cluster member may have by monitoring the status of the shards within the cluster members. In the performance evaluation, we prove that shard leader distribution significantly improves the throughput of the datastore by load balancing data update requests to the cluster.

We propose a threshold secret sharing scheme for secure communications in multiple input and multiple output wireless networks. In the proposed scheme, the base station divides the secret data into Nmin parts using a polynomial of degree T − 1 (T ≤ Nmin) and transmits the divided data to the legitimate user by beamforming with multiple spatial dimensions. Then, at the user, the secret data can be reconstructed with a sufficient number (≥T) of divided parts by using the Lagrange interpolating polynomial. However, it is difficult for the eavesdropper to correctly estimate the T parts due to the difference between the main channel with beamforming and the eavesdropping channel in the physical layer, which results in the failure of secret data reconstruction. The numerical results show that the eavesdropping probability of the proposed scheme is lower than those of conventional schemes. Moreover, we analyze the symbol‐error‐rate and show that the theoretical result is well aligned with simulation results.

In recent years, the prevalence of Wi‐Fi‐enabled devices such as smartphones, smart appliances, and various sensors has increased. As most IoT devices lack a display or a keypad owing to their tiny size, it is difficult to set connectivity information such as service set identifier (SSID) and password without any help from external devices such as smartphones. Moreover, it is much more complex to apply advanced connectivity options such as SSID hiding, MAC ID filtering, and Wi‐Fi Protected Access (WPA) to these devices. Thus, we need a new Wi‐Fi network management system which not only facilitates client access operations but also provides a high‐level authentication procedure. In this paper, we introduce a remote connectivity control system for Wi‐Fi devices based on software‐defined networking (SDN) in a wireless environment. The main contributions of the proposed system are twofold: (i) it enables network owner/administrator to manage and approve connection request from Wi‐Fi devices through remote services, which is essential for easy connection management across diverse IoT devices; (ii) it also allows fine‐grained access control at the device level through remote control. We describe the architecture of SDN‐based remote connectivity control of Wi‐Fi devices. While verifying the feasibility and performance of the proposed system, we discuss how the proposed system can benefit both service providers and users.

In this paper, we present an IoT architecture which handles stream sensor data of air pollution. Particle pollution is known as a serious threat to human health. Along with developments in the use of wireless sensors and the IoT, we propose an architecture that flexibly measures and processes stream data collected in real-time by movable and low-cost IoT sensors. Thus, it enables a wide-spread network of wireless sensors that can follow changes in human behavior. Apart from stating reasons for the need of such a development and its requirements, we provide a conceptual design as well as a technological design of such an architecture. The technological design consists of Kaa and Apache Storm which can collect air pollution information in real-time and solve various problems to process data such as missing data and synchronization. This enables us to add a simulation in which we provide issues that might come up when having our architecture in use. Together with these issues, we state r easons for choosing specific modules among candidates. Our architecture combines wireless sensors with the Kaa IoT framework, an Apache Kafka pipeline and an Apache Storm Data Stream Management System among others. We even provide open-government data sets that are freely available.

SDN controller platforms have supported clustering architecture to meet high scalability and availability requirements for large scale carrier grade networks. As a well-known open source project, OpenDaylight provides a clustering and distributed datastore architecture. Datastore is distributed into shards such that a subset of shard can be located in any cluster member. To guarantee strong consistency of datastore that all shard replicas have same value, only the shard leader has responsibility for accepting data updates. Even though OpenDaylight supports a distributed shard leader election algorithm, the elected leaders are not distributed over cluster members due to the lack of centralized control. This results in an unbalanced CPU and network usage of the controller cluster. In this paper, we review the current status of the leader election algorithm for OpenDaylight controller cluster. Then, we propose a shard leader distribution algorithm to fully utilize the distributed datastore. The performance evaluations prove that shard leader distribution enhances the aggregated datastore throughput by distributing the requests for data updates to the distributed datastore.

Summary A number of web applications provide completely automated machine translation services, allowing users to easily translate information of interest. However, these services still generate inaccurate results when translating technical terminologies. Therefore, we propose a new method that collects reliable pairs of English–Korean technical terms and translates the given English terminology to Korean. To collect the pairs, we utilize textual big data, such as Korean academic papers, and develop a new statistical model to determine appropriate characteristics. Our method is evaluated in terms of the reliability of English–Korean pairs and the precision of translation. We thus confirm that our method can produce highly reliable data and can positively influence the translation quality of technical terminologies. Copyright © 2014 John Wiley & Sons, Ltd.

