Metamorphic mechanisms are widely studied and applied in theory and practice because of their characteristics of variable topological structures. The adjacency matrix is used in representation and operation of link changes. But this method is not available in configuration transformation with multiple joints and characteristics change of joints of metamorphic mechanisms, also the information of the matrix is poor. So it doesn't represent the configurations of metamorphic mechanisms intuitively. This article used adjacency graph to represent the topology of mechanism. The adjacency matrix of joint was proposed to represent the concrete information of the metamorphic mechanism. The metamorphic equation is constructed by multiplying elementary matrixes, and it is used to describe the transformations from one configuration to another. The verified examples indicate that the results obtained with the method have intuitive and comprehensive information of the configuration, and the method is easy to computer aided analysis.
In order to eliminate the interference in WLAN, an interference self-optimization scheme was proposed based on polynomial window function interference model. The dynamic channel allocation was adopted to acquire optimum channel configuration for each AP. The channel configuration strategy was based on adjacent channel interference measurement and data fitting by polynomial window function while the object was minimizing the average interference. Simulation and demonstration show that the proposed scheme can effectively reduce the network interference meanwhile increase the network throughput.
A new approach for multiple wireless networks access selection based on finite Moran process, namely ASFMP, was proposed, in which the process of selection strategy is described as a finite population game model, and each strategy evolution is proved to be a random birth-death process. The local update mechanism is then improved from a multi-strategy perspective, which is used to reveal the revolution mechanism of selection strategies. Simulation shows that the proposed method can convergence to Nash equilibrium within finite evolution times, and can also avoid the "Ping-Pong effect" caused by infinite population game.
A model of sectionalize control based on predictive control was presented for problems of vision-guided AGV path tracking. Control problems with inequality constraints were sectioned. The turning radius as constraint conditions was made, so that, the problem of the selection of objective function weights during the optimization process is solved. In each control period, a single-step prediction method was used to find the optimal target point, and obtain the optimal control variables directly by analytical methods under the input constraints and state constraints. The speed of the optimization problem is improved. This method fully meets the need of real-time controls, and it is used to calculate and small amount of computation. The validity of the model predictive is proved by the Simulation results.
A related contents gathering based cache policy called cache gathering (CG) was proposed to solve the routing scalability problem brought by existing cache policies in content centric network. By the mutual attraction between the contents with same feature, called related contents, the related contents gathered on nodes, which facilitated the cache contents feature abstraction, so the routing advertisement was decreased and the routing scalability was improved. Meanwhile, with the strategy of lifetime-increasement between contents with main feature, CG was able to reduce the update frequency of cache content, so the routing reliability was improved. On this basis, the general CG algorithm was proposed. The simulation proved the reasonableness and efficiency of the schemes in this thesis.
Depending on the intrinsic weakness and advantages of back propagation(BP) neural network and Fast Independent Component Analysis(FastICA), a Fast Independent Component Analysis(FastICA) Genetic Neural Networks Method was proposed for fault characteristic signal recognition. The FastICA is used to decompose signals to obtain the independent components successively, each of Independent components corresponding to an energy band, and feature vector of each energy band is used as input sample to optimize neural network. Secondly, the genetic algorithm is used to optimize the weights and thresholds of BP neural network to obtain the Genetic Neural Network. Thirdly,the feature vector is used as input sample of the genetic neural network to identify the fault. Using this method can analysis and identify many kinds of rolling bearings fault signal, and through this method the ability of fault identification isimproved.
To speed up the convergence of the bilinear alternating least squares (BALS) algorithm of fitting the parallel factor (PARAFAC) model, an improved algorithm of fitting the PARAFAC model was proposed. In each iteration, the proposed algorithm sets up their own relaxation factors for two loading matrices which are required to be estimated, and gets the optimal couple of two relaxation factors by the joint optimization. Analysis and simulation show that the proposed algorithm improves the speed of fitting the PARAFAC model without performance deterioration compared with the existing BALS algorithm.
An adaptive federated Kalman filter model for multi-positioning system used in urban was proposed.Firstly, the credibility of subsystems was evaluated by estimating the position errors and the credible factors was obtained.Secondly, the information sharing factors of federated Kalman filter were assigned by credible factors to adaptive filtering.To assess the filtering effect, a new assessment method was proposed. Simulation demonstrates the effectiveness of the filtering algorithm and assessment method.
