A lot of the devices in the Internet of Things are sensors responsible for capturing environmental information and relaying it to a system or network. Oftentimes, it is important to know the location of these sensors to better contextualize the information received from them. However, the sensors are purposely simple and cheap, meaning that conventional localization techniques, such as global navigation satellite systems are not feasible. Research has been done on using unmanned aerial vehicles to estimate the location of the sensors, but issues with signal strength fluctuation and location approximation because of it are still prevalent. In this work, we propose a new method for estimating the location of sensors by exploiting the characteristics of radio wave signal propagation and creating a new flight path design, a solution to minimize the impact of variation in measurements, a novel candidate point generation through signal strength analysis, and a method to find the location based on the candidates. Through a thought-out experiment, we show that the proposed algorithm is overall significantly better than the existing solution when it comes to identifying the location of outdoor sensors.

Recognizing Customer Behavior (CB) from videos of instore cameras is important to smart retail solutions. Because of possible changes in retail needs and environments, a high degree of adaptability to different target CBs is required for Customer Behavior Recognition (CBR) methods. Existing CBR methods are mainly machine learning based models due to their remarkable recognition accuracy. However, trained models are not reusable for different target CBs. Consequently, existing CBR methods are hard to adapt to different target CBs because the necessary recollecting data and retraining models. In this paper, we propose a CBR method that recognizes CBs by combinations of primitives, each of which represents an object’s motion or objects’ relationship. Since primitives can be reused in combinations for various CBs, the proposed method is easily adaptable to changed target CBs. Experiments on two datasets indicate the good adaptability and sufficient recognition accuracy of our method.

This paper proposes a Software-Defined Network (SDN)-based Moving Target Defense (MTD) mechanism to protect the network from potential scans. The proposed mechanism can work in combination with an IPS without affecting its normal behavior. To do so, an SDN controller changes the packets' headers passing through switches using virtual IP addresses while the operation of IPS continues monitoring the devices' actual IP addresses. Preliminary results in an emulated environment show that it is possible to achieve a seamless collaboration between the MTD and IDS to detect low and high-rate scans.

SDN is an emerging network paradigm whose main characteristic is the separation of the Control from the Data Plane, allowing the implementation of innovative, robust, and flexible ways to program networks. Moreover, SDN has recently extended its coverage to Hybrid wired and/or wireless environments. However, SDN-enabled devices must be connected to a controller (i.e., a central entity) to be programmable. Therefore, if multiple controllers fail in a short period, regardless of the ample resource availability in Hybrid SDN environments, the whole network infrastructure, and the overall service are compromised. This article presents a mechanism capable of handling multi-controller failure in SDN by considering the hybrid nature of the infrastructure and the scale of the failure. Preliminary results show that by applying the proposed mechanism, it is possible to increase the device's controller coverage by up to 70% even if half of the controllers are unavailable compared to conventional approaches.

Software Defined Networking (SDN) has recently drawn attention by changing the network control paradigm. SDN devices handle the network traffic based on entries stored in their flow table. However, due to the flow table’s limited size, attacks such as DDoS can cause the table to overflow. Some of the current table overflow mitigation methods distribute flow entries to other network devices on the physical infrastructure, which might not be effective in small to medium-sized private networks. Therefore, this research proposes a method to mitigate table overflow attacks on SDN by dynamically filtering attackers using Network Function Virtualization (NFV). Preliminary results show that the proposed approach can effectively mitigate table overflow attacks without relying on network infrastructure.

Due to the diversification of devices and the spread of the Internet, many people are handling information over the network and the threat of cyber-attacks by unauthorized access and malware is also considered as a problem, and countermeasures against these threats are an urgent issue. SDN and OpenFlow are useful for network cybersecurity countermeasures because they can collect traffic statistics from switches and can interrupt communication at the switch level. In this study, we focus on detecting a port scanning as a preparation stage for cyber-attacks. In this paper, we propose port scan detection method based on the characteristics of Packet-In messages in an OF network. Also, we show its effectiveness through the experiments.

There are various information flows in the network because of the diverse ways in which companies and individuals communicate. Therefore, OpenFlow is attracting much attention since it can flexibly deal with various types of information flow requirements through centralized management of devices. However, whenever is necessary to change routes due to a network failure or a topology change, there is a delay in the service as processes, such as the transmission of the commands and switch processing time. In this paper, we describe the design a method to control information flow effectively based on various network switch costs such as route switching time in OpenFlow.

