Mohammad Masoud

On preventing ARP poisoning attack utilizing Software Defined Network (SDN) paradigm

In this work, SDN has been utilized to alleviate and eliminate the problem of ARP poisoning attack. This attack is the underlying infrastructure for many other network attacks, such as, man in the middle, denial of service and session hijacking. In this paper we propose a new algorithm to resolve the problem of ARP spoofing. The algorithm can be applied in two different scenarios. The two scenarios are based on whether a network host will be assigned a dynamic or a static IP address. We call the first scenario SDN_DYN; the second scenario is called SDN_STA. For the evaluation process, a physical SDN-enabled switch has been utilized with Ryu controller. Our results show that the new algorithm can prevent ARP spoofing and other attacks exploiting it.

A graph-theoretic study of the flattening Internet AS topology

The Internet topology at the autonomous system (AS) level has been under heavy investigation in recent years. Previous studies have shown that the Internet AS topology exhibits a power-law node degree distribution and a small-world structure. With the proliferation of peering links and IXPs between ASes, the traditional hierarchical Internet has been more flattened. In this paper, we conducted a graph-theoretic study of the Internet AS topology and applied the centrality metrics (including betweenness and closeness) to quantify the AS core. To this end, we implemented a progressive node deletion algorithm to eliminate high-rank AS nodes from the AS topology in order to study whether the AS topology maintains as a tiering graph. To evaluate the proposed algorithm, we constructed the Internet AS topology based BGP dumps harvested from 50 route servers. Our results show a strong trend quantitatively that the Internet AS topology has become much more flattened. The Internet is more resilient to node and link failures and is difficult to be torn down. This AS topology evolution trend may result in significant traffic shifting and potentially reshaping the global telecom industry.

Constructing a locality-aware ISP-friendly peer-to-peer live streaming architecture

Recently peer-to-peer (P2P) live streaming has become a promising approach for delivering videos on massive scale on the Internet. The popularity of various P2P streaming systems have now driven Internet Service Providers (ISPs) to carry huge intra/inter-domain video traffic. To reduce costs, ISPs have attempted to deploy traffic shaping measures while P2P streaming systems have attempted to apply escape strategies to avoid the rate-throttling punishment. Nevertheless, the emerging cooperation between ISPs and P2P systems may bring forth benefits for both parties by constructing a locality-aware ISP-friendly P2P live streaming architecture. Lacking of incentive, however, makes collaboration difficult if not impossible. Therefore, we propose a new collaboration-enhanced live streaming architecture (CELS) to encourage cooperation between ISPs and P2P streaming systems. Within this architecture, ISPs contribute amplifiers to reinforce the bandwidth provisioning for streaming channels. Reciprocally, peers cooperatively provide their locality information to ISPs. ISPs then use this information to assist peers to select appropriate neighbors which can reduce intra/inter-domain traffic load. This architecture utilizes the locality information in a three-tier hierarchy: AS, PoP and IP blocks, to reduce inter/intra-domain traffic. We conducted a simulation study to evaluate the performance of the proposed CELS architecture. We implemented a push-pull live streaming protocol, CoolStreaming+ and the CELS architecture in ns2. Our simulation experiments demonstrate that CELS reduces inter-domain traffic but maintains satisfied streaming performance of the CoolStreaming+ system. Our study may provide some insights into the design trade-off for exploring the full potentials of the ISP-P2P cooperations.

A measurement study of AS paths: Methods and tools

Many Internet applications are designed and deployed as overlay applications. The potential mismatch between the application overlay and the network underlay has driven the demand for designing locality-aware applications in order to reduce emerging huge inter-domain traffic load. In this paper, we study a fundamental problem of measuring AS paths between two Internet hosts using three methods including traceroute-based direct measurement, BGP-based indirect inference and graph-based shortest AS path. We conducted a measurement study of AS paths to evaluated the accuracy and complexity of the above three AS path measurement methods and the corresponding tools. Inspired from our experiment results, we proposed a hybrid progressive method to combine the traceroute probes and the BGP tables to enhance the IP-to-AS mapping process to achieve a more accurate estimation of AS paths. We also found that the missing IP addresses in the traceroute measurement decrease the accuracy of the traceroute-based method; however, this performance degradation can be compensated using BGP tables. Our study leads a more accurate IP-to-AS mapping tool and it can provide a solid support for locality-aware Internet applications.

