This paper presents an investigation of statistically based approaches to the design of buffer control algorithms for interfacing a compressed digital video source to a constant rate channel. Since video compression t...
详细信息
This paper presents an investigation of statistically based approaches to the design of buffer control algorithms for interfacing a compressed digital video source to a constant rate channel. Since video compression techniques such as DPCM or transform coding with Huffman-type statistical encoding are characterized by a variable output bit-rate, rate buffering and adaptive encoding are used to maintain high average image quality, while avoiding buffer overflow events. As a first step to a quantitative methodology for the design and evaluation of adaptive buffer control algorithms, a detailed statistical characterization of the various encoding modes of an example "broadcast quality" intra/interframe DPCM algorithm is obtained from extensive simulation. The statistics presented include simple measures such as the overall encoded rate distributions for each encoding mode (which are useful for general mode selection), as well as more detailed intermode and intramode statistics (which provide useful information for mode switching). Two specific adaptive mode control algorithms are proposed. These are a) the expected "rate policy line" algorithm, driven by conditional intermode expectation statistics, and b) the "constant overflow probability" algorithm driven by conditional intramode distributions. Using simulation over a large set of representative images, the performance of the proposed statistically based algorithms is compared to that of a conventional buffer-level-based control heuristic which does not require source characterization. It is demonstrated that the proposed buffer control methods potentially provide substantial performance improvement over unaided heuristics, without significantly increasing implementation complexity.
The feedback lattice filter forms, including the two-multiplier form and the normalized form, are examined with respect to their relationships to the feedback direct form filter. Specifically, the transformation matri...
详细信息
The feedback lattice filter forms, including the two-multiplier form and the normalized form, are examined with respect to their relationships to the feedback direct form filter. Specifically, the transformation matrix between the lattice forms and the direct form is dervied;parameter and state relationships between the lattice forms and the direct form are therefore obtained. An IIR filter structure-the cascade lattice IIR structure-is constructed. Based on this structure, three IIR adaptive filtering algorithms in the two-multiplier form can then be developed following the gradient approach, the Steiglitz-McBride approach and the hyperstability approach. Convergence of these algorithms is theoretically analyzed using either the ODE approach or the hyperstability theorem. These algorithms will then be simplified into forms computationally as efficient as their corresponding direct form algorithms. Relationships of the simplified algorithms to the direct form algorithms are also studied, which disclose a consistency in algorithm structure regardless of the filter form. Three normalized lattice algorithms can also be derived from the two-multiplier lattice algorithms. Experimental results show much improved performance of the normalized lattice algorithms over the two-multiplier lattice algorithms and the direct form algorithms.
We propose the dual phase consensus algorithm (DCPA) to solve distributed sensor-target allocation for multitarget tracking. DPCA combines the benefits of greedy and consensus algorithms to converge to a feasible solu...
详细信息
We propose the dual phase consensus algorithm (DCPA) to solve distributed sensor-target allocation for multitarget tracking. DPCA combines the benefits of greedy and consensus algorithms to converge to a feasible solution, and then iteratively improves the allocation to aproach the global optimal. Theoretical analysis for convergence and computational and communication complexity of the algorithm is included. This analysis is validated using representative simulation results including sensitivity studies for increasing the number of targets and varying network topologies.
This paper presents the development of several Efficient LEarning algorithms for Neural NEtworks (ELEANNE). The ELEANNE 1 and ELEANNE 2 are two recursive least-squares learning algorithms, proposed for training single...
详细信息
This paper presents the development of several Efficient LEarning algorithms for Neural NEtworks (ELEANNE). The ELEANNE 1 and ELEANNE 2 are two recursive least-squares learning algorithms, proposed for training single-layered neural networks with analog output. This paper also proposes a new optimization strategy for training single-layered neural networks, which provides the basis for the development of a variety of efficient learning algorithms. This optimization strategy is the source of the ELEANNE 3, a second-order learning algorithm for training single-layered neural networks with binary output. A simplified version of this algorithm, called ELEANNE 4, is also derived on the basis of some simplifying but reasonable assumptions. The two algorithms developed for single-layered neural networks provide the basis for the derivation of ELEANNE 5 and ELEANNE 6, which are proposed for training multilayered neural networks with binary output. The ELEANNE 7 is an efficient algorithm developed for training multilayered neural networks with either binary or analog output. The proposed algorithms are experimentally tested and compared with algorithms already existing in the literature.
