- EI核心期刊
A completely new approach to artificial intelligence, named mechanism approach, was presented in the paper to cope with the advanced intelligence(AI) research. It was discovered that the core mechanism of intelligence formation is a series of conversions—the one from information to knowledge and the one from knowledge to intelligence (intelligent strategy), or more concisely informationknowledge-intelligence transform. It was also discovered that there is a system of knowledge ecology from experimental knowledge to regular knowledge and further to commonsense knowledge. To combine the mechanism approach and the ecology system of knowledge, it is discovered that the three existed schools of advanced intelligence theory (the structuralism, functionalism, and behaviorism) are unified within the framework of the mechanism approach. More interestingly, it can also be proved that the general form of mechanism approach is the information conversion which is an effective approach to advanced intelligence and as significant as that of material conversion and energy conversion in physics.
The metabolic grey model (MGM) is investigated for network traffic prediction. It is shown that, when the MGM modeling-length is shorter than the traffic primary-period length, the accuracy is satisfactory. Small modeling-length MGM is preferred. Residual grey model (RGM) is not needed for its contribution to accuracy improvement is limited. The prediction of MGM, auto regressive integrated moving average（ARIMA ）and Elman neural network(ENN) are compared. The MGM accuracy is better than the ARIMAs, and equals to the ENNs. Furthermore, MGM is adaptive to traffic changes.
The problems of the joint subcarrier allocation, power allocation, transmission mode selection and relay selection optimization which maximizes the weighted sum rate are addressed in orthogonal frequency division multiple access-based decode-and-forward(OFDMA-DF) relaying systems. Based on convex optimization theory, an optimal resource allocation algorithm is proposed. The complexity of this algorithm is only linear in the number of subcarriers. Simulation demonstrates that the scheme can achievenot only fair resource allocation but also effective relay utilization that enhances the system capacity through adjusting weighting factors for each user.
Aiming at the high complexity of non-linear combined quantization caused by constellation biased distribution in lattice reduction aided(LRA) multiple input multiple output (MIMO) detection, a sorted feedback non-linear quantization scheme is proposed. It considers the border of transformed constellation and the correlations between elements, meanwhile, the element is selected to be quantized according to the value of the quantization errors. Simulation shows that the performance of LRA MIMO detection based on this quantization scheme is superior to the performance of LRA MIMO detection based on non-sorted feedback and independent element quantization. Furthermore, the LRA MIMO detection based on the proposed scheme can approach the performance of maximum likelihood detection in polynomial complexity in a flat fading channel.
Based on equivalent matrice and tree search, codebook-based precoding selection strategy is presented for four transmission antenna ports in the 3rd Generation Partnership Project long term evolution（3GPP LTE） system. Sixteen codebooks are equivalent to five ones, and then the best codebook is found by the tree search method. Compared with the optimal codebook selection scheme, the proposed selection strategy can reduce codebook number and computation complexity effectively. Simulation results demonstrate that the performance of the proposed strategy near to that of the optimal codebook selection scheme.
A wireless mesh transport control protocol (TCP) network coding scheme based on rate-matching receiver is proposed. A network coding(NC) layer to code transport control protocol data(TCP-DATA) and transport control protocol ack(TCP-ACK) on mesh connectivity layer (MCL) is constructed, and modified distribution coordination function request to send/clear to send (DCF RTS/CTS) mechanism to match data rate in the receiver is adopted. Actually a real 3 hops wireless mesh linear network has been constructed, capturing the TCP network coding data based on the proposed scheme, therefore, it is concluded that the TCP throughput of the end-to-end system is approximately 20% higher than non
-NC scheme.
To build an efficient, scalable, and security trust model in peer-to-peer (P2P) networks, the P2P trust model is studied and a risk-aware P2P trust model is built. This model constructs a new relationship model between risk and reputation, and interlards recommendation reputation with risk. By means of the new risk evaluation function and the computation of probability model in series, the risk brought by direct exchange and indirect recommendation is well quantified. To avoid slander or coordinated cheat attack among malicious nodes, this model filters the local recommendation reputation and local recommendation risk with the center deviation method. Through P2P networks simulations, the performance gains brought by this model are significant and the system security is greatly improved.
To decrease the acquisition time and complexity of Gold code acquisition, m-sequence iterative acquisition algorithm is applied to Gold code. In considering global position system (GPS) C/A code, a group of sparse check relations are found according to creating polynomials of Gold code. And for the purpose of decreasing the complexity of iterative computation, the ordinal check and mixed check method are proposed. Finally, the influences on iterative acquisition performance made by different check relations and different check method are verified by simulation, it shows that the ordinal check with degressive iterative number is getting optimized and effectively decrease the complexity and acquisition time.
Joint design of the carrier frequency offset (CFO) estimator and adaptive training sequence for multiinput multioutput orthogonal frequency division multiplexing (MIMO-OFDM) system is investigated. With adaptive training sequence, which is optimized according to channel state information (CSI) available at the transmitter, proposed CFO estimator can update Schenks estimator and then reduce the effect of channel noise and further increase the accuracy of CFO estimation in fading channels.
