摘要：Hardware/software partitioning is an essential step in hardware/software *** large size problems,it is difficult to consider both solution quality and *** paper presents an efficient GPU-based parallel tabu search algorithm(GPTS)for HW/SW partitioning.A single GPU kernel of compacting neighborhood is proposed to reduce the amount of GPU global memory accesses theoretically.A kernel fusion strategy is further proposed to reduce the amount of GPU global memory accesses of *** further minimize the transfer overhead of GPTS between CPU and GPU,an optimized transfer strategy for GPU-based tabu evaluation is proposed,which considers that all the candidates do not satisfy the given *** show that GPTS outperforms state-of-the-art work of tabu search and is competitive with other methods for HW/SW *** proposed parallelization is significant when considering the ordinary GPU platform.

摘要：There is a sharp decline in the network performance when the wireless link fails as a data path in the Wireless Network-on-Chip(WiNoC).To counteract this problem,we propose a fault-tolerance mechanism for the efficient retransmission of data in the *** an error is detected in the data transmission process,this mechanism works to feed back the fault information to the source node in real time via fault signal *** the source node,the highest transmission priority is assigned to the backup retransmitted data,and the corresponding direct link is positioned to enable the data packet for its efficient retransmission to the destination node,thereby ensuring efficiency in fault ***,we have improved the receiving port of the wireless router,added the corresponding redundant buffers and mux,and dynamically selected the retransmitted non-faulty data packets to be written to the local router in order to avoid the disorderly retransmission of the data *** evaluation results of this paper demonstrate that compared with the methods which are under different fault conditions,this fault-tolerant method drastically improves the data throughput rate,reduces the delay,effectively guarantees the reliability of the network,and improves the system performance.

摘要：The dispersion characteristics and transverse field distribution of the fundamental and higher order modes are analyzed for polymeric ridge multimode waveguide by a new technique which is based on the combination of the effective index method and the variational method. An algorithm is implemented to study the effect of the structure parameters and dimensions on the dispersion curves. The optical field distribution of the fundamental and higher order modes for TM modes are computed. The single mode conditions of polymeric ridge waveguide are obtained. The relationship between the curvature radius and the bending loss of S-shaped ridge waveguide are studied with wide-angle finite-difference beam propagation method and effective index method. The conclusion is: when the curvature radius is larger than 5000 μm, the bending loss will not decrease distinctly even if the curvature radius increases, and the light can propagate stably in the S-shaped ridge waveguide.

摘要：A reconfigurable cipher chip for accelerating DES is described, 3DES and AES computations that demand high performance and flexibility to accommodate large numbers of secure connections with heterogeneous clients. To obtain high throughput, we analyze the feasibility of high-speed reconfigurable design and find the key parameters affecting throughput. Then, the corresponding design, which includes the reconfignration analysis of algorithms, the design of reconfignrable processing units and a new reconfignrable architecture based on pipeline and parallel structure, are proposed. The implementation results show that the opcrating fiequency is 110 MHz and the throughput rate is 7 Gbps for DES, 2.3 Gbps for 3 DES and 1.4 Gbps for AES. Compared with the similar existing implementations, our design can achieve a higher performance.

摘要：The internal flow field of an axial-flow pump under low flow rate conditions is extremely turbulent. The unstable flow region is formed due to the tip leakage and the rotating stall, and is distinguished by a saddle patterned zone in its flow-head curve that demonstrates restrictions in its workable flow range. It is therefore important to understand and improve the operational characteristics of an axial-flow pump under low flow rate conditions. In this study, experiments are performed for the performances of an axial-flow pump at the flow rate in a range between 0.8Qd and 1.2Qd, with the internal flow field measured by the particle image velocimetry (PIV), and with the pump performances and the internal flow field simulated by the Ansys CFX. The simulation results agree well with the experimental results. From the predicted heads at the flow rate in the range between 0.1Qd and 0.7Qd by the numerical simulation, the complete flow-head curves are obtained. The streamlines and the velocity contours in the region in front of the impeller leading edge under different flow conditions are analyzed. By adopting the double-inlet structures, the flow-head characteristics are studied by varying the values of δ and σ respectively, where δ denotes the gap between the inner cylinder end and the impeller leading edge, and σ denotes the gap between the inner cylinder and the outer cylinder. The findings indicate that with the double-inlet design, the velocity distribution can be effectively improved in the region in front of the impeller leading edge, as well as the head performance under the low flow rate conditions, and the positive slope value of the head curve is reduced in the unstable flow range. The optimal δ and σ values are determined.

