Aiming at the problems of incomplete segmentation of lung lesions and fuzzy prediction of regional boundaries in mainstream deep learning networks， a Multiscale Dense Fusion Network （MDF-Net） based on U-Net was proposed. Firstly， multi-branch dense skip connections were introduced to capture multi-level contextual information， and Information Weighted Fusion （IWF） module was introduced at the end of the network for level-by-level fusion to solve the feature loss problem in the network. Secondly， a self-attention pyramid module was designed. Each pyramid layer was used to segment the feature map in different scales， and the self-attention mechanism was applied to calculate the pixel correlation， thereby enhancing the saliency of the infection features in local and global regions. Finally， unlike the up-sampling form in traditional U-Net， a Up-sampling Residual （UR） module was designed. The multi-branch residual structure and channel feature excitation were used to help the network restore more abundant features of micro lesions. Experimental results on two public datasets show that compared with UNeXt， the proposed network improves the ACCuracy （ACC） by 1.5% and 1.4% respectively， and the Mean Intersection over Union （MIoU） by 3.9% and 1.9% respectively， which verify that MDF-Net has better lung lesion segmentation performance.

Focused on the issue that complex image texture can not be fully expressed by single dictionary in image Super-Resolution (SR) reconstruction, a remote sensing image super-resolution algorithm based on compressive sensing and wavelet theory using multiple dictionaries was proposed. Firstly, the K-Singular Value Decomposition ( K-SVD) algorithm was used to establish the different dictionaries in the different frequency bands in wavelet domain. Secondly, the initial solution of SR image was obtained by using global limited condition. Finally, the sparse solution of multiple dictionaries in wavelet domain was implemented using Orthogonal Matching Pursuit (OMP) algorithm. The experimental results show that the proposed algorithm presents the better subjective visual effect compared with the single dictionary based algorithm. The Peak Signal-to-Noise Ratio (PSNR) and the Structural SIMilarity (SSIM) index increase more than 2.8 dB and 0.01 separately. The computation time is reduced as the dictionaries can be used once again.

Aiming at the problems of network flow monitoring (classification) in complex network environment, a stochastic artificial neural network learning method was proposed to realize the direct classification of multiple classes and improve the training speed of learning methods. Using Plane-Gaussian (PG) artificial neural network model, the idea of stochastic projection was introduced, and the network connection matrix was obtained by calculating the pseudo-inverse analysis. Theoretically, it can be proved that the network has global approximation ability. The artificial simulation was carried out on artificial data and standard network flow monitoring data. Compared with the Extreme Learning Machine (ELM) and PG network using the random method, the analysis and experimental results show that: 1)the proposed method inherits the geometric characteristics of the PG network and is more effective for the planar distributed data; 2)it has comparable training speed to ELM, but significantly faster than PG network; 3)among the three methods, the proposed method is more suitable for solving the problem of network flow monitoring.

Concerning the problems that Two-Dimensional Multiple-Input Multiple-Output (2D MIMO) systems only consider the effects of horizontal radiation pattern, ignoring the effects of vertical radiation pattern, and the closed-form on the sum rate of 2D MIMO system over K (Rayleigh/Gamma) fading channels involves special functions, two closed-form upper bounds on achievable sum rate of Three Dimensional Distributed Multiple-Input Multiple-Output (3D D-MIMO) systems with Zero-Forcing (ZF) receivers over K composite fading channels were proposed. The upper bounds considered Rayleigh multipath fading, Gamma shadow fading, geometric path-loss, 3D antenna radiation loss, and user distribution. The experimental results show that the obtained expressions accurately match with the Monte Carlo simulation conclusions.

Based on smooth spline principle, a self-adaptive multi-threshold segmentation algorithm HistSplineReg (Spline Regression for Histogram) was proposed. HistSplineReg is a two-step method. Firstly, a smoothing spline function was regressed to fit the one-dimensional image histogram, and then the extreme value was found by the regression function to achieve multi-threshold automatic segmentation of the image. Compared to the existing multi-threshold methods, the advantages of HistSplineReg lie in 5 aspects:1) it is quite consistent with the human intuition; 2) it is constructed on the smoothing spline, which is a solid mathematic basis; 3) both the number and the size of multiple thresholds can be automatically determined; 4) HistSplineReg can be analytically solved, and its computing burden is mainly concentrated on Cholesky decomposition of the matrix, while the size of matrix depends on the pixel level of the image, rather than the scale of the image; 5) it has only one trade-off parameter to balance the empirical error and regressor's smoothness. Furthermore, for the forest fire recognition task, an experimental reference value was provided. Finally, experiments were conducted on some digital forest fire images in the RGB (Red, Green, Blue) mode. The experimental results show that the histSplineReg method is more effective than Support Vector Regression (SVR) and Polynomial Fitting (PolyFit), which is based on the grayscale image, the color channel, the color image synthesized by each channel segmentation. And the three methods all reflect the red channel information is most significant to the forest fire image segmentation effect.