To improve the calculative precision of gaze direction, an eye gaze tracking system using stereovision information is designed, it composes of two charge coupled device cameras and a novel loop illuminator. Pupil position is got through doing subtraction of the two adjacent images, “bright pupil” and “dark pupil”, which are gained by illuminating users eyes alternately. Local gazing direction is the vector difference between the Purkinje image in “dark pupil” and pupil center located by ellipse fitting. Pupil tracking is designed using neural network combined with Kalman filtering, and mapping model from 3D parameters to gaze point is worked out via support vector regression. Experiment proves that this system can most eliminate the interference from users head motion.
In distributed multi-user opportunistic spectrum access (OSA) environment, user-synchronization problem is the vital reason for network throughput deteriorating. To resolve this problem, a changeable-domain based sensing policy is proposed. By adopting randomized sensing policy it can weaken user-synchronization. When changing the number of channels which are available for randomized access (i.e. the width of the random domain), the proposed policy can achieve a tradeoff between optimal myopic sensing and user-synchronization avoidance. Simulations show that the policy can achieve higher network throughput compared to the randomized sensing policy over the whole channels.
An image segmentation method based on the modified intersecting cortical model (MICM) is proposed to set the MICM parameters adaptively according to the different characteristics of images and choose the optimal segmentation results automatically, which are two main obstacles for the basic intersecting cortical model(ICM) to be used in practice. Experiments show that the comprehensive evaluation value of MICM is close to those of basic pulse coupled neural network(PCNN) and basic intersecting cortical model. Compared with the fuzzy C-means algorithm and OTSU algorithm, MICM is of visually better segmentation and the comprehensive evaluation value of MICM increases by approximately 15% and 13% respectively.
Based on discrete Fourier transform(DFT) estimator, the effect of virtual subcarriers is analyzed. It is shown that, on edge subcarriers, the estimation errors caused by channel information missing are much larger than that caused by noise. Therefore the estimation errors of edge subcarriers are getting worse. A low complexity channel estimator is proposed. The proposed estimator takes the least square(LS) estimation instead of the DFT estimation on edge subcarriers. Then replacement criteria and the bound of mean square error(MSE) are presented. The simulation shows that the MSE and BER performances of the proposed method are better than that of LS estimator or DFT-based estimator.
A relay cooperative scheme is proposed for cellular orthogonal frequency division multiplexing (OFDM) system: source station reorganizes and recombines the bits stream, destination station combines the signals from the two independent paths: direct and relay links, signals are redesigned. Furthermore, a supplementary strategy of sub-carrier symbol re-allocation is proposed to enhance the system. Simulations show that the proposed scheme achieves transmit diversity and improves system performance.
A flexible and portable WiMAX MAC (FPWiMAC) protocol stack software system is proposed to provide a reference design for WiMAX system. By introducing an operating system (OS) abstract layer and a physical layer (PHY) management subsystem, FPWiMAC is independent to operating systems, PHY implementations and hardware platforms. System flexibility is improved by features of dynamic algorithm library, core database and so on. Moreov er, the key mechanisms, including downlink data scheduling, two-dimension data mapping, dynamic ranging region allocation, are designed to improvesystem performances. The tests show that the flexibility and portability of FPWiMAC are good. The features of high bandwidth and low latency can support the quality of service (QoS) requirements of real-time services in IEEE80216 networks.
In order to perform service monitoring and classifying with high accuracy but lower error rate, and to expand the usage scope of the algorithm, a notion of constant-step circular track is presented, turns into residue class circular track problem. Meanwhile, a multi-track separating algorithm based on bubble principle is proposed. It separates tracks in accordance with descending order in attribute values by using the feature of variances about multi-tracks in turn, without applying the traditional separating method by track interval. It is indicated that this approach is with highly accurate and convergence for detecting shared access hosts so as to meet needs of existing network.
A proposal of efficient certificateless signature scheme is presented to insecure against public key replacement attack. It is shown that an adversary who replaces the public key of the original signer can forge valid proxy delegations for the corresponding proxy signer without knowledge of the signers private key, and can even forge valid proxy signatures. To thwart this attack, an improved scheme is further proposed, which is not only more securer but also avoids the problems of the original scheme.
To classify users according to the importance of their business with the information system and establish an evaluation model on service quality of information system based on classified users appraisal, the weights of relevant indexes are proposed using analytical hierarchy process(AHP) and expert scoring. Simulation shows that, under distribution denial of service (DDoS) attack, it can effectively reduce the loss of business of the information system and improve its service quality evaluation to ensure advanced users business firstly.
In order to improve communication quality and promote spectral efficiency, a combination of irregular low-density parity-check (LDPC) codes and high order modulation technique is analyzed. A simple and efficient combination scheme is proposed, in which, the effect of degree distribution of irregular LDPC codes and the unequal error protection (UEP) property of high order modulation technique are considered. Different degree variable nodes have different protection levels during modulation mapping. It is proved that the system performance is promoted when using Gaussian approximation (GA) algorithm. Simulation shows that new scheme can reduce the system error rate.
