محمد فالح حسن الجنابي

Anomalous events such as link failure, misconfiguration, and Denial of Service attacks can affect the Internet inter-domain routing protocol. This effect can range from small to large-scale impact. While large-scale events can be detected using one or multiple global monitoring points, small-scale events need monitoring at the Autonomous System (AS) level. This paper presents a Real-time Detection Tool for Internet routing protocol Disruptions (RDTD) at AS-level. RDTD is a black-box statistical approach that detects disruptions based on observing changes in the underlying behaviour of a series of inter-domain routing updates rather than information contained in inter-domain routing updates. The RDTD can be connected to a designated AS to detect disruptions at that AS or to one of the collectors at public vantage points to detect the Internet routing disruptions from the public vantage-point’s view. The evaluation of the detection tool has been made through replaying route traffic related to one of the most well-known events within a controlled testbed. Our evaluation shows the ability of the detection tool to detect route leak in near real-time without requiring a long history of data. RDTD can also detect hidden anomalous behaviour in the underlying traffic that may pass without detection.

Open Shortest Path First (OSPF) is one of the most widely used intra-domain routing protocols. Unfortunately, it has many serious security issues. Falsification over OSPF is one of the most critical vulnerabilities that can cause routing loops and a black hole. In this paper, we introduce a novel approach by using a technique from non-linear statistical analysis to identify OSPF attacks. Firstly, we evaluate the capability of the non-linear technique to identify OSPF attacks using a controlled testbed where we introduce different types of LSA falsifications. Secondly, we evaluate our approach to detect different types of OSPF attacks using OSPF traffic associated with a single OSPF router and OSPF traffic associated with a set of OSPF routers. In both cases, our approach can detect anomalous behaviour quickly. Finally, we use various successful machine learning classifiers to analyze the outputs obtained from the non …

Due to growth in demand for high-performance applications that require high numerical stability and accuracy, the need for floating-point FPGA has been increased. In this work, an open-source and efficient floating-point unit is implemented on a standard Xilinx Sparton-6 FPGA platform. The proposed design is described in a hierarchal way starting from functional block descriptions toward modules level design. Our implementation used minimal resources available on the targeting FPGA board, tested on the Sparton-6 FPGA platform and verified on ModelSim. The open-source framework can be embedded or customized for low-cost FPGA devices that do not offer floating-point units.

With the ensuing massive communication demands required for IoT Wireless Sensor Networks (WSNs), conventional routing protocols in wireless networks are not applicable for the massive IoT sensors networking. Consequently, a new type of routing protocol called energy-aware routing protocols has been proposed. Major design goals of such protocols are to uniformly distribute energy consumption among the IoT nodes and minimize the energy dissipation optimally. In this paper, we develop a novel Energy Balanced Distributed Clustering protocol (EBDC) to minimize the energy consumption among sensor nodes uniformly. Moreover, the proposed EBDC is based on an adaptive clustering and re-clustering process. We evaluate the proposed protocol on different energy-based WSNs protocols. Our evaluation shows EBDC achieves a notable enhancement in terms of balanced energy consumption and extended network lifetime compared to the other existing protocols. In addition, we propose a new metric for the evaluation of protocols.

The history of the works in Multiple Classifier Systems (MCS) goes back to the 1979 paper by Dasarathy and Sheela [1]. In their work, the training space is partitioned into several combined classifiers. In 1990, the work given by Hansen and Salamon [2] improved the generalization error by combining several neural networks classifiers. The main contribution into ensemble systems is done by Schapire [3]. In his work, a strong classifier is generated from combining several weak classifiers using boosting. From there, research in ensemble systems grew rapidly under different names. The short summary of names and algorithms that have described ensemble systems are a combination of multiple classifiers [4]–[8], classifier fusion [9]–[11], classifiers ensembles (ensemble systems)[12],[13], mixture of experts [14],[15] and many others. In addition, several books have already been written that focus on topics that deal with the development of ensemble systems such as [16],[17] and from 2000 until now there are series of annual workshops on multiple classifier systems [18]. The purpose of these workshops is to organize and improve progress in the area of combined classifier systems.

In order to improve classification accuracy, multiple classifier systems have provided better pattern classification over single classifier systems in different applications. The theoretical frameworks proposed in [5], [7] present important tools for estimating and minimizing the added error of linearly combined classifier systems. In this work, we present a theoretical model that estimates the added error using geometric mean rule which is a nonlinear combining rule. In the derivation, we assume classifiers' outputs are uncorrelated and have identical distributions for a given class case. We also show that by setting the number of classifiers to one (a single classifier system), the derived formula is modified and matches the results given in [5]. We validated our derivations with computer simulations and compared these with the analytical results. Due to the nonlinearity of the geometric mean, theoretical results show that the bias and variance errors are mixed together in their contribution to the added error. It was shown that the bias error dominated the contribution to the added error compared to the variance error. It is possible to minimize the variance error by increasing the ensemble size (number of classifiers) while the bias error is minimized under certain conditions. The proposed theoretical work can help in investigating the added error for other nonlinear arithmetic combining rules.