In recent years, as an effective method to mine information from the complex network, community discovery has been widely used in social network, financial risk control and other fields. However, the existing community discovery algorithms are not effective in dealing with complex network which always contains fuzzy community structure. With the help of graph convolution, the proposed algorithm defines the connectivity between any nodes in a network and constructs the symmetric doubly stochastic matrix. Then, the algorithm enhances the network by the nonlinear transformation of the eigenvalues of the symmetric doubly stochastic matrix and makes the original fuzzy community structure become clear. Experimental results show that this method can effectively sharpen the community structure of a network and improve the effect of community partitioning.

Discovering mixed-drove spatiotemporal co-occurrence patterns (MDCOPs) is important for network security such as distributed denial of service (DDoS) attack. There are usually many features when we are suffering from a DDoS attacks such as the server CPU is heavily occupied for a long time, bandwidth is hoovered and so on. In distributed cooperative intrusion, the feature information from multiple intrusion detection sources should be analyzed simultaneously to find the spatial correlation among the feature information. In addition to spatial correlation, intrusion also has temporal correlation. Some invasions are gradually penetrating, and attacks are the result of cumulative effects over a period of time. So it is necessary to discover mixed-drove spatiotemporal co-occurrence patterns (MDCOPs) in network security. However, it is difficult to mine MDCOPs from large attack event data sets because mining MDCOPs is computationally very expensive. In information security, the set of candidate co-occurrence attack event data sets is exponential in the number of object-types and the spatiotemporal data sets are too large to be managed in memory. To reduce the number of candidate co-occurrence instances, we present a computationally efficient MDCOP Graph Miner algorithm by using Time Aggregated Graph. which can deal with large attack event data sets by means of file index. The correctness, completeness and efficiency of the proposed methods are analyzed.

In recent years, invaders have been increasing rapidly in the internet world. Gen- erally, to detect anonymous attackers, the algorithm needs more features. Many algorithms fail in the efficiency of detecting malicious activity. The deep learning approach has been used in cyber security use cases, namely, intrusion detection, malware analysis, traffic analysis, spam and phishing detection etc. In this work, to leverage the application of deep learning architectures towards cyber secu- rity, we consider malicious activity detection using Bi-LSTM. In the experiments of intrusion detection using the dataset UGR’16, the deep learning approach performed better when compared to the combination of Bi-LSTM with an autoen- coder neural network model. Moreover, the approach without autoencoder, both precision and recall are 99 Percentage for just the Bi-LSTM model in detecting malicious activities in cyber security. Moreover by using Autoencoder as feature enginerring does not yeild any higher performance when modelling deep learn- ing algorithm using Bi-directional LSTM. However, when using with Bi-LSTM without Autoencoder, the performace are more efficient and better.

Barrier coverage is a fundamental application in wireless sensor networks, which are widely used for smart cities. In applications, the sensors form a barrier for the intruders and protect an area through intrusion detection. In this paper, we study a new branch of barrier coverage, namely warning barrier coverage (WBC). Different from the classic barrier coverage, WBC has the inverse protect direction, which moves the sensors surrounding a dangerous region and protects any unexpected visitors by warning them away from the dangers. WBC holds a promising prospect in many danger keep out applications for smart cities. For example, a WBC can enclose the debris area in the sea and alarm any approaching ships in order to avoid their damaging propellers. One special feature of WBC is that the target region is usually dangerous and its boundary is previously unknown. Hence, the scattered mobile nodes need to detect the boundary and form the barrier coverage themselves. It is challenging to form these distributed sensor nodes into a barrier because a node can sense only the local information and there is no global information of the unknown region or other nodes. To this end, in response to the newly proposed issue of the formation of barrier cover, we propose a novel solution AutoBar for mobile sensor nodes to automatically form a WBC for smart cities. Notably, this is the first work to trigger the coverage problem of the alarm barrier, wherein the regional information is not pre-known. To pursue the high coverage quality, we theoretically derive the optimal distribution pattern of sensor nodes using convex theory. Based on the analysis, we design a fully distributed algorithm that enables nodes to collaboratively move toward the optimal distribution pattern. In addition, AutoBar is able to reorganize the barrier even if any node is broken. To validate the feasibility of AutoBar, we develop the prototype of the specialized mobile node, which consists of two kinds of sensors: one for boundary detection and another for visitor detection. Based on the prototype, we conduct extensive real trace-driven simulations in various smart city scenarios. Performance results demonstrate that AutoBar outperforms the existing barrier coverage strategies in terms of coverage quality, formation duration, and communication overhead.