Aiming at the problem of robust speech recognition for Vocal Effort (VE) variability, a speech recognition algorithm based on multi-model framework was presented. Firstly, changes of acoustic characteristics under different VE modes, as well as influence of these changes on speech recognition, were analyzed. Secondly, a VE detection method based on Gaussian Mixture Model (GMM) was proposed. Finally, the special acoustic models were trained to recognize whisper speech if the result of VE detection was whisper mode; otherwise articularoty features, in company with spectrum features, were introduced to recognize speech of the remaining four VE modes. The experiments conducted on isolated-word recognition show that significant improvement of recognition accuracy can be achieved when using proposed method: compared with the baseline system, the mixed corpus training method and the Maximum Likelihood Linear Regression (MLLR) adaptation method, the average character error rate of five VE modes is reduced by 26.69%,14.51% and 15.30% respectively. These results prove that articularoty feature is more robust than the traditional spectrum feature when confronting VE variability, and the multi-model framework is an efficient method for robust speech recognition related to VE variability.

To overcome the salient extraction results cannot preserve edge and enrich the inner details when extracting image salient region, a new multi-scale extraction approach based on frequency domain was proposed. In order to remove redundant information and get the innovation, the image was Fourier-transformed to get the spectral residual on multiple resolutions. Then normalization processing was applied to obtain the final saliency image. The simulation results show that the proposed method has good visual effect, which can keep the edges of salient region and highlight the whole significant target uniformly at the same time. The area under Receiver Operating Characteristic (ROC) curve of these results also has satisfied performance.

With the rapid development of Service Oriented Computing (SOC), more and more Web services with the same functionalities and different Quality of Service (QoS) are available on the Web, QoS becomes an important basis for evaluating and selecting Web services. As the environment of Web services is open, the QoS of Web services are dynamic. How to accurately predict Web services' QoS becomes a key factor that affects the success of QoS based Web service selection and composition. Here existing Web service QoS prediction methods were deeply analyzed, and the advantages and disadvantages of these existing methods are pointed out, and the future research directions of Web service QoS dynamic prediction were explored.

Considering that error propagation cannot be avoided at relay node in Orthogonal Amplify-and-Forward (OAF) scheme for network coding, orthogonal cooperative scheme for network coding with Decode-and-Forward (DF) protocol was proposed in this paper, in which whether to forward the source signals was determined by the decoding results at the relay node. In the proposed scheme, the error propagation was avoided and system reliability was improved. The theoretical analysis shows that the proposed scheme outperforms OAF scheme in terms of average mutual information and reliability. The simulation results demonstrate the validity of the theoretical analysis.

Random number and broadcast polynomial for every session were created. The legal nodes could recover the lost legal group keys by themselves independently according to their private information and broadcast messages, and the collusion attack between newly joined nodes and revoked nodes was resisted. New private information was distributed to revoked nodes so as to rejoin group in later sessions. Through the security and efficiency analysis, the scheme has less communication cost yet still achieves security property, suitable for mobile Ad Hoc network.