Information-Centric Networking (ICN) separates the identifier and address of the host, which enables the multi-homed host to own multiple addresses. The packet carrying multiple destination addresses could obtain multiple output ports after matching the routing table. Therefore, in order to select the best output port for the packet with multiple destination addresses, the router needs to maintain a status table to record the dynamic status values of different ports. Since the status table directs the forwarding of the flow, the accuracy of the status value is crucial to the performance of the transmission. In the paper, we propose an improved maximum flow routing algorithm to calculate the status value of the status table to improve the efficiency of the transmission. The proposed algorithm utilizes the dynamic network information to maximize the transmission rate of the flow. Besides, status table is also dynamically updated to adapt to the change of the flows in the network. Experimental results show this routing algorithm could increase the throughput by up to 40% and decrease the flow completion time by about 30%. The maximum flow routing algorithm achieves good performance even in heterogeneous path environment and high network load. The dynamic update of the status table could also improve the routing robustness of the network.

Video services are flourishing, and the same content might be hosted by different Web sites. A user typically does not care about which provider provides the service to him/her but rather the quality of the service. However, the network condition is not stable and it is hard to obtain the service performance in real time during the online service. Therefore, it is very important to understand the potential characteristics of video services and design the predictors to forecast the service performance. In our work, we design a measurement system deployed in 11 provinces and cities in China, monitor and analyze two popular Web sites, Youku and Tudou. Based on the analysis of the measured data, we see that the performance trend of two service providers has the diurnal pattern, and the worst performance is typically during the prime time, 8 - 9 pm. To understand the service structure of the website, the underlying CDN architecture is studied and the performance impact factors are analyzed. For performance forecasting, a modified time series model is proposed and evaluated, which shows that it can obtain obvious better performance than baseline models. Moreover, a new predictor by combining different information sources is designed, which can improve the forecasting precision significantly, and it may be useful in service source analyzing and recommendation systems in the future.

Research on caching strategy is the key to improving network performance for Information-Centric Networks (ICNs). But it is still a great challenge to better utilize in-network caching of ICNs with low costs. In this paper, we propose a popularity-based neighborhood collaborative caching algorithm for ICNs. In the algorithm, in-network nodes track the popularity of contents, and a novel process is used for quick comparison of popularity in the algorithm. En-route and one-hop neighborhood nodes make caching decision collaboratively. Real-world topologies and different client placed scenes are used in the simulation experiments, and our algorithm performs better in terms of latency, cache hit ratio and path stretch compared with the state-of-the-art algorithms and ideal situations. The overhead and tradeoff of the algorithm on estimation of popularity and node interaction are also explored in details, and the proposed algorithm provides a practical choice for ICN caching decision strategy with its good performance and acceptable overhead.

Traffic classification plays an important and basic role in network management and cyberspace security. With the widespread use of encryption techniques in network applications, encrypted traffic has recently become a great challenge for the traditional traffic classification methods. In this paper we proposed an end-to-end encrypted traffic classification method with one-dimensional convolution neural networks. This method integrates feature extraction, feature selection and classifier into a unified end-to-end framework, intending to automatically learning nonlinear relationship between raw input and expected output. To the best of our knowledge, it is the first time to apply an end-to-end method to the encrypted traffic classification domain. The method is validated with the public ISCX VPN-nonVPN traffic dataset. Among all of the four experiments, with the best traffic representation and the fine-tuned model, 11 of 12 evaluation metrics of the experiment results outperform the state-of-the-art method, which indicates the effectiveness of the proposed method.

Research has proven that in-network caching is an effective way of eliminating redundant network traffic. For a larger cache that scales up to terabytes, a network element must utilize block storage devices. Nevertheless, accessing block devices in packet forwarding paths could be a major performance bottleneck because storage devices are prone to be much slower than memory devices concerning bandwidth and latency. Software-defined networking (SDN) has entered into all aspects of network architecture by separating the control and forwarding plane to make it more programmable and application-aware. Protocol-oblivious forwarding (POF), which is an enhancement to current OpenFlow-based SDN forwarding architecture, enhances the network programmability further. In this paper, we proposed a novel split architecture to cope with the problem of speed mismatch between high-speed packet forwarding and low-speed block I/O operation over POF switches. The issues raised by this split architecture were first explored and could be summarized as packet dependency and protocol conversion. Then, we focused on solving these two problems and proposed an efficient and scalable design. Finally, we conducted extensive experiments to evaluate the split architecture along with the proposed approaches for packet dependency and protocol conversion.

