This paper presents a new architecture of a high-speed ATM switch. The switch, called FSM (Factorial Style Memory) switch, uses Dual-Port memories to construct a factorial style memory for eliminating the bus contention problem. In order to fit the various applications, based on the proposed switch architecture, several kinds of models are also proposed to construct the larger size of switches. With the same required throughput and cell loss probability, the performance analysis of the switch shows that the number of buffers and average cell delay can be significantly reduced in the proposed switch while comparing to the ATM switches with central shared memory. For satisfying applications that require larger switches, three types of expansion methods which contain size expansion, memory expansion, and size-memory combined expansion are discussed.

In wireless communications, a smart antenna system utilizes an antenna array to acquire the spatial signatures of transmitted signals. This system uses the difference in the spatial signatures or the direction of arrival (DOA) of signals to correctly obtain the desired signal. This can reduce co-channel interference, mitigate the fading phenomenon caused by multipath transmissions, improve the communication quality and increase the system capacity. The purpose of this paper is to evaluate the performance of smart antennas using four beamforming algorithms applied to a wideband code division multiple access (W-CDMA) system. The simulation results show that, based on the same power consumption at the transceiver and using a Rake receiver, a W-CDMA system with a smart antenna can operate at a lower bit error rate at the specific signal to noise ratio (SNR). Moreover, the smart antenna system accommodates more users at the specific signal to interference ratio (SIR), even though a certain angle difference exists between the actual DOAs and the estimated DOAs.

In this study, a novel decoupling method using parasitic elements (PEs) connected by a bridge line (BL) for two planar inverted-F antennas (PIFAs) is proposed. The proposed method is developed from a well-known decoupling method that uses a BL to directly connect antenna elements. When antenna elements are connected directly by a BL, strong mutual coupling can be reduced, but the resonant frequency shifts to a different frequency. Hence, to shift the resonant frequency toward the desired frequency, the original size of the antenna elements must be adjusted. This is disadvantageous if the method is applied in cases where the design conditions render it difficult to connect the antennas directly or adjust the original antenna size. Therefore, to easily reduce mutual coupling in such a case, a decoupling method that does not require both connecting antennas directly and adjusting the original antenna size is necessitated. This study demonstrates that using PEs connected by a BL reduces the mutual coupling from -6.6 to -14.1dB, and that the resonant frequency is maintained at the desired frequency (2.0GHz) without having to adjust the original PIFAs size. In addition, impedance matching can be adjusted to the desired frequency, resulting in an improved total antenna efficiency from 77.4% to 94.6%. This method is expected to be a simple and effective approach for reducing the mutual coupling between larger numbers of PIFA elements in the future.

In this paper, we proposed a novel entropy-based image watermarking method in wavelet domain. Unlike traditional entropy, we use the normalized energy instead of the probability which is called energy-based entropy (EBE). Based on EBE, the watermark can be embedded robustly and imperceptibly. In our proposed method, the wavelet-trees are grouped into super-trees. Then each super-tree is also divided into five sub-blocks. According to the watermark bit state, the EBE of each sub-block will be modified respectively. In an experiment, three images (Lenna, Goldhill and Peppers) are chosen for evaluating the performance. The PSNR of these watermarked images are 44.039, 43.51 and 43.67. Compared with Wang et al.[18], it greatly increases the PSNR, by about 5.8, 4.8 and 3.9 dB respectively. For the consideration of the capacity for embedding, the maximum number of watermark bits is also increased. The experimental results show that the proposed entropy-based watermarking method performs well in JPEG compression, filtering (Gaussian filter, median filter and sharpen) and geometrical attacks (pixel shift and rotation). In addition, it is also very robust to against the multiple watermark attack.

This paper puts forward the concept of cellular network location with less information which can overcome the weaknesses of the cellular location technology in practical applications. After a systematic introduction of less-information location model, this paper presents a location algorithm based on AGA (Adaptive Genetic Algorithm) and an optimized RBF (Radical Basis Function) neural network. The virtues of this algorithm are that it has high location accuracy, reduces the location measurement parameters and effectively enhances the robustness. The simulation results show that under the condition of less information, the optimized location algorithm can effectively solve the fuzzy points in the location model and satisfy the FCC's (Federal Communications Commission) requirements on location accuracy.

In this work, we propose a scheme of routing table compaction for IP forwarding engines based on ternary content addressable memory (TCAM). Our scheme transforms the original routing table into a form with only disjoint prefixes. The most prevalent next hop of the routing table is then calculated and the route prefixes corresponding to the next hop are replaced by one TCAM entry. In combination with Espresso-II logic minimization algorithm, the proposed scheme reduces the TCAM storage requirements by more than 75% compared to the original routing tables. We also present an effective approach to support incremental updates.

In this work, a simple method for extracting MOSFET threshold voltage, effective channel length and channel mobility by using S-parameter measurement is presented. In the new method, the dependence between the channel conductivity and applied gate voltage of the MOSFET device is cleverly utilized to extract the threshold voltage, while biasing the drain node of the device at zero voltage during measurement. Moreover, the effective channel length and channel mobility can also be obtained with the same measurement. Furthermore, all the physical parameters can be extracted directly on the modeling devices without relying on specifically designed test devices. Most important of all, only one S-parameter measurement is required for each device under test (DUT), making the proposed extraction method promising for automatic measurement applications.

Traditional sparse representation-based methods for human action recognition usually pool over the entire video to form the final feature representation, neglecting any spatio-temporal information of features. To employ spatio-temporal information, we present a novel histogram representation obtained by statistics on temporal changes of sparse coding coefficients frame by frame in the spatial pyramids constructed from videos. The histograms are further fed into a support vector machine with a spatial pyramid matching kernel for final action classification. We validate our method on two benchmarks, KTH and UCF Sports, and experiment results show the effectiveness of our method in human action recognition.

Current-voltage characteristics of Cr-doped hydrogenated amorphous silicon-V (Cr/p+a-Si:H/V) analogue memory switching devices have been measured over a wide range of device resistance from several kilo-ohms to several hundred kilo-ohms, and over a temperature range from 13 K to 300 K. Both the bias and temperature dependence of the conductance show similar characteristics to that of metal-insulator heterogeneous materials (i.e. discontinuous or granular metallic films), which are analysed in terms of activated tunnelling mechanism. A modified filamentary structure for the Cr/p+a-Si:H/V switching devices is proposed. The influence of embedded metallic particles on memory switching is analysed and discussed.

In recent years, the increasing complexity of deep network structures has hindered their application in small resource constrained hardware. Therefore, we urgently need to compress and accelerate deep network models. Channel pruning is an effective method to compress deep neural networks. However, most existing channel pruning methods are prone to falling into local optima. In this paper, we propose a channel pruning method via Improved Grey Wolf Optimizer Pruner which called IGWO-Pruner to prune redundant channels of convolutional neural networks. It identifies pruning ratio of each layer by using Improved Grey Wolf algorithm, and then fine-tuning the new pruned network model. In experimental section, we evaluate the proposed method in CIFAR datasets and ILSVRC-2012 with several classical networks, including VGGNet, GoogLeNet and ResNet-18/34/56/152, and experimental results demonstrate the proposed method is able to prune a large number of redundant channels and parameters with rare performance loss.

In this work, fluctuation patterns of ReRAM current are classified automatically by proposed fluctuation pattern classifier (FPC). FPC is trained with artificially created dataset to overcome the difficulties of measured current signals, including the annotation cost and imbalanced data amount. Using FPC, fluctuation occurrence under different write conditions is analyzed for both HRS and LRS current. Based on the measurement and classification results, physical models of fluctuations are established.