An unsupervised algorithm is proposed for learning overcomplete topographic representations of nature image. Our method is based on Independent Component Analysis (ICA) model due to its superiority on feature extraction, and overcomes the weakness of traditional method in fast overcomplete learning. Besides, the learnt topographic representation, resembling receptive fields of complex cells, can be used as descriptors to extract invariant features. Recognition experiments on Caltech-101 dataset confirm that these complex cell descriptors are not only efficient in feature extraction but achieve comparable performances to traditional descriptors.

The present paper gives a new formulation for rough surface scattering in terms of a stochastic integral equation which can be dealt with by means of stochastic functional approach. The random surface is assumed to be infinite and a homogeneous Gaussian random process. The random wave field is represented in the stochastic Floquet form due to the homogeneity of the surface, and in the non-Rayleigh form consisting of both upward and downward going scattered waves, as well as in the extended Voronovich form based on the consideration of the level-shift invariance. The stochastic integral equations of the first and the second kind are derived for the unknown surface source function which is a functional of the derivative or the increment of the surface profile function. It is also shown that the inhomogeneous term of the stochastic integral equation of the second kind automatically gives the solution of the Kirchhoff approximation for infinite surface.

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.

Execution performance is critical for large-scale and data-intensive workflows. This paper proposes DISWOP, a novel scheduling algorithm for data-intensive workflow optimizations; it consists of three main steps: workflow process generation, task & resource mapping, and task clustering. To evaluate the effectiveness and efficiency of DISWOP, a comparison evaluation of different workflows is conducted a prototype workflow platform. The results show that DISWOP can speed up execution performance by about 1.6-2.3 times depending on the task scale.

For the problem of Indoor Home Scene Classification, this paper proposes the BOW Model of Local Feature Information Gain. The experimental results show that not only the performance is improved but also the computation is reduced. Consequently this method out performs the state-of-the-art approach.

Reinforcement learning has been used to adaptive service composition. However, traditional algorithms are not suitable for large-scale service composition. Based on Q-Learning algorithm, a multi-task oriented algorithm named multi-Q learning is proposed to realize subtask-assistance strategy for large-scale and adaptive service composition. Differ from previous studies that focus on one task, we take the relationship between multiple service composition tasks into account. We decompose complex service composition task into multiple subtasks according to the graph theory. Different tasks with the same subtasks can assist each other to improve their learning speed. The results of experiments show that our algorithm could obtain faster learning speed obviously than traditional Q-learning algorithm. Compared with multi-agent Q-learning, our algorithm also has faster convergence speed. Moreover, for all involved service composition tasks that have the same subtasks between each other, our algorithm can improve their speed of learning optimal policy simultaneously in real-time.

In this article, we present a scattered pilot aided channel estimation and tracking technique for MIMO-OFDM systems. First we extend Li's estimation algorithm to scattered pilot-aided case. Then a scattered pilot-aided tracking scheme is developed to track the time-varying MIMO channel. The algorithms we present reduce the system overhead and channel estimation complexity to a low level with acceptable performance degradation. Simulations of a MIMO-OFDM system with four-transmit and four-receive antennas show that the new algorithm has better performance than Li's algorithm in time-varying MIMO channels.

Carrier frequency and symbol timing errors may greatly degrade the performance of the orthogonal frequency division mulitplexing (OFDM) system, especially in multipath environment. In this paper, we explore the cyclostationarity of OFDM signals, which only relies on second order statistics, to estimate the synchronization offset. First, a coarse carrier frequency offset estimator for multipath environment is developed using the second order statistics of the received OFDM signal. It has a wide capture range though not accurate. Second, we introduce a new synchronization algorithm based on cyclostationarity and matched filter theories, which can get the maximal estimation SNR in multipath environment. Both estimators utilize channel state information to achieve better estimation performance and are non-pilot aided. They can be combined to form a whole OFDM synchronizer for multipath environment. Finally, simulations confirm the performance of the estimation algorithm.