The neural network learning models was proposed, focusing on complexity and particularity of product sales in garment industry. The network model was established to predict the impact on garment sales on basis of factor analysis and optimized by using genetic algorithms for each connection weights of back propagation (BP) neural network. The method combines the strong learning ability of the BP neural and the global search capability of genetic algorithms.
An improved clipping and filtering algorithm was proposed to improve the currently used algorithms for reducing PAPR of OFDM with high computational complexity and serious distortion. The algorithm divides OFDM signals into odd signals and even signals. And the two kinds of signals are clipped and filtered. Simulation shows that, in comparison with currently used clipping and filtering algorithms, the algorithm is with better PAPR performance with lower computational complexity and filters in-band interference more obviously. The BER performance of the system can be improved through the algorithm.
To reduce the complexity, the existing algorithms treat the residual inter-carrier interference (ICI) as white noise,which causes a decline in system performance. Consequently, an optimal linear ICI cancellation algorithm based on whitening residual ICI and noise was proposed. Firstly, the whiten matrix is constructed to whiten the residual ICI and noise simultaneously. Then, the optimal linear ICI cancellation matrix was deduced by maximizing the achievable rate in orthogonal frequency division multiplexing (OFDM) systems, which can obtain the time diversity embedded in time variant channels combining with successive interference cancellation. Simulations show that the proposed algorithm obviates the performance loss introduced by band approximation in traditional algorithms and can apply to OFDM systems in the channel estimation accuracy at present.
To make network coding based multipath routing(NCMR)reliability transmission mechanism more practiced, local retransmission and network coding based multipath routing(LR-NCMR) reliability transmission mechanism was proposed for wireless Ad hoc network, transmitting coded packets over multipath and using retransmission within local area. Comparing with NCMR, the simulation shows that LR-NCMR is more reliable than NCMR and it can reduce the redundancy in the network.
To improve the accuracy and efficiency of cancer gene expressing data clustering, Quantum-behaved particle swarm with comprehensive learning strategy(CLQPSO) and generalized regression neural network (GRNN) are studied, A cancer gene clustering algorithm was generated based on CLQPSO. GRNN takes advantage of the implicit rules in a number of similar genes and the prediction of missing values for gene expression has higher credibility; CLQPSO algorithm can make full use of each particle best position and particle swarm social cooperation information offered, avoiding premature convergence in local optimum value. Experiments show that the integrated use of GRNN and CLQPSO algorithm has better clustering performance and global convergence compared with K-Means, spectral clustering, discrete particle swarm algorithm in the aspect of cancer gene expressing data clustering.
In conventional multicast scheme (CMS), the total throughput of multicast group is constrained by the user with the worst channel quality. In order to overcome the limited throughput problem, a multicast scheduling was considered based on layered coding. A new subcarrier and bit allocation algorithm is exploited for targeting the maximum throughput of a whole multicast group while at the same time guaranteeing the quality of services (QoS) requirements of all users. The article proposed an optimal resource allocation algorithm in the downlink of OFDMA-based wireless multicast group. A two-phase suboptimal algorithm was proposed as well to reduce the computational complexity. Simulations show that the performance gap between the optimal algorithm and the proposed suboptimal algorithm is quite small. The proposed algorithm significantly outperforms CMS. Moreover, it obtains more throughput than Tian's algorithm.
A new synchronization circuit was proposed. Due to requirement of the GHz sampling D/A converter, the circuit employs high-speed dynamic comparators and flip-flops to receive the input data from the low voltage differential signaling (LVDS) interface, which has the advantage of low power and low complexity. At the same time, this circuit adopts a low jitter analog delay locked loop and digital phase detector to obtain the proper synchronous clock, thereby, the clock frequency range of the synchronous circuit can be improved. Based upon the SMIC 0.18 um 1.8 V CMOS process, the simulation gives that the clock frequency of the synchronization circuit is within the range of 250~800 MHz, and the data rate is 500 Mbps~1.6 Gbps. The circuit can be used in the synchronization of the GHz sampling DAC core and the external LVDS transmitter interface.