In extreme weather or traffic, passengers cannot get a taxi to the destination quickly, thus a taxi unified recommendation algorithm based on region partition was proposed to provide common taxi service and carpooling service. First of all, the region was regarded as the logo of journey, making the journey matching possible. Secondly, in the carpooling service, the similar routes of two passengers were matched in real-time to help passenger carpool sharing. Finally, the taxi with the minimum percentage of bypass time was selected to recommend to the user. The Global Positioning System (GPS) data of 14747 taxis was used to evaluate the proposed algorithm. Compared with CallCab system, the total mileage of the proposed algorithm was dropped by about 10%, while the carpooling time was only raised by 6% on average, as well as the total passenger mileage was reduced by 30%. Experimental results show that the proposed algorithm not only can significantly reduce the emission of automotive exhaust, but also has better performance in terms of time consumption.

Neural networks have strong nonlinear learning ability, so the super-resolution algorithms based on neural networks are preliminarily studied. These algorithms can only be used in controlled microscanning, which has uniform displacement between frames. It is difficult to apply these algorithms to uncontrolled microscanning. In order to overcome the limiting condition and obtain better super-resolution performance, a deblurring algorithm using Radial Basis Function (RBF) neural network was firstly proposed, which was then combined with non-uniform interpolation step to form a new two-step super-resolution algorithm. The simulation results show that the Structural SIMilarity (SSIM) index of proposed algorithm is 0.55-0.7. The proposed two-step super-resolution algorithm not only extends application scope of RBF neural network but also achieves good super-resolution performance.

To improve the convergence rate of digital repeater echo cancellation, firstly, the echo cancellation technique based on adaptive filter was studyed; secondly, the recursion algorithm of adaptive filter was improved by the technical schemes that two adaptive filters compute in parallel and update the weights jointly and recursively. Since the error signal to adjust weights of the two adaptive filters was generated in different ways, the schemes were divided into two categories: scheme one is that the weights of two filters were adjusted by error signal of echo cancellation (simultaneously); scheme two is that the weights of the first filter were adjusted by error signal as the difference value of received signal from antenna and output signal of the first filter; and the weights of the second filter were adjusted by error signal as the difference value of the above-mentioned error signal and the output signal of the second filter (separately). The simulation results show that the convergence rate of the echo cancellation is increased by 11.11%~17.78% in the improved technique scheme, so as to improve the condition effectively with the slow convergence rate of digital repeater echo cancellation.

Most training simulators of military power cannot give reasonable evaluation to the operation process of soldiers, because of being lack of scoring system. Therefore, in this paper, the scoring system for training simulator of military power based on the operating rule of actually weapon and the knowledge of large-scale system theory and expert system was analyzed. The operating rule of the military power and the key technologies of scoring system such as building up scoring rule, selecting scoring standard, selecting coefficient for the score deductions, calculation of the scoring system and program process of the scoring system were deeply analyzed in this paper. At last, a very useful scoring system for the training simulator of gas turbine generator was developed.

According to the research of distributed SIP flooding attack detection and defense, in combination with the characteristics of IP-based distributed flood attack and SIP messages, the two-level defense architecture against SIP distributed flooding attacks (TDASDFA) was presented. Two-level defensive components made up TDASDFA logically: the First level Defense Subsystem (FDS) and the Second level Defense Subsystem (SDS). FDS coarse-grained detected and defended SIP signaling stream to filter out non-VoIP messages and discard SIP messages of the IP addresses exceeding the specified rate to ensure service availability| SDS fine-grained detected and defended SIP messages using a mitigation method based on security level to identify the cunning attacks and low-flow attacks with obvious features of malicious DoS attacks. FDS and SDS detected and defended network status in real-time together to weaken SIP distributed flooding attacks. The experimental results show that TDASDFA can detect and defend SIP distributed flooding attacks, and reduces the probability of SIP proxy server or IMS server being attacked when the network is on the abnormity.

security devices(e.g.firewalls,IDS’s,anti-virus tools etc) that have been widely adopted in enterprise environments may generate huge amounts of independent,raw attack alerts,which are characterized by high false positive ratio and false negative ratio.As a result,it is difficult for users to understand these alerts and respond correspondingly.Therefore,handling the huge number of alerts produced by security devices is becoming a critical and challenging task in network security research.A general approach for solving this problem is to do some correlation analysis with these alerts and build attack scenario.A general survey of the contemporary alerts correlation algorithms was given in this paper by a straight forward classification paradigm,and some problems for future research were addressed.

TURN protocol is a technique for simple traversal of UDP through NAT. On the basis of RFC3489, the draft for TURN protocol was researched and analyzed in detail. Its address translation table was modified and simplified, which stored dynamic allocated addresses, and the working mode and application model of TURN technology were designed. Then, consulted STUN design ideas, a prototype system of TURN server was designed and implemented, which solved the problem that SIP UA cant traverse symmetric NAT by STUN.