In view of the problem that conventional incremental learning models may go through a way of performance degradation during the update stage, a kernelized incremental learning method was proposed based on objective equilibrium measurement. By setting the optimization term of "empirical risk minimization", an optimization objective function fulfilling the equilibrium measurement with respect to training data size was designed. The optimal solution was given under the condition of incremental learning training, and a lightweight incremental learning classification model was finally constructed based on the effective selection strategy of new data. Experimental results on a publicly available fall detection dataset show that, when the recognition accuracy of representative methods falls below 60%, the proposed method can still maintain the recognition accuracy more than 95%, while the computational consumption of the model update is only 3 milliseconds. In conclusion, the proposed method contributes to achieving a stable growth of recognition performance as well as efficiently decreasing the time consumptions, which can effectively realize wearable devices based intellectual applications in the cloud service platform.

The network replication technology has become the cornerstone of the evaluation platform for network security experiments and the system for network emulation. Facing the requirements of fidelity and scalability of network replication, a multi-scale network replication technology based on cloud platform for the fusion of lightweight virtualization, full virtualization and digital simulation was proposed. The architecture of the seamless fusion of these three scales was introduced at the beginning; And then the network construction technology based on the architecture was studied. The emulation experimental results show that the emulation network which is built with the construction technology has the characteristics of flexibility, transparency and concurrency; in addition, the construction technology is capable of emulating networks with high extensibility. At last, communication tests for a variety of protocols and simple network security experiments on the large-scale emulation network were conducted to verify the availability of this large-scale emulation network. The extensive experimental results show that the multi-scale network replication technology for the fusion of virtualization and digital simulation can be used as the powerful support for creating large-scale emulation networks.

Aiming at the problem of advanced persistent threats in power information networks, a hybrid Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN) intrusion detection model was proposed, by which current network states were classified according to various statistical characteristics of network traffic. Firstly, pre-processing works such as feature encoding and normalization were performed on the network traffic obtained from log files. Secondly, spatial correlation features between different hosts' intrusion traffic were extracted by using deformable convolution kernels in CNN. Finally, the processed data containing spatial correlation features were staggered in time, and the temporal correlation features of the intrusion traffic were mined by RNN. The experimental results showed that the Area Under Curve (AUC) of the model was increased by 7.5% to 14.0% compared to traditional machine learning models, and the false positive rate was reduced by 83.7% to 52.7%. It indicates that the proposed model can accurately identify the type of network traffic and significantly reduce the false positive rate.

Bernoulli-Gaussian (BG) model in Expectation-Maximization Bernoulli-Gaussian Approximate Message Passing (EM-BG-AMP) algorithm is constrained by its symmetry and restricted in the approximation of the actual signal prior distribution. Gaussian-Mixture (GM) model in Expectation-Maximization Gaussian-Mixture Approximate Message Passing (EM-GM-AMP) algorithm is a high-order model of BG model and has quite high complexity. In order to solve these problems, the Bernoulli-Asymmetric-Gaussian (BAG) model was proposed. Based on the new model, by further derivation, the Expectation-Maximization Bernoulli-Asymmetric-Gaussian Approximate Message Passing (EM-BAG-AMP) algorithm was obtained. The main idea of the proposed algorithm was based on the assumption that the input signal obeyed the BAG model. Then the proposed algorithm used Generalized Approximate Message Passing (GAMP) to reconstruct signal and update the model parameters in iteration. The experimental results show that, when processing different images, compared to EM-BG-AMP,the time and the Peak Signal-to-Noise Ratio (PSNR) values of EM-BAG-AMP are increased respectively by 1.2% and 0.1-0.5 dB, especially in processing images with simple texture and obvious color difference changing, the PSNR values are increased by 0.4-0.5 dB. EM-BAG-AMP is the expansion and extension of EM-BG-AMP and can better adapt to the actual signal.

Because existing passenger-finding algorithms do not consider taxi's spatio-temporal context, a collaborative filtering recommendation algorithm of taxi passenger-finding based on spatio-temporal context was proposed. The proposed algorithm mapped potential passenger locations to space network, and introduced time delay factor to similarity measure to get the neighbor set which was similar to a target taxi's driving behavior. Based on location context, the proposed algorithm chose the target taxi's most interest potential passenger location from similar neighbor set. The experimental results on Fuzhou taxi trajectory data show that the proposed algorithm can get the best recommendation result when the time delay factor is 0.7. Meanwhile, compared to the traditional collaborative filtering recommendation algorithms, the proposed algorithm obtains better recommendation result under the neighbor sets with different size, which means the proposed algorithm is more accurate than the traditional collaborative filtering algorithms.

