HKUST Institutional Repository
(5.016 recursos)
Repository of Hong Kong University of Science and Technology. Managed by the HKUST Library.
Mostrando recursos 1 - 20 de 216
1.
Adaptive filtering for collision detection - Shi, Bertram
This paper presents a theoretical analysis of the convergence properties of a parameter estimation based cellular neural network algorithm for estimating the time-to-contact from a one dimensional image. The algorithm uses a model of the image motion e based upon the assumption that the camera is approaching a planar object oriented parallel d to the image plane, which also implies that the time to contact is identical for every point in nce the image. Using Gabor-type spatial filtering and adaptive temporal filters, the algorithm 1 adjusts the parameters of the model to match the image velocity variation across the image....
2.
Dealing with multilinguality in a spoken language query translator - Fung, Pascale N.; Shi, Bertram; Wu, Dekai; Lam, Wai Bun; Wong, Shuen Kong
In this paper, we examine three issues concerning the robustness of multilingual speech interfaces for spoken language translation systems: accent differences, mixed language input, and the use of common feature sets for HMM-based speech recognizers for English and Cantonese. From the results of our preliminary experiments, we find that accent difference causes recognizers performance to degrade. For mixed language input, we found out that a straight forward implementation of a mixed language model-based speech recognizer performs less well than the concatenation of pure language recognizers due to the increase in recognition candidate numbers. Finally, our experimental results show that the...
3.
Second order CNN arrays for estimation of time-to-contact - Shi, Bertram
This paper describes a cellular neural network (CNN) for estimating the time-to-contact from a one dimensional image. The CNN arrays used for this algorithm consist of cells with second order dynamics. The key feature of these arrays is that the spatial information in a region around each cell is represented by the phase of a complex number. The velocity is encoded as the temporal variation of that phase. By modelling this variation using adaptive temporal oscillators, the velocity can be estimated. Velocity information extracted over the entire array can be combined to estimate time-to-contact.
4.
Performance analysis of least load multicast routing for single rate loss networks - Cheung, Chi-Chung; Tsang, Danny H. K.; Gupta, Sanjay; Chu, Hon-Wai
In this paper, we investigate a state dependent multicast routing algorithm called Least Load Multicast Routing (LLMR), for single rate loss networks. The algorithm is based on Least Load Routing (LLR) concept and the approach is to select the least load links for establishing connections. The networks considered are assumed fully connected. In addition, connection requests are Poisson arrival and the holding times of accepted calls are exponentially distributed. The analytical model that we developed for calculating blocking probabilities is based on the link independence assumption and the Reduced Load Approximation (RLA). Analytical results are compared with simulation results and...
5.
Modified fair queueing for finite buffer in ATM networks - Lai, Arthur M. O.; Tsang, Danny H. K.
Service scheduling scheme that controls the order of servicing cells within an ATM node is very important in providing guaranteed services. Much attention of recent research have been paid on emulating the General Processor Sharing (GPS) system as closely as possible with low computational complexity. The primary motivation of emulating the GPS system is to provide traffic isolation and thus to achieve a maximum delay bound. This delay bound is guaranteed by guaranteeing the minimum bandwidth. However, after guaranteeing the minimum bandwidth, some excess bandwidth might be leftover and we argue that it should be used more intelligently to improve...
6.
Hierarchical scheduling with adaptive weights for W-ATM - Huang, Hui; Tsang, Danny H. K.; Sigle, Rolf; Kuehn, Paul J.
Medium access control (MAC) protocol is one of the key components for providing quality of service (QoS) in wireless ATM (W-ATM) networks. In this paper, we propose a hierarchical scheduling scheme coupled with fair queueing algorithms with adaptive weights. This scheme is intended to be applicable to a TDMA/TDD based MAC protocol. Specifically, the performance of the fair-queueing algorithm using fixed weights and adaptive weights are investigated. Simulation results show that the proposed hierarchical fair-queueing scheduling with adaptive weights (HAW) can yield a lower cell transfer delay and higher channel utilization while maintaining fairness among multiple users.
7.