An increasing number of recent experimental works have demonstrated that the supposedly secure channels in the Internet are prone to privacy breaking under many respects, due to packet traffic features leaking information on the user activity and traffic content. We aim at understanding if and how complex it is to obfuscate the information leaked by packet traffic features, namely packet lengths, directions, and times: we call this technique traffic masking. We define a security model that points out what the ideal target of masking is, and then define the optimized traffic masking algorithm that removes any leaking (full masking). Further, we investigate the tradeoff between traffic privacy protection and masking cost, namely required amount of overhead and realization complexity/feasibility. Numerical results are based on measured Internet traffic traces. Major findings are that: 1) optimized full masking achieves similar overhead values with padding only and in case fragmentation is allowed, and 2) if practical realizability is accounted for, optimized statistical masking attains only moderately better overhead than simple fixed pattern masking does, while still leaking correlation information that can be exploited by the adversary.

The ZigBee tree routing is widely used in many resource-limited devices and applications, since it does not require any routing table and route discovery overhead to send a packet to the destination. However, the ZigBee tree routing has the fundamental limitation that a packet follows the tree topology; thus, it cannot provide the optimal routing path. In this paper, we propose the shortcut tree routing (STR) protocol that provides the near optimal routing path as well as maintains the advantages of the ZigBee tree routing such as no route discovery overhead and low memory consumption. The main idea of the shortcut tree routing is to calculate remaining hops from an arbitrary source to the destination using the hierarchical addressing scheme in ZigBee, and each source or intermediate node forwards a packet to the neighbor node with the smallest remaining hops in its neighbor table. The shortcut tree routing is fully distributed and compatible with ZigBee standard in that it only utilizes addressing scheme and neighbor table without any changes of the specification. The mathematical analysis proves that the 1-hop neighbor information improves overall network performances by providing an efficient routing path and distributing the traffic load concentrated on the tree links. In the performance evaluation, we show that the shortcut tree routing achieves the comparable performance to AODV with limited overhead of neighbor table maintenance as well as overwhelms the ZigBee tree routing in all the network conditions such as network density, network configurations, traffic type, and the network traffic.

Tree-based routing protocols in low-rate wireless mesh networks usually have the detour problem in return for the no route discovery overhead. In this paper, we propose a novel algorithm, named Incremental Path Reduction (IPR), which incrementally shortens inefficient detoured path as more data packets are delivered. In IPR, data packets are delivered along the tree route in use by using 1-hop broadcast, enabling neighbor nodes to learn about the data packets' hop count. Using the hop counts, each node estimates their residual hop count to destination. As a result, each forwarder selects next hop node that has small residual hop count. In this way, IPR incrementally shortens the detoured route as more data packets are delivered. To verify our algorithm, we applied IPR to the representative tree routing protocols, and evaluated the path stretch and packet delivery ratio as well as control packet overhead. Simulation results show that IPR significantly enhances the overall routing metrics for any types of tree-based routing protocols.

As wireless sensor networks are becoming more and more commercialized in many applications such as smart home network, building automation system, the needs for network scalability are increasing to support several hundreds of nodes and massive amount of data from them. In this paper, we propose the hierarchical network protocol (HNP) to provide the efficient communication, network reliability, and network management as well as network scalability.

For adopting wireless sensor network in consumer networks, easy configuration and standard based interoperable operations are important. General device-controller approach in service discovery protocols like UPnP allows sensor networks to be discovered and accessed as general UPnP devices via gateways. However, users also need standard capability information that fully describes sensor types, attributes, operations, and calibration to utilize sensor nodes. Integration of IEEE1451 architecture and UPnP network enables self-identification of sensor nodes and "plug and play" capability. In this short paper, we present a management system based on integration of IEEE1451 sensor networks and UPnP.

Service-oriented computing which provides flexible composition of various applications using multiple reusable services has getting more attractive. We propose a service oriented wireless sensor networks toolbox, which encapsulates complexities of sensor networks and enables simple service compositions using an intuitive GUI. We utilize sensor networks as a collection of services that are discovered and coordinated by users. Service metadata which are self-describing service capabilities and interfaces are defined and exposed via our service metadata repositories. A service oriented sensor network toolbox consisting of sensor networks, an intuitive GUI application that retrieves metadata for discovered services and helps users to composite various roles are presented.