Nowadays, with the continuous integration of production network and business network, more and more Industrial Internet of Things and Internal Office Network have been interconnected and evolved into a large-scale enterprise-level intraindustry network. Terminal devices are the basic units of internal network. Accurate identification of the type of device corresponding to the IP address and detailed description of the communication behavior of the device are of great significance for conducting network security risk assessment, hidden danger investigation, and threat warning. Traditional cyberspace surveying and mapping techniques take the form of active measurement, but they cannot be transplanted to large-scale intranet. Resources or specific targets in internal networks are often protected by firewalls, VPNs, gateways, and other technologies, so they are difficult to analyze and determine by active measurement. In this paper, a passive measurement method is proposed to identify and characterize devices in the network through real traffic data. Firstly, a new graph structure mining method is used to determine the server-like devices and host-like devices; then, the NAT-like devices are determined by quantitative analysis of traffic; finally, by qualitative analysis of the NAT-like device traffic, it is determined whether there are server-like devices behind the NAT-like device. This method will prove to be useful in identifying all kinds of devices in network data traffic, detecting unauthorized NAT-like devices and whether there are server-like devices behind the NAT-like devices.

MOTIVATION: Previous researches showed that phytoene synthase (Psy) from Dunaliella bardawil is the first regulatory point in carotenogenesis. We hypothesize certain interactions between the environmental stress factors and the regulatory sequences of Psy in D.bardawil (DbPsy). Consequently, LA PCR-based genomic walking approach was performed for isolation of psy promoter and terminator, respectively. The obtained nucleic acid sequences and the corresponding protein structure of DbPsy were analyzed and predicted using various bioinformatics tools. Finally, we presented some hints for the regulation mechanisms of DbPsy at the molecular level according to the computed results. RESULTS: LA PCR-based genomic walking results showed that the isolated sequences are the promoter and terminator of psy, correspondingly. Computational analysis demonstrated several candidate motifs of the promoter exhibiting hypothetic UV-B-, norglurzon- and salt-induced characteristics, as well as some typical domains universally discovered in promoter sequences, such as TATA-box, CCAAT-box and GATA-box, etc. Furthermore, the structure of Psy was also predicted and aligned along with many counterparts at the protein level. Low homology of N-terminus was found in D.bardawil, while a relatively conserved C-terminus was predicted to be involved in the catalytic activity and substrate recognization/binding. Phylogenic analysis classified the DbPsy into a cluster with other algae. These results implied that Psy may share similar regulation mechanisms among algae with respect to their C-termini; while the diversity in N-terminus among Psys, along with the predicted inducible motifs in psy promoter from D.bardawil, may confer the fine tuning differences between D.bardawil and other algae. CONCLUSION: By means of computer techniques, we found in D.barawali that two interesting conserved motifs of psy promoter may involve in UV-B, norglurzon and salt regulation correspondingly; and that the diversity of Psy protein mainly lies in the N-termini among algae. These results indicate some hints for regulation mechanisms of carotenogenesis in D.bradawil. CONTACT: [email protected].

In this paper we propose a digital watermarking algorithm based on balanced multi-wavelet. According to the characteristics of the balanced multi-wavelet, a visual masking model based on single wavelet transformation and the JND (just noticeable difference) formula are analyzed and modified, then the improved visual model and the modified JND formula for balanced multi-wavelet transformation are obtained. The watermark extraction can be performed without the original image, so the algorithm belongs to blind watermarking. Experimental results show that the algorithm has strong imperceptibleness as well as good robustness.