Shared-Memory Optical Packet (SMOP) switch architecture is very promising for significantly reducing the amount of required optical memory, which is typically constructed from fiber delay lines (FDLs). The current reservation-based scheduling algorithms for SMOP switches can effectively utilize the FDLs and achieve a low packet loss rate by simply reserving the departure time for each arrival packet. It is notable, however, that such a simple scheduling scheme may introduce a significant packet out of order problem. In this paper, we first identify the two main sources of packet out of order problem in the current reservation-based SMOP switches. We then show that by introducing a "last-timestamp" variable and modifying the corresponding FDLs arrangement as well as the scheduling process in the current reservation-based SMOP switches, it is possible to keep packets in-sequence while still maintaining a similar delay and packet loss performance as the previous design. Finally, we further extend our work to support the variable-length burst switching.

We focus on non-conflicting construction of an optical multistage feedforward network to emulate the N-to-1 output buffer multiplexer by using switched fiber delay line (SDL). In [1], Y.T. Chen et al. presented a sufficient condition (an upper bound) for the number of delay lines required for such a multiplexer with variable length bursts. In this paper, we first give an improved upper bound. Then we develop a framework to construct an arrival case of bursts which can be used to achieve a necessary condition (a lower bound). These results are further extended to the feedforward construction of N-to-N output buffer switch. Through simulation and performance comparison, we find that the new bounds can significantly decrease the hardware cost for constructing both the feedforward SDL-based multiplexer and output buffer switch while still provide the same performance as that of the old ones.

In this paper, we present a new all-digital carrier recovery loop for high-order quadrature amplitude modulation (QAM) signal constellations. The proposed approach is a blind phase-frequency detector structure that consists of a phase detector, a phase offset estimator, a frequency offset estimator, and a digital control oscillator. Compared to previous related approaches, the proposed algorithm provides a wider acquisition range and a more accurate estimation of frequency and phase offsets. These features are demonstrated by simulation results of the DOCSIS (Data-Over-Cable Service Interface Specifications) cable modem system.

One major drawback of orthogonal frequency-division multiplexing is the high peak-to-average power ratio (PAPR) of the output signal. The selected mapping (SLM) and partial transmit sequences (PTS) methods are two promising techniques for PAPR reduction. However, to generate a set of candidate signals, these techniques need a bank of inverse fast Fourier transforms (IFFT's) and thus require high computational complexity. In this paper, we propose two low-complexity multiplication-free conversion processes to replace the IFFT's in the SLM method, where each conversion process for an N-point IFFT involves only 3N complex additions. Using these proposed conversions, we develop several new SLM schemes and a combined SLM & PTS method, in which at least half of the IFFT blocks are reduced. Computer simulation results show that, compared to the conventional methods, these new schemes have approximately the same PAPR reduction performance under the same number of candidate signals for transmission selection.

In this letter, we learned overcomplete filters to model rich priors of nature images. Our approach extends the Gaussian Scale Mixture Fields of Experts (GSM FOE), which is a fast approximate model based on Fields of Experts (FOE). In these previous image prior model, the overcomplete case is not considered because of the heavy computation. We introduce the assumption of quasi-orthogonality to the GSM FOE, which allows us to learn overcomplete filters of nature images fast and efficiently. Simulations show these obtained overcomplete filters have properties similar with those of Fields of Experts', and denoising experiments also show the superiority of our model.

Power bus noise problem has become a major concern for both EMC engineers and board designers. A fast algorithm, based on the cavity-mode model, was employed for analyzing resonance characteristics of multilayer power bus stacks interconnected by vias. The via is modeled as an inductance and its value is given by a simple expression. Good agreement between the simulated results and measurements demonstrates the effectiveness of the cavity-mode model, together with the via model.

The researches on model-based testing mainly focus on the models with single component, such as FSM and EFSM. For the network protocols which have multiple components communicating with messages, CFSM is a widely accepted solution. But in some network protocols, parallel and data-shared components maybe exist in the same network entity. It is infeasible to precisely specify such protocol by existing models. In this paper we present a new model, Parallel Parameterized Extended Finite State Machine (PaP-EFSM). A protocol system can be modeled with a group of PaP-EFSMs. The PaP-EFSMs work in parallel and they can read external variables form each other. We present a 2-stage test generation approach for our new models. Firstly, we generate test sequences for internal variables of each machine. They may be non-executable due to external variables. Secondly, we process the external variables. We make the sequences for internal variables executable and generate more test sequences for external variables. For validation, we apply this method to the conformance testing of real-life protocols. The devices from different vendors are tested and implementation faults are exposed.