Modern webpages consist of many rich objects dynamically produced by servers and client terminals at diverse locations, so we face an increase in web response time. To reduce the time, edge computing, in which dynamic objects are generated and delivered from edge nodes, is effective. For ISPs and CDN providers, it is desirable to estimate the effect of reducing the web response time when introducing edge computing. Therefore, in this paper, we derive a simple formula that estimates the lower bound of the reduction of the response time by modeling flows obtaining objects of webpages. We investigate the effect of edge computing in each webpage category, e.g., News and Sports, using data measured by browsing about 1,000 popular webpages from 12 locations in the world on PlanetLab.

Web performance, the time from clicking a link on a web page to finishing displaying the web page of the link, is becoming increasingly important. Low web performance of web pages tends to result in the loss of customers. In our research, we measured the time for downloading files on popular web pages by running web browsers on four hosts worldwide using PlanetLab and detected the longest portion in download time. We found the longest portion in download time to be Blocked time, which is the waiting time for the start of downloading in web browsers. In this paper, we propose a method for accelerating web performance by reducing such Blocked time with a cache of rendering results. The proposed method uses an in-network rendering function which renders web pages instead of web browsers. The in-network rendering function also stores the rendering results in its cache and reuses them for other web browsers to reduce the Blocked time. To evaluate the proposed method, we calculated the web performance of web pages whose render results are cached by analyzing the measured download time of actual web pages. We found that the proposed method accelerates web performance of long round trip time (RTT) web pages or long RTT clients if the web pages' dynamic file percentages are within 80%.

Modern websites consist of many rich objects dynamically produced by servers and client terminals at diverse locations. Consequently, we face complications in understanding the communication structure generated when accessing websites. To reduce the response time at browsed websites, many website objects are delivered using content delivery networks (CDNs), in which data objects are delivered from cache servers located close to user terminals. Although the use of CDNs have been assumed to reduce web response time, the actual effect of CDNs on this reduction has not been clarified. To answer this fundamental question, we measured the communication structure of traffic generated when accessing the 1,000 most popular websites from 12 locations worldwide. We found, for example, that it will be desirable to give high priority to entertainment websites at night and to business-related websites during the day.

Current collaborative filtering can't represent various aspects of users' interests. We propose a recommendation method in which a user can find new interests that are partially similar to the user's taste. Partial similarity is an aspect of the user's preference which is projected by the community in which the user belongs. We developed a television program recommendation system which performs such recommendation with serendipity, conducted an actual experiment and evaluated its results.

Because of the development of recent broadband access technologies, fair services among users are becoming more important criteria. The most promising scheme of router mechanisms for providing fair service is per-flow traffic management. However, it is difficult to be implemented in high speed core routers because per-flow state management is prohibitive; thus, a large number of flows are aggregated into a small number of queues. This is not a preferable situation because the more number of flows aggregated into a queue increases, the worse fairness tends to become. In this paper, we propose a new traffic management scheme called hierarchically aggregated fair queueing (HAFQ) to provide per-flow fair service. Our proposed scheme can adjust flow aggregation levels according to the queue handling capability of various routers. That means the proposed scheme is scalably used in high-speed networks. HAFQ improves the fairness among aggregated flows by estimating the number of flows aggregated in a queue and allocating bandwidth to the queue proportionally. In addition, since HAFQ can identify flows having higher arrival rates simultaneously in estimating the number of flows, it enhances the fairness by preferentially dropping their packets. We show that our proposed scheme can provide per-flow fair service through extensive simulation and experimental studies using a network processor. Since the currently available network processors (Intel IXP1200 in our case) are not high capacity, we also give extensive discussions on the applicability of our scheme to the high-speed core routers.

With the growth of computing power and the proliferation of broadband access to the Internet, media streaming has widely diffused. Although the proxy caching technique is one method to accomplish effective media streaming, it cannot adapt to the variations of user locations and diverse user demands. By using the P2P communication architecture, media streaming can be expected to smoothly react to network conditions and changes in user demands for media-streams. We propose efficient methods to achieve continuous and scalable media streaming system. In our mechanisms, a media stream is divided into blocks for efficient use of network bandwidth and storage space. We propose two scalable search methods and two algorithms to determine an optimum provider peer from search results. Through several simulation experiments, we show that the FLS method can perform continuous media play-out while reducing the amount of search traffic to 1/6 compared with full flooding.

