sanqi li - Academia.edu (original) (raw)

Papers by sanqi li

Proceedings of INFOCOM '94 Conference on Computer Communications

This pape? presents a novel approach to dynamic transmission bandwidth allocation for transport o... more This pape? presents a novel approach to dynamic transmission bandwidth allocation for transport of real-time variable-bit-rate video in ATM networks. Describe video traffic in the frequency domain: the low frequency signal captures the slow time-variation of consecutive scene changes; the high frequency signal exhibits the feature of strong frame autocorrelation. Our study indicates that the video transmission bandwidth in a finite-buffer system is essentially characterized by the low frequency signal. Since the time scale of scene changes is usually in the range of a second or longer, the low frequency video signal i s defined in a well-founded low frequency band. Hence, it is feasible to implement dynamic allocation of video transmission bandwidth using on-line observation and prediction of scene changes. Two prediction schemes are examined: recursive least square method us. time delay neural network method. A time delay neural network with low-complexity high-order architecture, called "Pi-Sigma Network", is successfully used to predict scene changes. The proposed dynamic bandwidth allocation scheme is shown to be promising and practically feasible in obtaining efficient transmission of real-time video trafic with guaranteed quality of services.

IEEE International Conference on Communications, - Spanning the Universe.

The authors focus on a gateway design for the interconnection of ILAN (integrated local area netw... more The authors focus on a gateway design for the interconnection of ILAN (integrated local area network) and ISDN (integrated services digital network). A key question is how to maintain the quality of service for each class of traffic as guaranteed by both networks. The emphasis of the study is on the main issues involved in the interconnection of MAGNET (a

Computer Communications, 2001

This paper presents a new mechanism of weighted fair bandwidth sharing for differentiated service... more This paper presents a new mechanism of weighted fair bandwidth sharing for differentiated services (DiffServ) assured forwarding (AF) services. The mechanism is named SCALE-WFS, i.e. Scalable Core with Aggregation Level labEling-Weighted Fair bandwidth-Sharing. It aims to achieve near-optimal max-min weighted fairness without per-flow management at core routers. Through extensive simulation and simple analysis, we show its advantages over the current

Globecom '00 - IEEE. Global Telecommunications Conference. Conference Record (Cat. No.00CH37137)

We have studied the delay related rate oscillation within Diffserv. The selective attenuation fee... more We have studied the delay related rate oscillation within Diffserv. The selective attenuation feedback via estimation (SAFE) is proposed to reduce the oscillation while maintaining fast response to network dynamics. SAFE has no per-flow accounting. Furthermore, the hashing technique is adopted to keep the operating overhead of SAFE to its minimum. System analysis supports the effectiveness of SAFE. The simulation result also shows that SAFE significantly reduces the rate oscillation, therefore achieves a high link utilization and small queue size while maintaining very low control overhead

IEEE INFOCOM, 1998

The research on guaranteeing Quality of Services has been to a large extent focusing on satisfy- ... more The research on guaranteeing Quality of Services has been to a large extent focusing on satisfy- ing cell-level performance such as cell loss rate, cell de- lay variation, etc. In the authors' previous work, we have used stochastic modeling techniques to evaluate the packet-level performance of two existing packet discard- ing schemes, packet tail discarding and early packet dis- carding,

Proceedings of INFOCOM'95

In this paper ' we present a method to establish real-time connections with guaranteed quality-of... more In this paper ' we present a method to establish real-time connections with guaranteed quality-of-services, based on persession (a, p)-characterization. Under two distinctive service disciplines, rate proportional processor sharing and fixed rate processor sharing, we derive tighter probabilistic bounds on per-session end-to-end average cell loss rate, which is caused by either buffer overflow in the route o r excessive delay a t the destination. One remarkable feature of the bounding solutions is that they are solely determined by the probabilistic (a, p)characterization of each session itself, independent of the network environment and other connections. To improve network resource utilization, our method is extended to allow statistical sharing of buffer resources at each node. The admission control scheme presented in this paper has a greatflexibility in connection management since both bandwidth and buffer resources can be adaptively allocated among incoming and existing sessions according to present network resource availability. The (a, p)-characterization strongly depends on traffic characteristics. Our study of real multimedia traffic streams reveals the interrelationship among (a, p)-characterization, traffic statistics and QOS constraint and also provides many engineering aspects of (a, p)-characterization for connection management.