A scan-line z-buffer algorithm that takes advantage of the architecture of a vector computer for improved performance is presented. The changes that need to be considered when designing and implementing a vectorized s...
详细信息
A scan-line z-buffer algorithm that takes advantage of the architecture of a vector computer for improved performance is presented. The changes that need to be considered when designing and implementing a vectorized scan-line rendering algorithm are described. The resulting algorithm has about a three to one speed advantage over its scalar version. This speedup was achieved without sacrificing the efficiency of the scalar version. The performance of the vectorized algorithm is compared to that of three other rendering algorithms.
Subspace based direction-of-arrival (DOA) estimation has attracted many excellent performance studies, but limitations such as the assumption of an infinite amount of data and analysis of individual algorithms general...
详细信息
Subspace based direction-of-arrival (DOA) estimation has attracted many excellent performance studies, but limitations such as the assumption of an infinite amount of data and analysis of individual algorithms generally exist in these performance studies. We have previously proposed a unified performance analysis based on a finite amount of data, and achieved a tractable expression for the mean-squared DOA estimation error for the multiple signal classification (MUSIC), Min-Norm, estimation of signal parameters via rotational invariance techniques (ESPRIT), and State-Space Realization (SSR) algorithms. However, this expression uses the singular values and vectors of a data matrix which arc obtained by the highly nonlinear transformation of the singular value decompisition (SVD). Thus the effects of the original data parameters such as numbers of sensors and snapshots, source coherence and separations were not explicitly analyzed. Here we have made significant further unification and simplification of our previous result, and derived a unified expression based on the original data parameters. We then analytically observe the effects of these parameters on the estimation error. In addition, some interesting phenomena are discovered such as the fact that not all the algorithms have the property that additional sensors give better performance.
This paper investigates the statistical behavior of the finite precision LMS adaptive filter in the identification of an unknown time-varying stochastic system, Nonlinear recursions are derived for the mean and mean-s...
详细信息
This paper investigates the statistical behavior of the finite precision LMS adaptive filter in the identification of an unknown time-varying stochastic system, Nonlinear recursions are derived for the mean and mean-square behaviors of the adaptive weights, Transient and tracking algorithm performance curves are generated from the recursions and shown to be in excellent agreement with Monte Carlo simulations, Our results demonstrate that linear models are inappropriate for analyzing the transient and the steady-state algorithm behavior. The performance curves indicate that the transient and tracking capabilities cannot be determined from perturbations about the infinite precision case, It is shown that the transient phase of the algorithm increases as the digital wordlength or the speed of variation of the unknown system decrease, design examples illustrate how the theory can be used to select the algorithm step size and the number of bits in the quantizer.
The paper first summarizes a general approach to the training of recurrent neural networks by gradient-based algorithms, which leads to the introduction of four families of training algorithms. Because of the variety ...
详细信息
The paper first summarizes a general approach to the training of recurrent neural networks by gradient-based algorithms, which leads to the introduction of four families of training algorithms. Because of the variety of possibilities thus available to the ''neural network designer,'' the choice of the appropriate algorithm to solve a given problem becomes critical. We show that, in the case of process modeling, this choice depends on how noise interferes with the process to be modeled;this is evidenced by three examples of modeling of dynamical processes, where the detrimental effect of inappropriate training algorithms on the prediction error made by the network is clearly demonstrated.
A new constant false alarm rate (CFAR) detection algorithm operating in nonhomogeneous clutter is proposed. The proposed CFAR algorithm, cell-under-test (CUT) inclusive (CI) CFAR, utilizes a goodness-of-fit test for d...
详细信息
A new constant false alarm rate (CFAR) detection algorithm operating in nonhomogeneous clutter is proposed. The proposed CFAR algorithm, cell-under-test (CUT) inclusive (CI) CFAR, utilizes a goodness-of-fit test for determining and combining homogeneous windows, which results in a higher detection performance. The CI-CFAR algorithm layout and analytical properties for the ideal case are investigated. In comparison with the traditional cell-averaging (CA) CFAR and order-statistic (OS) CFAR, it is demonstrated that CI-CFAR outperforms both in most situations.
暂无评论