To provide distributed multimedia applications with end-to-end QoS (quality of service) guarantees, resource reservation-based control mechanisms should be employed in both networks and end systems. In this paper, we propose a resource allocation scheme for real-time video multicasting described as a utility maximization problem. In this scheme, clients are first divided into multicast groups by means of a clustering technique. Then system resources are allocated to each group so that the total utility is maximized. We have confirmed that our proposed scheme can achieve effective use of resources while providing high-quality video to users.

In this article, we propose a personalized sports digest composition system with scenario templates and show its architecture. While conventional methods only select and show important events in a simple series, our system is capable to add a direction to them with considering a story of a whole game. We apply our prototype system to the summer Japan championship tournament of high school baseball. We provided the personalized digest service of the summer championship tournament of high school baseball in 2000 to the limited audience of the Internet and received the positive feedbacks of our system from about 100 users.

The policy-based service is becoming more important for the Internet shared by various applications. To realize the policy-based service, the router is required to forward packets according to the requirements of the traffic flow that the packet belongs to. However, the packet (and flow) classification easily becomes a bottleneck because the router is necessary to handle multiple fields within the packet. We propose a new packet classification algorithm capable of following the packet forwarding rate of the high-speed routers with the commercially available CPU, RAM, and cache memories. Through our experiments, we show the effects of parameters of the cache structure on the processing rate of the packet classification.

As the use of real-time multimedia applications increases, the bandwidth available to TCP connections is oppressed by "greedy" UDP traffic and their performance deteriorates extremely. In order that both TCP and UDP sessions fairly co-exist in the Internet, UDP sessions should properly react against congestion as TCP. We implement a "TCP-friendly" rate control mechanism suitable to video applications and investigate its applicability to a real system through observation of the video quality at the receiver. It is shown through our experimental system that we can achieve high-quality and stable video transfer while fairly sharing the network bandwidth with TCP by applying our rate control at a control interval of 16 or 32 times as long as the round trip time (RTT).

We investigate the performance of TCP under three representatives of packet scheduling algorithms at the router. Our main focus is to investigate how fair service can be provided for elastic applications sharing the link. Packet scheduling algorithms that we consider are FIFO (first in first out), RED (random early detection), and DRR (deficit round robin). Through simulation and analysis results, we discuss the degree of achieved fairness in those scheduling algorithms. Furthermore, we propose a new algorithm which combines RED and DRR algorithms in order to prevent the unfairness property of the original DRR algorithm, which appears in some circumstances where we want to resolve the scalability problem of the DRR algorithm. In addition to the TCP Reno version, we consider TCP Vegas to investigate its capability of providing fairness. The results show that the principle of TCP Vegas conforms to DRR, but it cannot help improving the fairness among connections in FIFO and RED cases, which seems to be a substantial obstacle for the deployment of TCP Vegas.

A method of evaluating brain function using the metacomputer concept of the Globus system combined with a message-passing interface is described. The proposed method has the ability to exploit various geographically distributed resources and parallel computing linked to a high-technology medical instrumentation system, magnetoencephalography, to analyze the functional state of the brain. It is envisaged that the method will lead to the realization of an efficient telemedicine system for health care.

We focus on fairness and stability of the congestion control mechanisms adopted in several versions of TCP by investigating their time-transient behavior through an analytic approach. In addition to TCP Tahoe, TCP Reno, and TCP Vegas, we consider enhanced TCP Vegas which is proposed in this paper for fairness enhancements. We consider the homogeneous case, where two connections have the equivalent propagation delays, and the heterogeneous case, where each connection has different propagation delay. We show that TCP Tahoe and TCP Reno can achieve fairness among connections in the homogeneous case, but cannot in heterogeneous case. We also show that TCP Vegas can provide almost fair service among connection, but there is some unfairness caused by the essential nature of TCP Vegas. Finally, we explain the effectiveness of our enhanced TCP Vegas in terms of fairness and throughput.

Previous analytic approaches for all-optical networks have only allowed a random wavelength assignment policy in spite of the fact that network performance can be improved by other wavelength assignment policies such as first-fit wavelength assignment. We develop an approximate analytic method to allow flexible wavelength assignment policies by virtue of a layered-graph approach. Our analysis is also applicable to the networks that wavelength conversion is limitedly provided. By comparing with simulation results, we show that our analytic approach has good accuracies when the number of wavelengths is not large. We also show that our analysis is applicable to general network topologies.

