Jiali You

Identifying Language Origin of Person Names With N-Grams of Different Units

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

Nodes clustering and dynamic service balance awareness based virtual network embedding

Network virtualization is an important technique to allow multiple customized networks to cohabit on a shared substrate network. In order to put the network virtualization into practice, the virtual network embedding (VNE) problem that searching an efficient mapping of a virtual network request onto substrate network resources should be solved primarily. Since the VNE problem is known to be NP-hard, previous work focused on designing heuristic-based algorithms. In this paper, we proposed the nodes clustering and dynamic service balance awareness based virtual network embedding (NC-DSBA) that employs the divide-and-conquer strategy along with the concurrence policy. Besides, the proposed NC-DSBA introduced a dynamic service balance algorithm to assist the embedding and improve the efficiency. The simulation results show that the NC-DSBA reduces the computational complexity, increases the acceptance ratio and revenue-cost ratio, and balances the loads of virtual nodes and links in substrate network in the long run.

Optimal media service scheduling in clouds

Cloud becomes increasingly important currently, and media services, which usually have multiple steps and complexity computing requirement, are applied in it. Although this scheduling topic has been investigated a lot, most of works focus on balancing jobs or decreasing the execution time of one job. For large scale paralleling media services, the job request usually contains a series tasks, and how to schedule them to a cloud or a grid with high efficiency and low cost is an important problem. To achieve this target, we focus on the fine grain tasks for media services and propose a task-based scheduling algorithm, in which two-stage scheduling approach is designed, including initializing stage and adjustment stage. In this algorithm, after the initializing assignment process, in order to meet the need of the limited cost consumption, the characters of nodes or virtual machines(VMs), such as balance character, capability character, are considered deeply in scheduling adjusting process. Evaluation shows that our algorithm presents obviously better performance in terms of makespan, cost consumption, and balance factor than several benchmarks.

Source Routing Over Protocol-Oblivious Forwarding for Named Data Networking

Named data networking (NDN) has promising advantages, however, it is hard to deploy because of the lack of special hardware to support it. Besides, the lack of specific routing schema also restrains the advantages of NDN. To solve the problems of deployment and routing, we come up with the source routing schema over protocol-oblivious forwarding (POF) for NDN. POF is able to directly carry NDN packets without over IP because it provides an adaptable and programmable forwarding plane by defining a generic flow instruction set. Source routing is a flexible and scalable routing schema and provides the flexible and efficient implementation for the future network. The centralized management of the source routing schema over POF can provide the following properties: (i) quickly responding to the change of network, and (ii) forwarding the Interest packet to the best cache node. Experiments show that the source routing schema over POF for NDN can reduce more than 50% of the networking traffic flow and increase the cache hit rate by up to 25% more than the baseline schemas. Thus, the source routing schema over POF could be an implementation plan for NDN.

Analyzing and Forecasting the Performance of Video Service Providers

Video services are flourishing, and the same content might be hosted by different websites. 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 websites, 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.

An On-Site Elastic Autonomous Service Network with Efficient Task Assignment

With the increasing demand of the big data processing and the personalized services in the Internet, the current service platforms such as cloud are becoming hard to meet the requirement of multiple-level quality of services (QoS) within a short execution time. In this paper, an on-site elastic autonomous service network named SEA is proposed to cope with the above problems by utilizing the available resources from edge devices. The proposed SEA is constructed by the devices at the edge of the Internet and provides multiple-level QoS service to users. The architecture of the SEA is firstly designed to maintain the edge devices efficiently and monitor the status of the available resources. Then, an efficient scheduling algorithm based on discrete particle swarm optimization (PSO) is proposed to assign the tasks to proper devices in the SEA. The evaluation shows that the average execution time based on PSO is reduced more than 20% in comparison with a state-of-the-art algorithm. The simulation also shows that the hybrid SEA-cloud service is superior to the cloud-based service.

A distributed and virtualized infrastructure for networked multimedia services

This paper designs an infrastructure to meet the resource requirements, and more importantly, the QoS (quality of service) requirements for networked multimedia services. Learning from the successful experiences of distributed systems, the infrastructure deploys clusters of devices at the edge to improve the probability for end users to get satisfying QoS, and a two-layer management architecture is adopted accordingly. Moreover, a resource supply method based on virtualization and resource attributes is proposed. In the method, virtualized resources are packed and classified according to their purposes and functions, and each classification defines a minimum granularity, named service unit, to supply resources. Finally, to isolate traffic influence and guarantee network performance, each service is running on its isolated virtual network environment. The paper also presents the prototype system, and evaluates the feasibility and availability of it through experiments.

A behavior cluster based availability prediction approach for nodes in distribution networks

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 nodes 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.

Using Degree-Based Strategy for Network Coding in Content Distribution Network

In a content distribution network based on overlay networks, the peers upload bandwidth is always constrained. To optimize the system performance, one problem is how to assign the upload capacity to the right neighbors for transferring data blocks. In a system using network coding technology, a strategy which is to select the right neighbors based upon the degrees of them is proposed. The selection probability of each neighbor is determined by its degree. After. this strategy is applied, the utilization of network is improved, and the distribution progress is shortened. From the simulation, we found that the performance of content distribution is obviously better than the baseline system on three metrics, including maximum finish time, average finish time and standard deviation.

Improving letter-to-sound conversion performance with automatically generated new words

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.