Capacity of Wireless Networks With Directed Energy Links in the Presence of Obstacles

In this paper, we study the capacity of wireless networks with directed energy (DE) links in the presence of obstacles. DE links are highly focused wireless links that can be treated as “pencil beam,” an example of which is the E-band link (71–76 and 81–86 GHz) newly made available by the FCC. The 10-GHz spectrum of E-band is 50-times that of the entire cellular spectrum and provides the much needed spectrum in today’s world of exponential growth in mobile applications. Since the performance of highly focused DE links are highly susceptible to the presence of obstacles and real-world applications typically involve obstacles, it is important to study the capacity of wireless networks with DE links in the presence of obstacles, which is the subject of this paper. In the following, we first provide a study of probability distribution of DE links in the presence of obstacles and investigate how the number, shape, and size of obstacles impact the DE link probability distribution. Then, based on the probability distribution of DE links, we derive the capacity scaling laws for wireless networks in the presence of obstacles. Furthermore, our results can be extended to obstacles with arbitrary shapes quite accurately.

Capacity of Large-Scale Wireless Networks Under Jamming: Modeling and Analyses

Distributed jamming has important applications not only in the military context but also in the civilian context, where spectrum sharing is increasingly used and inadvertent jamming becomes a reality. In this paper, we derive the capacity bounds of wireless networks in the presence of jamming. We show that when the density of jammers is higher than that of target nodes by a certain threshold, the capacity of wireless networks approaches zero as the numbers of target nodes and jammers go to infinity. This is true even when the total power of target nodes is much higher than that of the jammers. We provide the optimal communication schemes to achieve the capacity bounds. We also describe the power efficiency of wireless networks, showing that there is an optimal target node density for power-efficient network operation. Our results can provide guidance for designing optimal wireless networking protocols that have to deal with large-scale distributed jamming.

The Impact of 16-bit and 32-bit ASNs Coexistence on the Accuracy of Internet AS Graph

Modeling Internet structure as an autonomous system (AS) graph has attracted researchers over years. AS graph model demonstrates the power-law distribution of the Internet. It also demonstrates the relationship between the cluster coefficient and the small-world structure of the Internet. To obtain an accurate AS graph model, the data used to generate the graph should be massive and correct. In this work, we studied the correctness of the data that is utilized to generate Internet AS graph. We conducted an experiment to measure the popularity of 32-bit AS numbers (ASNs) in the Internet. We examined the impact of the special purpose 16-bit reserved ASN AS23456 on the accuracy of the AS graph. To this end, we proposed a cleaning algorithm to correct the conflict that AS23456 produces. Our results show that the current existing method of mapping 16-bit and 32-bit ASNs reduces the fidelity of the constructed AS graph to various graph parameters.

Software Project Management: Resources Prediction and Estimation Utilizing Unsupervised Machine Learning Algorithm

Software project effort estimation is a major process in software development cycle. This process helps in decision making in resource allocation and distribution. In this work, a new effort estimation clustering method based on estimation maximization soft-clustering unsupervised machine learning algorithm is proposed. This model classifies any software project into one of four categories. An enterprise will accept to develop a software project if this project is clustered into a class that requires resources equal or less than the enterprises resources. The new model helps in decision making process in one hand and helps consumers in assigning projects to a developing enterprise in the other hand. COCOMO dataset has been used to implement, deploy and test the model. The propose model has been compared with K-means algorithm to show the differences between soft and hard clustering. The paper results show that soft-clustering has the ability to estimate efforts like any supervised machine learning algorithms.

Contention delay distribution in event driven wireless sensor networks

In this paper we derive the distribution of the total contention delay in wireless sensor networks (WSNs) tasked with detection of an intruder in a specific sensing field. We use a stochastic geometry framework to model the WSN, in which the sensor nodes send their local decisions and data to the fusion center over a shared communication channel employing a contention based MAC protocol. We first derive the mean and variance of the contention delay. Then we approximate the distribution with a negative binomial via moment matching. This statistical characterization is then used to find the optimum medium access probability to minimize both the mean and the variance of the total delay. Simulation results almost exactly follow our theoretical results.

LEACH-T: LEACH Clustering Protocol Based on Three Layers

Power consumption of routing protocols is one of the main issues that wireless sensor networks (WSNs) encounter in their lifetime. Low-Energy Adaptive Clustering Hierarchy (LEACH) clustering protocol was introduced to reduce power consumption. However, in LEACH, power consumption increases massively as the distance between sink node and cluster heads (CHs) increases. This drawback introduces distance as one major issue in LEACH since it does not contain routing. In this work, a LEACH based protocol consists of three layers (LEACH-T) is proposed. Each layer has its own CHs. The layers attempt to reduce the distance between sink node and CHs. The third layer is utilized if the distances between CHs and sink node exceed a threshold value. To measure the performance of LEACH-T, simulation is utilized to compare its consumption to pure LEACH. Results show that the proposed protocol extends network lifetime and reduces power consumption.