New state dependent multicast routing for single rate loss networks - Cheung, Chi-Chung; Tsang, Danny H. K.; Gupta, Sanjay
In this paper, we investigate the dynamic multicast routing problem for single rate loss networks and briefly discuss the dynamic multicast routing algorithm called Least Load Multicast Routing (LLMR). We propose a new multicast routing algorithm called Maximum Mean Number of New Calls Accepted Before Blocking Multicast Routing (MCBMR), which can more accurately reflect the current and future loading of a network. Simulation results show that this algorithm, compared with LLMR, not only has a smaller network revenue loss, but also results in smaller call blocking probabilities for all classes of traffic.
8.
Quorumcast routing by multispace search - Du, Bin; Gu, Jun; Tsang, Danny H. K.; Wang, Wei
In this paper, we present a multispace search algorithm, MUSQ, for the quorumcast routing problem. We first prune the candidate set to reduce the original search space. Some of the unqualified links for the opitmal tree are removed from the network. Then a multispace search algorithm is used to solve the quorumcast routing problem. By altering the original objective function, the original problem instance is transformed into a series of gradually more simplified problem instances with smoother terrain surfaces. A local search algorithm is used to solve each problem instance, from the simplest structure to the original structure, and the...
9.
Cell scheduling and bandwidth allocation for heterogeneous VBR video conferencing traffic - Yang, Tao; Tsang, Danny H. K.; McCabe, Patrick
We first propose a new cell loss scheduling scheme, referred to as the QoS-scheme. This scheme can satisfy different individual quality of service (QoS) requirements of a group of virtual channel connections with minimum required bandwidth. We next derive a simple formula based on Gaussian approximation to allocate bandwidth to heterogeneous VBR video conferencing connections. Comparing the proposed QoS-scheme with the simple last-in-first-out (LIFO) scheme, we find that if the LIFO scheme is adopted as the loss scheduling scheme, developing any more efficient bandwidth allocation algorithms can at most improve bandwidth utilization by about 10%. However, if the proposed QoS-scheme...
10.
Dynamic routing algorithms in VP-based ATM networks - Chu, Hon-Wai; Tsang, Danny H. K.
We consider a VP-based ATM network supporting multiple traffic classes with different traffic characteristics but with identical end-to-end quality of service (QoS) requirement. The concept of effective bandwidth is used to determine the required bandwidth to guarantee the QoS requirement of each traffic class. We study the problem of using dynamic routing to VP-based ATM networks by transforming it into an equivalent multi-rate circuit-switched network problem. To further simplify the analysis, we restrict the choice of path to single-link and two-link routes. We propose a dynamic routing algorithm based on the Least Loaded Routing (LLR) with packing. Simulation results are...
11.
Impacts of handoff on TCP performance in mobile wireless computing - Chan, Aldar Chun-fai; Tsang, Danny H. K.; Gupta, Sanjay
The effects of intermittent disconnections due to host motions on the performance of TCP connections are investigated. Fading and handoff due to host motion cause increased delay and packet losses to the active transport layer connection. TCP interprets these as signs of network congestion. As a result, it promptly throttles its transmissions and backoffs its timers, leading to slow post-handoff recovery of the transmission and long idle time. These cause severe end-to-end throughput degradation and unreasonably long interactive delay for human interaction. In this paper, we present three phenomena observed (long communication pause, slow post-handoff recovery, and successive timeouts) which...
12.
Generalized weighted fairness criterion : formulation and application on prioritized ABR service - Lee, Clarence Soul Chun; Cheung, K. F.; Tsang, Danny H. K.
In this paper, a generalized fairness criterion referred to as the Generalized Weighted Fairness Criterion (GWFC) for flow control on ABR service is presented. Within the GWFC framework, a weight is assigned to each ABR connection and bandwidth is allocated to it in proportion to the corresponding weight. The GWFC can generalize fairness sub-criteria for prioritized services, and in particular, subsume the Max-Min criterion as a uni-weight fairness criterion. Two weighted fairness sub-criteria are introduced and their performance on bandwidth allocation will be presented and compared with that of the Max-Min criterion.
13.
Credit-based fair queueing (CBFQ) : a simple and feasible scheduling algorithm for packet networks - Bensaou, Brahim; Chan, K. T.; Tsang, Danny H. K.
