Sudipta Chowdhury

Botnet detection using graph-based feature clustering

Detecting botnets in a network is crucial because bots impact numerous areas such as cyber security, finance, health care, law enforcement, and more. Botnets are becoming more sophisticated and dangerous day-by-day, and most of the existing rule based and flow based detection methods may not be capable of detecting bot activities in an efficient and effective manner. Hence, designing a robust and fast botnet detection method is of high significance. In this study, we propose a novel botnet detection methodology based on topological features of nodes within a graph: in degree, out degree, in degree weight, out degree weight, clustering coefficient, node betweenness, and eigenvector centrality. A self-organizing map clustering method is applied to establish clusters of nodes in the network based on these features. Our method is capable of isolating bots in clusters of small sizes while containing the majority of normal nodes in the same big cluster. Thus, bots can be detected by searching a limited number of nodes. A filtering procedure is also developed to further enhance the algorithm efficiency by removing inactive nodes from consideration. The methodology is verified using the CTU-13 datasets, and benchmarked against a classification-based detection method. The results show that our proposed method can efficiently detect the bots despite their varying behaviors.

In-situ monitoring of melt pool images for porosity prediction in directed energy deposition processes

ABSTRACT One major challenge of implementing Directed Energy Deposition (DED) Additive Manufacturing (AM) for production is the lack of understanding of its underlying process–structure–property relationship. Parts manufactured using the DED technologies may be too inconsistent and unreliable to meet the stringent requirements for many industrial applications. The objective of this research is to characterize the underlying thermo-physical dynamics of the DED process, captured by melt pool signals, and predict porosity during the build. Herein we propose a novel porosity prediction method based on the temperature distribution of the top surface of the melt pool as an AM part is being built. Self-Organizing Maps (SOMs) are then used to further analyze the two-dimensional melt pool image streams to identify similar and dissimilar melt pools. X-ray tomography is used to experimentally locate porosity within the Ti-6Al-4V thin-wall specimen, which is then compared with predicted porosity locations based on the melt pool analysis. Results show that the proposed method based on the temperature distribution of the melt pool is able to predict the location of porosity almost 96% of the time when the appropriate SOM model using a thermal profile is selected. Results are also compared with a previous study, that focuses only on the shape and size of the melt pool. We find that the incorporation of thermal distribution significantly improves the accuracy of porosity prediction. The significance of the proposed methodology based on the melt pool profiles is that this can lead the way toward in situ monitoring and minimize or even eliminate pores within the AM parts.

Sustainable Design of On-Demand Supply Chain Network for Additive Manufacturing

Abstract This study proposes a novel optimization framework that simultaneously considers interdependence of flow networks, resource restrictions, and process-and-system level costs under a unified decision framework for the design and management of an integrated Additive Manufacturing (AM) supply chain network. A two-stage stochastic programming model is proposed that minimizes the facility location and capacity selection decisions at the first-stage prior to realizing any customer demand information. However, after the demand information is revealed, a number of second-stage decisions, such as optimal layer thickness for AM products, production, post-processing, procurement, storage, and transportation decisions, are made. Due to the need to solve our proposed optimization framework in a realistic-size network problem, a hybrid decomposition algorithm, combining the Sample Average Approximation algorithm with an Adaptive Large Neighborhood Search algorithm, is proposed. The performance of the proposed algorithm is validated by developing a case study using data from Alabama and Mississippi. Based on a set of numerical experiments, the effect of process-and-system level factors on the design and management of an AM supply chain network are analyzed. Numerous managerial insights, particularly on layer thickness, customer demand variability, mean demand variation, powder safety stock, and wastage rate on overall system performance, are gained which are crucial for the sustainment of this new manufacturing and supply chain paradigm.