To predict the availability state of a node in a distribution network, its history trace is usually used. Sometimes, some usage behavior patterns cannot be captured precisely from the insufficient trace, which may lead to unreliable predictors. In this paper, to alleviate the data sparseness problem, the nodes with the similar behaviors are clustered, and all history information in a same cluster is seen as another information source for any node in it. For each node, an N-gram model is used to train the predictor by the combination of the new source and the node's own trace. In addition, because it is hard to capture the trace of all nodes in large scale networks, such as P2P networks, a bagging based prediction algorithm is proposed, which can be applied in the distribution environment and relieve the effect of the noisy data. In our experiments, three datasets are evaluated. Results show that the prediction performance of our cluster based N-gram predictor is better than the results of several other predictors. And the bagging based prediction algorithm presents its validity in the distribution environment.

We propose a novel way to alleviate the data sparseness problem in training letter-to-sound (LTS) N-gram models by adding automatically generated new words to the training set. The proposed method consists of two procedures: (1) generating a large pool of new words automatically; (2) selecting good new word candidates from the new word pool via semi-supervised learning. The new words are created by replacing stressed syllables of an existing word with other stressed syllables under specified contextual constraints. The new word selection by semi-supervised learning is based upon consistent pronunciation predictions by different LTS models. After adding new words to the training set, the performance of LTS conversion is significantly improved. For the NetTalk dictionary, compared with the performance from the N-gram baseline model, 21.6% relative word error rate reduction is obtained. For the CMU dictionary, 9.1% and 5.6% relative word error rate reductions are obtained, respectively, with/without considering the stress.

Although technologies of both low-level and high-level descriptors for music information retrieval (MIR) are advancing, there are some essential deficiencies while utilizing them separately. In this paper we propose a model where the low-level and high-level descriptors collaborate to support semantics-based MIR. The ontology of ldquomusic scenerdquo domain is constructed as a demonstration, and a set of domain related low-level and high-level descriptor analyses are introduced. Given the domain ontology and the analysis results as input, an abduction process is adopted to compute the semantics-based interpretations. Evaluations show that the collaborative model does not only give a better recall rate of semantics-based retrieval than separated models, but also maintains a promising precision meanwhile.

To identify the language origin of a named entity, morphological information associated with its letter spelling, such as letter N-grams, is commonly employed. However, with this information only, named entities with similar spellings but from different language origins are difficult to differentiate. In this paper, a measure of "popularity," in terms of frequency or page count of the named entity in language-specific Web search, is proposed for identifying its language origin. Morphological information, including letter or letter-chunk N-grams, is used to enhance the performance of language identification in conjunction with Web-based page counts. Six languages, including English, German, French, Portuguese, Chinese, and Japanese (Chinese and Japanese named entities are shown in their corresponding phonetic alphabets, i.e., Pinyin and Romaji), are tested. Experiments show that when classifying four Latin languages, including English, German, French, and Portuguese, which are written in Latin alphabets, features from different information sources yield substantial performance improvements in the classification accuracy over a letter 4-gram-based baseline system. The accuracy increases from 75.0% to 86.3%, or a 45.2% relative error reduction.

Identifying the language origin of a name in English is important for generating its correct pronunciation. In this paper, N-grams of syllable-based letter clusters are proposed for the task. The performance of the N-gram model of a set of frequently used letter clusters (correspond to syllables) is compared to that of letter N-gram model in a four-language task: English, German, French, and Portuguese. On average, the letter cluster N-gram, which has 26% error rate, is slightly better than the letter N-gram, which has 27.2% error rate. Furthermore, it is found that the error distributions from the two N-grams have fairly large differences. Therefore, AdaBoost is used to combine the results from N-grams of different units. The error rate is reduced to 22.5% or a relative 17.5% error reduction is achieved after the combination