In this paper, we propose a network control method considering the disaster risk of OpenFlow control and management plane (C/M-Plane). We consider the disaster risk of both data plane (D-Plane) and the C/M-Plane. Even if the C/M-Plane is damaged and the switch becomes uncontrollable, the switch operates in the state at that time. Our method controls the entire network to keep communication according to it to maintain data transfer even if some switches become uncontrollable.

Due to the increasing need for timely and reliable access to user-generated content, Distributed Storage Systems (DSS) became more relevant in the past years. However, since they have more interconnections among layers compared to traditional network applications, load imbalance issues arise. In this paper, we propose a hybrid approach combining server and link load balancing for multipath routing in DSS. The approach is Software Defined Networking (SDN) based, and uses a process we call on-demand inverse multiplexing. Preliminary results show that, by applying the proposal, the overall throughput considerably increases and resource usage remain balanced.

The use of Distributed Storage Systems (DSS) has considerably increased in the past years, alongside the need for effective data transfer from storage to storage. Although current network infrastructure can reliably handle large amounts of traffic, networking techniques have not changed for several years, leading to an under-use of resources, i.e. most routing solutions still use single-path routing. In this paper, we present a pragmatic approach for multipath routing in DSS, which is based on Software Defined Networking (SDN) that uses parallel links at the edge-side. Path discovery is calculated by finding the k-maximum disjoint paths in a multigraph. Preliminary results show that, by using our multipath solution, not only the overall throughput increases but also the efficiency of resources usage.

IoT environment has spread and the use of Internet services during move has increased recently. The demand for mobility management, the technology to keep communication even when a communicating mobile node moves, is growing due to it. We proposed a SDN based mobility management scheme in our prior work to deal with inter-domain handovers, but its routing algorithm only considers the number of domains to go through and the number of flow entries to install. This prevents us from selecting suitable routes. In this paper, we show the design of a new routing mechanism that considers various parameters. This mechanism suppresses costs for changing route to carry out handover smoothly and avoids lengthy routes.

In this paper, we propose a mobility management architecture based on Software Defined Network (SDN) for multiple domain networks. In this architecture, each SDN controller cooperates with each other to optimize communication route covering multiple domain networks for seamless IP services even when the mobile devices move to other domain. This paper presents the basic design of the architecture and protocol to cover wide area for mobility management. We also show the initial experiments to confirm the behavior of our proposed architecture.

We propose a novel background subtraction method for robust region extraction of moving objects in the dynamic background. In our method, a set of recently observed frame (reference) images is used as a background model. To withstand the constant fluctuations of background appearance, a current frame (input) image is compared with every reference image, and pixels in the input image deviated from those in the reference images are extracted as moving object regions. Such a simply-structured background model enables our method to adaptively vary the size of a spatial block for comparing the input image with the reference images. Setting an appropriate size block for each subject pixel in the input image, our method increases the stability of image comparison, and then improves the robustness of region extraction. Experimental results indicate that our method outperforms the existing methods in the region extraction accuracy of moving objects in the dynamic background.

Voxel based multiview stereo is a method for reconstructing a 3D scene from a set of images taken at different viewpoints by determining the voxels corresponding to object boundaries in a voxel space. To accelerate the voxel based multiview stereo, we propose a novel parallel method based on the voxel space division approach, which divides the voxel space into several domains and assigns them to multiple Processing Elements (PEs). The existing methods based on this approach cannot work well for viewpoints arranged in complex configurations, since their process decomposition increases communication between PEs. Unlike those methods, we take new process decomposition, where the calculation for every voxel is decomposed into its component for each viewpoint. Our method processes the calculations for each of two viewpoints in parallel applying the voxel space division, and then integrates their results. This method can reduce the amount of inter-PE communication for arbitrarily configured viewpoints.

To carry out a visual examination for chromosome abnormalities, individual chromosome regions need to be determined in a subject image and classified into distinct chromosome types in advance. We have proposed a subregion based approach to improve this process. The proposed approach regards each chromosome region as a series of subregions and iterates searches for subregions in the subject image consecutively. To make this subregion based approach fit for practical use, in this paper, we propose an effective method of searching for subregions in chromosome images and present the results of subregion search experiments.

