Z.-J. Qian's scientific contributions

Publications (5)

Article
To solve the problem of network traffic identification online, a clustering algorithm and a traffic identification scheme is proposed. The scheme uses a few number of the initial data packets in the flows as a sub-flow, extracts the statistical features from sub-flows, and extracts the best feature subset of sub-flows by applying correlation-based...
Article
The structure of optical code division multiple access (OCDMA) system based on Chebyshev-map chaotic codes is investigated in passive optical networks. The system can support more subscribers by providing the different codes at same wavelength channel. The scheme on chaotic spread spectrum sequences using Chebyshev-map is presented, which is taken...
Article
Machine learning with C4.5 algorithm is proposed for network traffic identification. The correlation feature selection algorithm and the genetic algorithm are adopted to select the attribute feature subset. A method of combining N-fold cross validation with testing set is suggested to assess the classification results of the current national broadb...
Article
An optical access network architecture is researched, with high capacity and large branches, named optical code division multiple access-wavelength division multiplexing-passive optical networks (OCDMA-WDM-PON). The scheme combines the advantages of both WDM and incoherent OCDMA, avoids the limitation of available wavelength channels and optical co...
Article
A dynamic scheduling algorithm in Ethernet passive optical networks (EPON) is proposed, and named as priority-based two-level cycle time-dynamic bandwidth allocation (PTC-DBA), it supports different priority services by allocating bandwidth with two-level cycle time. The scheme can enforce the bandwidth and delay guarantees of high priority service...

Citations

... The analysis showed that the advantages of this kind of method is: can make full use of the existing sample experience, control sample training, improve the accuracy of subjective factors; disadvantage is strong, training of excessive consumption of storage resources and time, and cannot identify the unknown exception types. The unsupervised learning method, K-Means literature [7,8] and DBSCAN methods were used. The advantages of this kind of method is that they do not need to have too much knowledge of the sample classification, reduce human error; disadvantage is that the classification results need to sample lots of analysis is reliable, the classification accuracy is low. ...