摘要：Passive filtering of neural networks with time-invariant delay and quantized output is considered.A criterion on the passivity of a filtering error system is proposed by means of the Lyapunov-Krasovskii functional and the Bessel-Legendre *** on the criterion,a design approach for desired passive filters is developed in terms of the feasible solution of a set of linear matrix ***,analyses and syntheses are extended to the time-variant delay situation using the reciprocally convex combination ***,a numerical example with simulations is used to illustrate the applicability and reduced conservatism of the present passive filter design approaches.

摘要：In this paper, we present a design method based on the concept of equivalent input disturbance （EID） to reject disturbances for a linear time-invariant system. A generalized state observer （GSO） is used to estimate an EID of the external disturbances, and the pole-assignment algorithm is employed to select the matrices of the GSO. Simulation and experimental results of a rotational speed control system demonstrate the validity of our method.

摘要：The method of condition number is commonly used to diagnose a normal matrix N whether it is ill conditioned state or not. For its shortcoming, a method to measure multi collinearity of a matrix was put forward. The method is that implement Gram Schmidt orthogonalizing process to column vectors of a design matrix A (α l ), then calculate the norms of every vector before and after orthogonalization process and their corresponding ratio, and use the minimum ratio among the group of ratios to measure the multi collinearity of A. According to the corresponding relationship between the multi collinearity and the ill conditioned state of a matrix, the method also studies and offers reference indexes weighing the ill conditioned state of a matrix based on the relative norm. The remarkable characteristics of the method are that the measure of multi collinearity has idiographic geometry meaning and clear lower and upper limit, the size of the measure reflects the multi collinearity of column vectors objectively. It is convenient to study the reason that results in the matrix being multi collinearity and to put forward solving plan according to the method which is summarized as the method of minimum norm and abbreviated as F method.

摘要：QLC（Quad-Level Cell） NAND flash will be one of the future technologies for next generation memory chip after three-dimensional（3D） TLC（Triple-Level Cell） stacked NAND flash. In QLC device, data errors will easily occur because of 2~4 data levels in the limited voltage range. This paper studies QLC NAND technology which is 4 bits per cell. QLC programming methods based on 16 voltage levels and reading method based on ＂half-change＂ Gray coding are researched. Because of the probable error impact of QLC NAND cell＇s voltage change, the solution of generating the soft information after XOR（exclusive OR） the soft bits by internal read mechanism is presented for Low-Density Parity-Check（LDPC） Belief Propagation（BP） decoding in QLC design for its system level application.

摘要：A wideband dual-feedback low noise amplifier （LNA） was analyzed, designed and implemented using SiGe heterojunction bipolar transistor （HBT） technology. The design analysis in terms of gain, input and output matching, noise and poles for the amplifier was presented in detail. The area of the complete chip die, including bonding pads and seal ring, was 655 μm × 495 μm. The on-wafer measurements on the fabricated wideband LNA sample demonstrated good performance： a small-signal power gain of 33 dB with 3-dB bandwidth at 3.3 GHz was achieved; the input and output return losses were better than - 10 dB from 100 MHz to 4 GHz and to 6 GHz, respectively; the noise figure was lower than 4.25 dB from 100 MHz to 6 GHz; with a 5 V supply, the values of OPtdB and OIP3 were 1.7 dBm and 11 dBm at 3-dB bandwidth, respectively.