This paper proposes a parallelization method for voxel-based 3D shape reconstruction. The voxel-based technique offers accurate 3D shape reconstruction by integrating information obtained from multiple images; however, it requires huge computation time for the reconstruction. Our main approach is parallel processing by voxel space division, which divides the voxel space by planes along with the lines passing through a virtual view point and assigns the process for each divided voxel space to a processing element (PE) individually. This makes it possible to reduce the amount of inter-PE communication. In addition to this, we introduce a method to decrease the amount of computation by controlling iteration count of iterative process for each voxel. Experimental results show that our method more efficiently reduces the computation time than simple horizontal domain division method.

To make a visual examination of various chromosome abnormalities, individual chromosome regions have to be determined in an image and classified into distinct chromosome types in advance. We propose a subregion based method to improve this process. The proposed method regards each chromosome region as a series of subregions, and iterates a search for subregions in the subject image consecutively. In this method, chromosome region classification can be performed simultaneously with its determination for each subregion, and the features in subregions can be integrated effectively for recognizing an entire chromosome region.

To make the visual examination of a chromosome image for various chromosome abnormalities, individual chromosome regions have to be extracted from the image and classified into the distinct chromosome types in advance. To improve the accuracy and flexibility in this process, a subregion (local band pattern) based method has been proposed for recognizing individual chromosome regions in the image. This method regards each chromosome region as a series of subregions, and iterates a search for subregions in the image consecutively. As a result, chromosome region classification is performed simultaneously with its extraction for each subregion. Since the dimensions and intensities of chromosome regions vary with every image, effective subregion searches require templates whose dimensions and intensities correspond with those of chromosome regions in the subject image. In this paper, to develop an efficient subregion search, we present a method for adjusting the dimensions of templates to those of chromosome regions in the subject image and adapting the intensities in the subject image to those of the templates.

We have been working on construction scheme of multimedia communication system based on QoS (quality of service)-aware agent organization. In this paper, we propose a behavioral characteristics model in multiagent systems to support their design and operation. We also introduce an observation and control scheme in flexible distributed system (FDS) and configure the behavioral characteristics model to analyze the behavior of multiagent systems based on QoS of FDS. In addition, this model is applied to multimedia communication system configured by QoS-aware agent organization, and discussed the effectiveness of the proposed scheme.

To cope with the complexity and broad scope of areas and functions that are involved with a network management system (NMS), a knowledge model should be provided for the system. The aim of this paper is to propose the design of a multi-agent based network management support system with well-organized representation of the formal semantics of networks' experiential knowledge related to NMS fault diagnosis functional area. The modeling of the network knowledge is performed using domain ontology, fault-state causal reasoning models, and problem solving methods (PSMs) in a generic and reusable manner. Hence, knowledge acquisition in the NMS domain can be done by using these reusable structures, thereby eliminating the need to build the new knowledge-based systems from scratch. This paper presents an approach towards the agent-mediated ontology-driven automated NMS to benefit the network administrators through the automatic provision of just-in-time and context-dependent knowledge for resolving network failures.

The explosive growth in online information is making it harder for large, globally distributed organizations to foster collaboration and leverage their intellectual assets. Recently, there has been a growing interest in the development of next generation knowledge management systems focussing on the artificial intelligence based technologies. We propose a generic knowledge management system architecture based on ADIPS (agent-based distributed information processing system) framework. This contributes to the stream of research on intelligent KM system to supports the creation, acquisition, management, and sharing of information that is widely distributed over a network system. It will benefit the users through the automatic provision of timely and relevant information with minimal effort to search for that information. Ontologies which stand out as a keystone of new generation of multiagent information systems, are used for the purpose of structuring the resources. This framework provides personalized information delivery, identifies items of interest to user proactively and enables unwavering management of distributed intellectual assets.

Generally, in order to manage and maintain a network system, it is necessary to carry out a series of operations including assessing the network status, determining the network errors, selecting/approving the countermeasures and applying the countermeasures. Since network systems are becoming more complicated and larger in scale, network operations increasingly require special professional knowledge and effort, thereby imposing heavy workloads on network administrators. In this paper, we propose and design the AIR-NMS (Active Information Resource Architecture based Network Management Support System) for decreasing network administrator workloads.

To reduce the burden for users that utilize information from distributed information resources, the deployment of distributed active information resource (AIR) is proposed. An AIR is a information resource that is enhanced with the knowledge of the content and the functionality for actively and flexibly supporting use/reuse of the content. Multiple AIRs could cooperate actively and automatically to target the proper information resources, carry out complex information processing and then provide composite result. In this paper the concept and main use of AIR are discussed. Also, the construction and information processing of AIR, as well as an experimental implementation of agent-based AIR system, are introduced.