A simple rate-based scheduling algorithm for packet-switched networks (e.g. Internet, ATM) is proposed. By using simple counters to trace the amount of credits accumulated by each traffic stream, the algorithm decides which stream to serve next based on the bandwidth share allocated to each traffic stream and the sizes of their respective head of line (HOL) packets. Our proposed CBFQ algorithm achieves the same fairness and delay bounds as the alternative approaches based on virtual time, while avoiding the implementation problems inherent in these methods. In addition, to take advantage of the fixed packet size in ATM networks, a variant...
14.
CBR transportation of VBR MPEG-2 video traffic for video-on-demand in ATM networks - Ni, Jian; Yang, Tao; Tsang, Danny H. K.
In this paper, a CBR transport scheme for VBR MPEG-2 video-on-demand traffic in ATM networks is proposed. The basic idea is to smooth the VBR traffic to match a given bandwidth. In order to keep the continuous playback of the video, our scheme requires a suitable amount of buffer space at the receiver side and introduces an initial playback delay. We study these two performance measures as functions of allocated bandwidth. For given buffer size and bandwidth, it is sometimes necessary to control the transmission in order to satisfy the required QoS. Two control methods are examined and the corresponding...
15.
Least-loaded-first dynamic multicast routing for single rate loss networks - Cheung, Chi-Chung; Tsang, Danny H. K.; Gupta, Sanjay; Chu, Hon-Wai
In this paper, a new dynamic multicast routing algorithm for single rate loss networks is proposed. The algorithm is called Least-Loaded-First Multicast Routing (LLFMR) and is based on Least Loaded Routing (LLR) concept. The basic idea is to select the least loaded links for establishing the connection. Simulation results show that the normalized revenue loss is significantly reduced when compared to the dynamic multicast routing algorithms proposed in [3]. The results also show that the connection blocking probabilities of calls of different number of destination nodes decrease as the number of destination nodes in a connection request increases.
17.
Comparison of static & dynamic bandwidth allocation schemes for multiple QOS classes in ATM networks - Chan, Jimmy Hung Shing; Tsang, Danny H. K.
In ATM networks, the bandwidth of virtual paths (VPs) are allocated individually for the support of different traffic classes with different quality of service (QOS) requirements within a VP. A VP combination scheme defined in [8] is a way to assign traffic classes to different VPs for satisfying all the QOS requirements. It is also referred to as static bandwidth allocation schemes which may not have good utilization of network resources because of no sharing of bandwidth among VPs. Another approach that makes use of some priority handling mechanism and allows VP bandwidth to be shared among the VPs can...
18.
Bandwidth allocation of multiple QOS classes in ATM environment - Chan, Jimmy Hung Shing; Tsang, Danny H. K.
For future Broadband-ISDN, Asynchronous Transfer Mode (ATM) is designed not only to support a wide range of traffic classes with diverse flow characteristics (e.g., burstiness, bit rate and burst length), but to guarantee the different quality of service (QOS) requirements as well. The QOS may be measured in terms of cell loss probability and maximum cell delay. In this paper, we consider the ATM networks in which the virtual path (VP) concept is implemented. By applying the Markov Modulated Deterministic Process method, we develop an efficient algorithm to compute the minimum capacity required to satisfy all the QOS requirements when...
19.
Bandwidth scheduling of prerecorded VBR video sources for ATM networks - Lauderdale, John; Tsang, Danny H. K.
This paper presents network bandwidth scheduling algorithms for the transmission of pre-encoded MPEG VBR video over ATM networks. A system model is developed for a Video-on-Demand application which includes a video server, an ATM network, a fixed-size playback buffer, and an MPEG decoder. Bandwidth scheduling algorithms are developed which use the known frame sizes of the encoded video and the known playback buffer size to produce a rate profile. The rate profile breaks a long MPEG sequence into several segments and specifies the rate required for each segment. The segment rates allow the network to deliver VBR encoded video using...
20.
Optimal bandwidth allocation of multiple VBR video sources with priorities - Chu, Hon-Wai; Tsang, Danny H. K.; Liu, Liming
Using variable bit rate (VBR) coding allows ATM networks to transmit video efficiently by statistical multiplexing. The two-layer coding schemes which separate the coded bits into two priority streams are particularly well-suited for ATM networks. Separation is achieved by giving different access priorities of the shared buffer to the two streams. In this paper, we derive formulae to determine the cell loss rates of the two priority streams under three different access control policies. Based on the results obtained, we then study the trade-off of the cell loss rates as the parameters of the control policies vary. We also consider...
1.