Previous works have shown that wavelength conversion can considerably improve the performance in all-optical networks. However, it is an unrealistic assumption under the current technology and wavelength conversion with a limited range, not a full range, is more promising. In this paper, we propose a new analytic approach for such networks by virtue of the layered-graph where each wavelength is virtually assigned a separate link. Our analytic approach can allow a flexible routing method. For example, wavelength assignment and routing selection can be considered at the same time, which has never been treated in the past literature.

A rate-based congestion control algorithm regulates cell emission rate of source end systems based on feedback information from the network. It was standardized by the ATM Forum for application to an ABR (available bit rate) service class. In the standard, two types of congestion notification methods of the switch are specified: EFCI marking and explicit-rate marking. In this paper, we focus on the explicit-rate marking switch. We propose our enhancements on a recently proposed switch algorithm known as the max-min scheme. The main objective of our enhancements is to control the queue length of the switch for preventing cell loss and achieving full link-utilization. We show the effectiveness of our switch algorithm by simulation experiments.

We study routing methods in all-optical switching networks. In all-optical switching networks, the connection with more hops encounters more call blocking, and it is especially true in optical networks with no wavelength conversions. We therefore consider an alternate routing method with limited trunk reservation in which connections with more hops are prepared more alternate routes. Through developing an approximate analytic approach, we show that our method keeps good performance when compared with the existing alternate routing methods, and also that the fairness among connections can be improved. Further performance improvement is investigated by introducing a wavelength assignment policy and a dynamic routing method. An effectiveness of the proposed method is investigated through simulation.

The rate control method of MPEG-2 over the ATM DBR service class is proposed. It is intended to guarantee the quality of service required by video applications. The disadvantages of MPEG-2 Test Model 5 are resolved by introducing the bandwidth re-negotiation with the network. And user-oriented high quality video transfer can be guaranteed with the method.

An ATM switch with both input and output buffers provided with a back-pressure function has been proposed as a cost-effective switch architecture. The back-pressure function prohibits cell transmission from the input buffer to the corresponding output buffer to avoid cell loss at the output buffer due to a temporary congestion. Especially when this switch is applied to ATM LANs for data transfer services, its performance should be evaluated by taking into account bursty traffic. In this paper, we show the maximum throughput, the packet delay distribution, and the approximate packet loss probability of such an ATM switch for bursty traffic through an analytic method. In addition to a balanced traffic condition, an unbalanced traffic and a mixture of bursty and stream traffic are also analyzed. Through several numerical examples, we quantitatively show the effects of the average packet length and the output buffer size on its performance. Key words: ATM LAN, Input/Output Buffered Type S...

The FRP (fast reservation protocol) utilizes a unique feature of ATM (asynchronous transfer mode) technology. The FRP is supposed mainly to be applied to LAN interconnection. We provide an exact analysis for a class of the FRP to obtain the burst (a protocol data unit of FRP) level performance of the FRP. One of the main features of the ATM is that the transmission rate can be adjusted according to the network congestion states. In our modeling, this feature is incorporated in such a way that the bandwidth a source requests is reduced if an attempt to reserve the bandwidth is rejected by the network. A rationale behind this is that rejection of the bandwidth request indicates network congestion. Therefore, the request with the smaller bandwidth after the reservation failure enables better sharing of network resources, which results in a performance improvement. To solve our model, which contains a very huge number of system states, we introduce a new numerical approach which is an extension of Sumita and Rieders's (1991) replacement process approach. Through numerical examples, the appropriate transmission rate control method is examined, and considerations on backoff time and overhead of RM cells are also presented.