To improve the region extraction accuracy in active contour model (ACM), we propose a novel method for employing the a priori knowledge of a target region shape. The proposed method is based on an idea that a user-drawn initial contour for ACM can include the shape features of a target region. In this method, the sequence of shape features (e.g., corner, line, and curve) is obtained from an initial contour, and this sequence is used as the a priori knowledge of a target region. During region extraction procedure, by finding the optimal assignment of the shape features to contour control points and determining the optimal positions for these points, ACM energy is minimized and the knowledge is reflected on an extracted region shape. Using an initial contour not only as an initial value for the energy minimization but also as the a priori knowledge of a target region shape, the proposed method can employ the a priori knowledge effectively without time-consuming preparation work.

Corners in images include useful information and they play an important role in computer vision. However, stable and accurate comer detection is very difficult, because the edges composing a comer break easily around the corner. To overcome this difficulty, a corner detection method using slit rotational edge-feature detector (SRED) has been proposed. By evaluating the edge certainties at each pixel in all directions, this method can detect corners stably. Nevertheless, this method cannot detect the corners in the region of low contrast, and its detection accuracy depends on the edge directions. To solve these problems, we propose a new comer detection method using weighted and interpolated SIZED (WI-SRED). By using the edge certainties based on the region separability, the proposed method achieves stable corner detection in the region of low contrast, and moreover, by interpolating and weighting pixels, this method prevents the detection accuracy from depending on the edge directions.

For accurate region extraction, region-based active contour models (ACM) have been proposed. Compared with ordinary ACM, they require heavier loads of initial settings and further processing time. Considering these problems, for effective and efficient extraction, we propose a new region-based ACM. In the proposed method, first, users draw an initial curve in an object. Along this curve, definite length scanlines are set perpendicular to the initial curve and across the object contour. Depending on the image properties around scanlines, the scanlines are separated into several groups and each group is divided into two subregions (object and background regions). Scanline grouping and group dividing are iterated until the image properties are sufficiently uniform in each subregion. Through this clustering, a region along an object contour is segmented into subregions of uniform image properties, and consequently the image properties in a necessary and sufficient area are introduced into the extraction process effectively.

A new method for extracting an object region in a complex scene image is proposed. For improving the accuracy and the robustness of region extraction, to the region of a single object, the proposed method applies multiple active contour models (ACMs) controlled by the statistical characteristics of image data. Around the object boundary, these ACMs compete with each other and each ACM extracts a subregion of uniform image properties. As a result of this competition, the entire region of an object is extracted as a set of several subregions. For the proposed method it is necessary to set many initial contours. To lighten this load, a procedure for making multiple initial contours from a single initial curve is also proposed. In this procedure, a few initial curves are set in an image, each initial curve is divided into segments at the optimal loci, and initial contours are made by dilating these segments.

For interactive image manipulation, a new region extraction method using active contour model (ACM) is proposed. In this method, to the region of a single object, multiple ACMs controlled by the statistical characteristics of image data are applied. These ACMs compete with each other, and each ACM extracts a subregion of uniform image properties. The entire region of the object is extracted as a set of several subregions. This method requires multiple initial ACMs set. To ease this initial setting, a procedure for making initial ACMs from a few initial curves is also proposed. In this procedure, a few initial curves are set in an image, each initial curve is divided into appropriate number segments at the optimal loci, and multiple initial ACMs are made by dilating these segments.

A novel method for hierarchical-clustering of parametric data is proposed (in this article, we assumed that these data were parameterized by a single parameter in multidimensional spaces). In the proposed clustering method, the continuity of a parameter is preserved in each class, and furthermore, the optimal loci of class boundaries and the appropriate number of classes are determined. To improve the recognition performance of the parametric eigenspace method, which is an object recognition method based on the visual learning approach, the proposed clustering method is applied to the construction of tree-structured dictionaries for this recognition method. Experimental result shows that these tree-structured dictionaries improve the recognition performance of the parametric eigenspace method without a decrease in recognition accuracy.

Most of conventional active contour models are not capable of extracting objects with complex image features or background, because they are deformed mainly based on the local image features along the contours. To deal with this problem, we propose a novel extracting method which reflects wide-ranging region information to region extracting process through competition of active contours. In the proposed method, firstly we set the initial curves in object and background, then divide these curves into segments as cores of initial contours. Secondly, we estimate feature distribution of inside of each contour, and determine the likelihood of control points to each contour with respect to image features. Each contour performs region competition based on the likelihood, and finally an object is extracted as a set of multiple active contours.