The chirp extraction of self-similar pulses was studied based on passively mode-locked Yb-doped fiber laser system and passively mode-locked Er-doped fiber laser system. The authors use the time-frequency analysis methods of Short-time Fourier Transform and Wigner-Ville Distribution to extract the chirp in mode-locked fiber laser system. The result was proved through simulation experiments analysis The advantages and disadvantages of the two time-frequency analysis methods by statistical characteristics were given. The results provide the theoretical and experimental foundation for judging pulses are self-similar pulses generated by the mode-locked fiber laser system.
The network throughput is an important specification to evaluate the network performance, and the ability of network processing is difference due to the proportion of overhead in packets. The test network was designed based on FPGA, which rate can reach line speed. The work status of slave machine is controlled by host, which sends back the test results. The uplink and downlink throughput can be tested by this way, further, the network performance is evaluated. It shows that method with cross subnet test throughput of the network can be used for different nodes of the network.
An energy-saving scheme for wireless sensor networks based on network coding and duty-cycle (NCDES) was proposed. The scheme determines the node's status based on the ID information which embedded in data information. When combining network coding and duty-cycle in wireless sensor networks, it will reduce transmission coding coefficients and retransmissions. The energy efficiency of the network increases as more volume of data will be transmitted to the sink with same number of transmissions. The optimal value of energy consumption for the multi-hop network was proved by analyzing the network topology and energy model. In contrast to RDCNCode and AdapCode++, the simulation shows that NCDES has a great improvement on data delivery ratio and energy efficiency, it extends the network lifetime about 4.02% and 8.51%, and improves the packet delivery rate about 14.83% and 4.65%.
3D-beamforming could use the vertical dimension for antenna pattern adaption, it can provide an additional degree of freedom for transmission signal optimization. A global adaptive 3D-beamforming algorithm based on uniform planar array and a dimension reduction vision algorithm based local adaptive algorithm was proposed. 3D-beamforming algorithm could adjust the antenna weight adaptively according to the user's move or the change of the wireless condition and alter the beam pattern in real time. Simulation shows that the 3D-beamforming aimed at specific user can be implemented by both of the proposed algorithms under a certain number of antenna arrays.
For purpose of enhancing network bandwidth utilization and for a balance between network traffic and transfer efficiency, based on the network traffic autoregressive technology, a new network traffic redundancy elimination algorithm called ANTREA was proposed. It splits data transfer missions into transfer units. The data in one transfer unit are executed in two ways, one is traditional traffic redundancy elimination, and the other is direct data transfer. A transfer unit makes up models of network situation, and predicts the time cost of checking duplications and the available bandwidth. So, it adjusts the size of direct data transfer according to the result of prediction. Experiments show that ANTREA algorithm can adjust its transfer strategy according to the network situation and utilize network bandwidth sufficiently to achieve higher transfer efficiency. It is of better flexibility on network situation than EndRE and has almost 7 times transfer throughput than EndRE in network with 10MB/s bandwidth.
To solve the problem of low recommendation precision and data sparsity in recommendation, a recommendation method based on user trust network was proposed. A weighted user trust network was constructed as well based on basic social network, including the trust relationship, the influence of user role, the similarity of user attribute and similarity of user preferences. Based on the trust network, a key path discovery algorithm was proposed to find user trust network within constrains, which is further utilized for recommendation. The comparison experiments were conducted on Filmtipset dataset. A comparative analysis was made aimed at factors that influence the quality of recommendation. It was shown that the method based on user trust network can achieve better recommendation.
A fault detection method for wireless sensor network based on improved Kruskal algorithm was proposed. It adopts the centralized improved Kruskal algorithm to obtain credible node set. According to the credible node, it uses the distributed adjacent node comparing algorithm to locate the fault WSN node by analyzing and processing the sensing value. Time redundancy is also employed for tolerating the transient faults in sensing and communication. Simulation shows that, even the fault note rate raises to 35%, the proposed method can still locate the fault node quickly, and can also guarantee a high accuracy.
In multi-cell scenario, the uneven distribution of machine-to-machine (M2M) terminals seemly leads to problems of system access throughput decrease and user access delay increase. To address these issues, a M2M load balancing algorithm was proposed. The algorithm builds up a model associated with cell signal quality and cell load factor. The problem approximates a convex optimization one, thus realizing load balancing. Simulation shows that, through employment of the proposed algorithm, the performance of access throughput and access delay is improved greatly.