In ATM networks, preventive congestion control is widely recognized for efficiently avoiding congestion, and it is implemented by a conjunction of connection admission control and usage parameter control. However, congestion may still occur because of unpredictable statistical fluctuation of traffic sources even when preventive control is performed in the network. The authors study another kind of congestion control, i.e., reactive congestion control, in which each source changes its cell emitting rate adaptively to the traffic load at the switching node (or at the multiplexer). The intention is that, by incorporating such a congestion control method in ATM networks, more efficient congestion control is established. They develop an analytical model, and carry out an approximate analysis of reactive congestion control algorithm. Numerical results show that the reactive congestion control algorithms are very effective in avoiding congestion and in achieving the statistical gain. Furthermore, the binary congestion control algorithm with push-out mechanism is shown to provide the best performance among the reactive congestion control algorithms treated.>

The authors consider a three-stage switching configuration with no internal buffers, i.e., bufferless switches are used at the first and second stages, and output buffered switches at the third stage. Short-term cell loss probability is analyzed in order to examine the influence of bursty traffic on the performance of the bufferless switches used at the first two stages. A four-stage switching configuration with traffic distributors added at the first stage is proposed. This switch provides more paths between a pair of input and output ports than the three-stage configuration. Some schemes for distributing cells are compared. It is shown that the distributor successfully reduces cell loss probability due to bursty traffic by splitting incoming cells into several switching modules.>

The authors analyze the flow enforcement algorithm suitable for regulating the flow of the bursty traffic in asynchronous transfer mode (ATM) networks (for broadband packet switching). The flow enforcement algorithm has two objectives: to force the input traffic to conform to the traffic parameters which are specified by the source itself at its connection time; and to obtain greater performance gain at the cell multiplexer by regulating the flow to some extent. The authors reveal the tradeoff relationship between the regulation level of the cell flow and the improvement of the link performance. The authors first analyze the flow enforcement mechanism to derive the delay distribution and the interdeparture time distribution for cells at the flow enforcer. For the input traffic, they allow the general independent arrivals of cells to represent the traffic burstiness larger than a Poisson distribution. The obtained interdeparture time distributions from the flow enforcer are then used to approximately analyze the performance of the cell multiplexer. For this purpose, a Markov-modulated Poisson process (MMPP) approximation method is extended to investigate the performance of individual traffic stream in the superposed arriving traffic streams.>

The performance of nonblocking packet switches such as the knockout switch and Batcher banyan switch for high-speed communication networks can be improved as the switching capacity L per output increases; the switching capacity per output refers to the maximum number of packets transferred to an output during a slot. The N*N switch with L=N was shown to attain the best possible performance by M.J. Karol et al. (1987). Here a N*N nonblocking packet switch with input and output buffers is analyzed for an arbitrary number of L such that 1>

Admission control in asynchronous transfer mode (ATM) networks is considered. An ATM switch is modeled as a discrete-time single-server queue at which the following three different kinds of arrival processes are allowed to join together: arrivals of cells with a general interarrival time distribution; Bernoulli arrivals of cells in batches; and interrupted Poisson processes. An exact analysis is given to derive the waiting-time distributions and interdeparture-time distributions for arriving cells subject to admission control in ATM networks. The model is extended to approximately obtain the end-to-end delay distributions for the designated traffic stream. Such an analysis is important for voice or coded video because they require a playout mechanism at the destination. Since the analysis is approximate, the simulation results needed to assess its accuracy are provided. Numerical examples show how the traffic characteristics of the designated traffic stream are affected by the total traffic load at the switching node, the number of other bursty traffic sources, and the degree of burstiness of the traffic source itself.>

An analytical model for predicting the performance of protocol software implementation in a layered communication architecture is proposed. Each layer is modeled by a queuing network. The model is constructed by focusing the behavior of the protocol entity in a layer. A first come, first served (FCFS) server queue with preemptive priority is employed for an entity and an infinite-server queue for response time of acknowledge packets. Respective layer models are piled up by embedding the lower services in the infinite-server queue. From the perspective of end-to-end performance, the model provides a solution for the following problems, which arise in implementing a layered communication protocol: assigning each layer's process to a processor; giving priority to respective processes in each processor; and providing an error recovery service. The authors present an approximate analysis to obtain an average end-to-end transmission delay and discuss the possible solutions for the problems mentioned.>

The major ATM switch architectures and their improvement techniques are categorized and discussed, focusing on performance issues. The performance measures of interest are the maximum throughput, the delay time, and the cell loss probability. The major assumptions and notations used are summarized. A survey is presented of nonblocking switches and their improvement techniques. The performances of a variety of nonblocking switches are compared.>

An exact analysis is presented for the 2*2 switching element with input port buffers that use the cell bypass queueing discipline. An approximate analysis is developed for the Banyan network composed of 2*2 switching elements with buffers that use the cell bypass queuing discipline. An investigation is made of the effect of this discipline on performance through numerical results obtained by analyses as well as simulation results. The authors also examine the nonuniform input traffic models. The analysis shows that the cell bypass queuing discipline considerably improves the performance of the ATM (asynchronous transfer mode) switches based on Banyan networks with respect to both throughput and cell blocking probability.>

