Modern video compression usually consists of ME/MC (Motion Estimation/Motion Compensation), transform, and quantization of the transform coefficients. Efficient bit allocation technique to distribute available bits to motion parameters and quantized coefficients is an important part of the whole system. A method that is very complex and/or needs buffering of many future frames is not suitable for real time application. We develop an efficient bit allocation technique that utilizes the estimated effect of allocated bits to motion parameter and quantization on the overall quality. We also propose an hierarchical block based ME/MC technique that requires less computations than classical BMA (Block Matching Algorithm) while offering better motion estimation.

We investigate traffic performance of CBWL schemes with multimedia services on non-homogeneous cellular network in which cut-off priority is given to handoff calls. Two generic routing schemes are analyzed: one is the randomized routing and the other is the least loaded routing. The performance measures that we focus on are the new call blocking probabilities and the handoff failure probabilities. To evaluate blocking probabilities of interest, we construct a generalized access network whose blocking probabilities are same as ones for CBWL systems. For analysis of generalized access network, we apply the reduced load approximation. The computational complexity and memory requirements of proposed algorithm are linear so that we can use this algorithm to approximate blocking probabilities of CBWL systems of large size. The proposed approximations are tested on a number of simple examples. Numerical results for 12 cells and 36 cells networks are given. The comparison between approximation and simulation results shows that the proposed approximation method is quite accurate.

In this paper, effectiveness and applicability of the models presented in the preceding paper are discussed. First, a collection of manually placed blocks employed for a 32-bit CPU design have been examined. Postulated linearities of the placement and partitioning are seen to be acceptable with reasonable statistical accuracy. One finding is that the extrapolation of the linear partitioning model gives a statistical 'first-order' value of the circuit area for custom chip design, typically done automatically on long one-dimensional rows of standard cells or gate arrays. Several statistical properties of terminals in a manually placed block are also presented.

Localization of mobile terminals has received considerable attention in wireless communications. In this letter, we present a covariance shaping least squares (CSLS) estimator using time-of-arrival measurements of the signal from the mobile station received at three or more base stations. It is shown that the CSLS estimator yields better performance than the other LS estimators at low signal-to-noise ratio conditions.

The future third generation mobile communications system, named IMT-2000, is expected to provide mobile users with voice, high-rate data and their combined multi-media services with the same QoS as in the fixed networks. As a radio access standard for the IMT-2000, W-CDMA and cdma2000 have been selected in Europe and North America, respectively. In this paper, we present an analytic model of the cdma2000 data mobile servicing a connected data service. In order to do this, we first model the traffic generated at mobile by a discrete-time Batch Markovian Arrival Process (D-BMAP). Next, we model the Radio Link Protocol (RLP) Queue in the cdma2000 MAC protocol by a D-BMAP/D/1 queueing system with batch service and setup times. Finally, we analyze this queueing system and get the performance measures such as the mean delay and the loss probability. Analytic results are compared with simulation ones for accuracy.

A Virtual Cellular Network (VCN) is a wireless cellular network wherein a single Mobile Station (MS) can communicate simultaneously with more than one Base Station (BS). In this paper, we analyze handoff for two kinds of VCN: a 'Distributed Resource-control VCN' (DR-VCN) and a 'Centralized Resource-control VCN' (CR-VCN). A VCN can take advantage of the fact that the same data is received by multiple base stations. The DR-VCN is a system in which every BS controls its own channels, while the CR-VCN is a system wherein a central station controls all system channels. Results from analysis and simulation show that both the new call drop rate and handoff refusal rate of the CR-VCN are much lower than those of the DR-VCN.

Cellular Digital Packet Data (CDPD) provides wireless data communication services to mobile users by sharing unused RF channels with AMPS on a non-interfering basis. To prevent interference on the voice activities, CDPD makes forced hop to a channel stream when a voice request is about to use the RF channel occupied by the channel stream. The number of forced hops is affected by the voice channel selection policy. We propose analytic models to investigate the CDPD channel holding time for the the least-idle and random voice channel selection policies. Under various system parameters and voice channel selection policies, we provide guidelines to reduce the number of forced hops.

This paper presents several statistical and empirical approaches to estimating circuit area required for laying out a given number of random-logic gates. The approaches are developed for MOS layout methods in which terminals for load and driver devices account for independent wiring tracks. For placements and interconnections, two statistical models involving (a) linear superposition of one-dimensional placements and (b) uniform, diffused connections of terminals on a wiring track, are introduced. Block width dependent wiring track requirements are determined by incorporating one-dimensional logic partitioning across the block. An empirical relationship for the spatial partitioning is established from a collection of placements done with a manual layout method.

For producing high quality synthesis, a concatenation-based Text-to-Speech (TTS) system usually requires a large number of segmental units to cover various acoustic-phonetic contexts. However, careful manual labeling and segmentation by human experts, which is still the most reliable way to prepare such units, is labor intensive. In this paper we adopt a two-step procedure to automate the labeling, segmentation and refinement process. In the first step, coarse segmentation of speech data is performed by aligning speech signals with the corresponding sequence of Hidden Markov Models (HMMs). Then in the second step, segment boundaries are refined with a proposed Context-Dependent Boundary Model (CDBM). Classification and Regression Tree (CART) is adopted to organize available data into a structured hierarchical tree, where acoustically similar boundaries are clustered together to train tied CDBM models for boundary refinement. Optimal CDBM parameters and training conditions are found through a series of experimental studies. Comparing with manual segmentation reference, segmentation accuracy (within a tolerance of 20 ms) is improved by the CDBMs from 78.1% (baseline) to 94.8% in Mandarin Chinese and from 81.4% to 92.7% in English, with about 1,000 manually segmented sentences used in training the models. To further reduce the amount of manual data for training CDBMs of a new speaker, we adapt a well-trained CDBM via efficient adaptation algorithms. With only 10-20 manually segmented sentences as adaptation data, the adapted CDBM achieves a segmentation accuracy of 90%.