Due to the dynamic coupling of space robot system, the evaluating method of manipulability used on ground can not be directly applied to the space robot. By moving frame method in differential geometry, a volume element concept is developed to evaluate the manipulability and disturbance of a two-degree space robot system. Based on volume element model, the manipulability of manipulator and its influence on the base are analyzed as well. Their effectiveness and accuracy are verified with help of virtual manipulator approach.
Specifically for challenge to the power dissipation and area during high-level synthesis in electronic design automation（EDA）, An imprecise arithmetic automation synthesis and optimization approach is proposed, the word-length optimization are based on finite field characteristic. The imprecise polynomial circuit optimization is based on arithmetic transformation method, which considering the fixed point and floating point representation condition first, then a realization of circuits under different accuracy boundary is given, the power dissipation and area error is traded off, which applying branch and bound method to find the solution with minimal cost, to determine the minimal word length which satisfies the error bound, making use of the automation synthesis approach to get ultimate optimize structure at the same time. The experiment results demonstrate power dissipation and area got a 35% and 20% reduction respectively by increasing 05%～1% round off error.
The short-period problems of Chebyshev sequences had been researched, and aconclusion that Chebyshev sequences of random integer order will degenerateinto short-period sequences while Chebyshev sequences initial value is afinite decimal had been proved. The equivalent conditions about Chebyshevsequences degenerating into short-period sequences had been shown first, andthen an integer order Chebyshev sequences whose initial value is integermust degenerate into short-period sequence has been proved. Finally, theinitial value conditions has been extended to finite decimal by usingmathematical induction method.
The effect on the received signal to interference plus noise ratio (SINR) of spatial division multiple access (SDMA) systems with zero-forcing beamforming due to the channel estimation error and feedback delay is investigated. To reduce the negative effect of channel estimation error, a dynamic adaptive modulation scheme is proposed. This scheme dynamically adjusts the SINR thresholds for different modulation orders according to the feedbacks information from the receiver until the all the data are received correctly. Without increasing the signaling load and modifying original system parameters (e.g., modulation order and threshold), it can be easily extended to the cross-layer design existing channel estimation error.
an adaptive feedback algorithm for multi-user multi-input multi-output （MIMO） system based on hybrid automatic repeat request (HARQ) is proposed in this paper. In the proposed scheme, base station receives the acknowledgement/no acknowledgement (ACK/NACK) from all users in HARQ processes, analyses the channel state and pairing state of scheduled users, controls the feedback of channel state information （CSI） from users, schedules and precodes the users to enhance the system capacity. The proposed scheme has the advantage of low feedback as well as low complexity and outperforms the traditional multi-user MIMO system. Especially, when high SNR regime or in high-speed mobility, this scheme alleviates the ceiling effect of multi-user MIMO.
For the radio resource management (RRM) of single-cell multimedia broadcast/multicast service （MBMS） with feedback, a simple and effective mechanism is proposed, in which users with G-factor lower than a threshold are not taken into account in RRM, to trade considerable throughput gain a small loss in coverage (percentage of users receiving data with block error rate lower than 1%). This mechanism can be combined with frequency domain packet scheduling (FDPS) of MBMS, and its beneficial impact on the scheduling gain is analyzed. Simulation shows that the proposed mechanism and FDPS can boost each other, that will further improve throughput of MBMS. With moderate to larger number of users per cell, 50% throughput gain can be obtained with coverage loss lower than 1%.
The optimal power allocation strategy of cognitive orthogonal frequency division multiplexing (OFDM) system is studied. In order to limit primary user’s outage probability, the optimization is formulated maximizing the capacity of cognitive user under the constraint of outage probability at the primary receiver, and the optimal solution to the problem is obtained by utilizing the Lagrange method. Simulation shows that this scheme can efficiently limit primary user’s outage probability to its tolerable range while maximizing secondary user’s transmission rate.
A credit scoring model depending on customer application data and behavior data is developed in order to obtain the score of telecom customers’ initial credit. The model includes: the definition of the attribute domain and its weight; some rules on the weight; the formula to calculate initial credit; and the evaluating function. Meanwhile, an allocation algorithm for customers’ attribute weight is proposed using ants algorithm based on the model. Simulation demonstrates the rationality and feasibility of the model.
The double-serially concatenated convolutional codes (D-SCCC) method applied in underwater acoustic communication system was studied. In view of the acoustic communication’s requirement of high reliability, a new coding method was proposed to improve the reliability of communication through analyzing bit error rate (BER), and a simulation based on a given underwater channel has been made. Simulations show that at the same bit rate DSCCC increases the waterfall area’s steepness of BER compared with SCCC and has 0.5dB coding gain in the BER of 10-5, and D-SCCC improves error-floor of Turbo code’s BER when BER is between 10-4 and 10-5. This can satisfy requirement of high reliability when the underwater acoustic transmits remote measurement and remote-controlled commands.