Computer Networks and Systems, 1994

The study of continuous time queueing systems goes back to the early work of Erlang. However a so... more The study of continuous time queueing systems goes back to the early work of Erlang. However a somewhat different model, that of discrete time queueing systems, has recently been the subject of increasing attention. In a discrete time model time is assumed to be “slotted”. That is, time consists of a concatenation of fixed length intervals known as slots. Events are constrained to take place during these slots. For instance, a discrete time queue might accept at most one packet during a slot and service at most one packet during a slot. On a networkwide basis, multiple events may occur during each slot.

IEEE/ACM Transactions on Networking - TON, 1993

Communications in Statistics. Stochastic Models, 1996

In this paper we propose and implement a generalized Folding-algorithm for sojourn time analysis ... more In this paper we propose and implement a generalized Folding-algorithm for sojourn time analysis of finite QBD processes. It is a numerical method for the direct computation of where P is the QBD generator matrix in block tri-diagonal form. Define the QBD state space in two dimensions with Nphases and Klevels, so that and . The time and space complexity

IEEE Transactions on Communications, 1992

This work studies the performance of a nonblocking space-division packet switch in a correlated i... more This work studies the performance of a nonblocking space-division packet switch in a correlated input traffic environment. In constructing the input traffic model, the author considers that each input is a time division multiaccess (TDM) link connecting to multiple sources. Every source on a link supports one call at a time. Each call experiences the alternation of ON and OFF periods, and generates packets periodically while in ON period. The stochastic property of each call does not have to be identical. Packets from each individual call are destined to the same output. The output address of each call is assumed to be uniformly assigned at random. The author derives both upper and lower bounds of the maximum throughput at system saturation. His study indicates that, if the source access rate is substantially lower than the link transmission rate, the effect of input traffic correlation on the output contentions can generally be ignored. Also, the analysis of each input queue becomes separable from the rest of the switch. The same study is carried out with nonuniform call address assignment

IEEE Journal on Selected Areas in Communications, 1997

In this paper we present a method to establish realtime connections with guaranteed quality of se... more In this paper we present a method to establish realtime connections with guaranteed quality of service (QOS), based on a per-session probabilistic burstiness curve (PBC). Under two distinctive service disciplines, rate proportional processor sharing and fixed rate processor sharing, we derive useful probabilistic bounds on per-session end-to-end loss which is caused by either buffer overflow in the path or excessive delay to the destination. One remarkable feature of the bounding solutions is that they are solely determined by the PBC of each session itself, independent of the network environment and other connections. To improve network resource utilization, our method is extended to allow statistical sharing of buffer resources. The admission control scheme presented in this paper has a great flexibility in connection management since bandwidth and buffer allocations can be adaptively adjusted among incoming and existing sessions according to present network resource availability. We also present a novel method to compute the PBC of multimedia traffic based on the measurement of two important statistics (rate histogram and power spectrum). Our study of MPEG/JPEG video sequences reveals the fundamental interrelationship among the PBC, the traffic statistics, and the QOS guarantee, and also provides many engineering aspects of the PBC approach to real-time multimedia services in ATM networks. Index Terms-ATM networks, call admission control, end-toend performance bounds, traffic characterization, VBR video. I. INTRODUCTION O NE OF THE MOST challenging issues in supporting real-time multimedia communications in a high-speed network is to provide quality-of-service (QOS) guarantees to sessions. The QOS requirements of multimedia services are typically stringent, and differ depending on media and applications. The function of admission control is to determine whether or not an incoming session can be accepted at its requested QOS without violating QOS guarantees of ongoing sessions. In ATM networks, the QOS is mainly measured by end-to-end cell delay and loss performance. In this paper, each session connection is defined by where and represent, respectively, buffer space and transmission bandwidth allocated for the session at each switching Manuscript

IEEE Communications Magazine, 1998

IEEE/ACM Transactions on Networking, 1994

In this paper we use the well known technique of power spectral representation to characterize th... more In this paper we use the well known technique of power spectral representation to characterize the second order statistics of packet loss during congestion at a statistical multiplexer. Unlike the steady state statistics, the second order statistics measures the time correlation behavior of loss. Typically, the low frequency loss power is associated with long periods of highly consecutive loss. The e ect of the loss rate power spectrum on transmission qualities is service dependent. Our study indicates that, by selective packet discarding, one can properly change the distribution of the overall loss rate spectrum among the multiplexed tra c streams, which can signi cantly improve the transmission quality of individual services. Moreover, with the design of multi-level bu er overload control, one can tune the loss rate steady state distribution of di erent priority streams to a piece-wise step function, which minimizes the packet loss impact on service qualities.

Computer Networks, 2000

ABSTRACT