The model parameters of Video Compressed Sensing of Linear Dynamic System (CS-LDS) can be estimated directly from random sampling data. If all video frames are sampled in the same way, the sampling data will be redundant. To solve this problem, an adaptive improvement algorithm based on adaptive compression sampling technology was proposed in this paper. Firstly, a Linear Dynamic System (LDS) model of the video signal was established. And then the sampling data of video signal was obtained by using the adaptive compression sampling method. Finally, the model parameters were estimated and the video signal was reconstructed by the sampling data. Without affecting the video reconstruction quality, the experimental results show that the proposed algorithm is better than the CS-LDS algorithm, it can not only reduce 20%-40% sampling data in the uniform measurement process, but also save the average running time of 0.1-0.3 s per frame. The improved algorithm reduces the number of samples and the algorithm's running time.

In the prediction of moving object trajectory, concerning the low accuracy rate of low order Markov model and the expansion of state space in high order model, a dynamic adaptive Probabilistic Suffix Tree (PST) prediction method based on variable length Markov model was proposed. Firstly, moving objects trajectory path was serialized according to the time; then the probability characteristic of sequence context was trained and calculated from the historical trajectory data of moving objects, the probabilistic suffix tree model based path sequence was constructed, combined with the actual trajectory data, thus the future trajectory information could be predicted dynamically and adaptively. The experimental results show that the highest prediction accuracy was obtained in second order model, with the order of the model increasing, the prediction accuracy was maintained at about 82% and better prediction results were achieved. In the meantime, space complexity was decreased exponentially and storage space was reduced greatly. The proposed method made full use of historical data and current trajectory information to predict the future trajectory, and provided a more flexible and efficient location-based services.

The identification of one-way road state can provide relevant information of road network to the public timely and accurately, improve the efficiency of public travel, and enhance the service level of dynamic traffic information. This paper presented a dynamic identification algorithm of one-way road state based on Floating Car Data (FCD). Firstly the line feature information of maps was got, and the matching of spatial information grid with the traffic roads was pretreated to achieve fast matching for massive FCD; Then statistical characteristics of FCD direction information was analyzed to filter dual-threshold information and direction information; Finally one-way road state information was got dynamically. The actual road network tests show the algorithm can identify one-way road state information effectively.

In order to achieve dynamic update of digital map data to support the geographic information services in traffic network with rapid development, a new-road automatic detection algorithm was proposed based on the Floating Car Data (FCD) technology. In this method, the moving trajectories of massive floating cars were calculated in real-time, then the suspected new road sets were extracted with the image matching between the existing map layers and the trajectories. After applying a filtering algorithm to the data sets for cleaning, the new road detection reports covering the new roads' location and length were generated automatically and saved as temporary map layers. The field test results show that this algorithm can detect the new roads quickly, so far as to detect new road within five minutes. It is a cost-effective solution for the real-time road map layer update.

Based on the research of two-dimensional code and recognition of mobile phone, the authors have developed a two-dimensional code recognition system on Android platform. In the system architecture design, the two-dimensional code recognition system platform support layer, client application layer, cloud handle layer were hierarchically designed to ensure the system security. According to the requirements, the system function models and cases analyses were given. Finally, the two-dimensional code core encode and decode functions of mobile phone were achieved.

The multi-hop transmission of multimedia emergency communication system based on Wireless Mesh Network (WMN) in underground mine have two problems: low basis bandwidth and high multi-hop transmission attenuation. This paper aimed to improve the multi-hop transmission performance for the system. In this paper, a trunk line network structure of multimedia emergency communication system based on WMN in under-ground mine was proposed. The authors established its transmission model, and then had a research on the main factors that affected the transmission performance. The multi-radio node structure of multi-hop mesh backbone network based on 802.11n was proposed and solved the two problems of multi-hop transmission. The experimental results show that it has more than 165Mbps basis bandwidth, and under the limited 60Mbps environment, the bandwidth attenuation of per hop is less than 1%, basically satisfying the application requirements of multimedia transmission in underground mine.

The multicast routing method has been applied into the Network on Chip (NoC) since traditional unicast communication cannot meet the increasingly rich application requirements of NoC. Three kinds of path-based multicast routing algorithms including XY routing, UpDown routing and SubPartition routing algorithms were applied to 2D Mesh or Torus NoC. The congestion control strategy was proposed. The simulation results show multicast routing algorithms have shorter average latency and higher throughput and balanced applied load compared with unicast routing algorithms. SubPartition routing algorithm was confirmed to have a more stable and better performance as the network size increases. Finally, multicast congestion control techniques for NoC were employed to make multicast communications more efficient and enhance the NoC performance.

Argumentation model is the theoretical basis to establish group argumentation environment. Based on Issue-Based Information System (IBIS) model and Toulmin' argument schema, a group argumentation model was proposed, which was able to evaluate the argumentative utterance. With this model, the group argumentative information could be structured as a graph which consisted of utterance nodes and semantic links. A method of evaluating utterance nodes based on Language Weighted Aggregation (LWA) operator and node reduction was proposed. A group argumentation on the issue of system architecture design was illustrated as an example to show the usability and effectiveness of the proposed model.