The performance of an asynchronous transfer mode (ATM) network subject to admission control and traffic smoothing is analyzed. Basically, an ATM switch is modeled as a discrete-time single-server queuing system in which a new call joins existing calls. Cell arrivals from a new call are assumed to follow a general distribution. It is also assumed that aggregated arrivals of cells from existing calls form batch arrivals with a general batch size distribution and a geometric distribution of the interarrival times between batches. Both finite- and infinite-buffer cases are considered. An exact analysis yields the waiting time distribution and cell loss probability for a new call and for existing calls. Numerical examples are given to show how the network performance depends on the statistics of a new call (burstiness, time that a call stays in an active or inactive state, etc.) and to demonstrate the effectiveness of admission control and traffic smoothing.>

An analytical method is developed for a token ring network with single-buffer terminals and an infinite-capacity bridge in which the bridge has priority over the terminals is developed. The priority-mode operation realized by the reservation discipline of IEEE standard 802-5 is considered. An approximate analysis is considered for both terminals and the bridge in two reservation disciplines: the T-reservation discipline in which the reservation field resides in the message trailer, and the H-reservation discipline in which the reservation field resides in the message header. System performance measures (mean waiting times and throughput) derived for these disciplines are compared to those for a mixed discipline which is an alternative method for priority-mode operation in token ring networks. The accuracy of the approximate analytical results is assessed for both terminals and the bridge by comparing them to the simulation results. For almost all tried cases, it is shown to be within 5-20% relative error in the mean message waiting time.>

An analysis is made of the throughput and delay performance of two classes of free-access tree algorithms with minislots. In one class, binary feedback information is available in minislots, and in the other, ternary feedback information is available. It is shown that the highest maximum throughput 0.56714 is achieved in the limiting case where the number of minislots in a (large) slot is infinity and minislot overhead is zero. A lower bound of the average transmission delays in these algorithms is analytically derived. The obtained lower bound is also a lower bound of the average delay of the whole class of the free-access algorithms.>

A token ring network is considered, with priority-mode operation by a reservation discipline which is adopted by an IEEE standard 802.5. In the reservation discipline, a high-priority station can reserve the next transmission right in the reservation field of the message transmitted by the low-priority station when the reservation field passes through the high-priority station. Then the free token at the high-priority level is passed to the high-priority station. After all messages from the high-priority stations are transmitted, the free token is returned to the station in the downstream of the previous low-priority station. To apply this reservation discipline to an interconnected token ring network system, an analytical model is constructed which consists of low-priority stations with single buffer and a single high-priority station with infinite buffer, which is intended to represent a bridge. An exact analysis for the low-priority stations and an approximate analysis for the high-priority station in the reservation discipline is presented. System performance measures (mean waiting times and throughput) derived from the present analysis are compared with a mixed discipline which is an alternative method for priority-mode operation in token ring networks.>

A design method for packet-switched networks is presented that enhances the reliability and prevents the performance degradation caused by network failures. The method is based on a design technique wherein the redundant capacity is assigned to appropriate links in advance to be capable of accommodating the detoured traffic caused by failed links. Using this technique, it is crucial to accurately evaluate to how large an amount the capacity for each link must be augmented. An augmented capacity function is introduced, which determines the capacity of a link based on the probability of a particular network state and a parameter which represents how important the link is in the state. Numerical examples are given to demonstrate the feasibility of the method, in terms of cost/performance effectiveness. The relation between the reliability level and the network cost is discussed.>

Introduction of the packet switching technique into digitized voice communication may afford great advantages in efficient use of the channel, compared to both circuit-switched and DSI systems. Detailed characteristics, however, have not been obtained because of difficulty in the exact analysis. Hence, simalation models are developed in this paper for the packetized voice transmission system, and various characteristics such as tranmission delays and loss probability of voice packets are obtained. We further evaluate three types of voice packet reassembly strategy at the receiving terminal, and obtain the optimal packet length, which keeps both overall packet transmission delay and packet loss probabilty less than a certain permissible value. Comparison among three strategies is also stated.

A method for realizing a circuit and packet integrated access scheme in a satellite communication channel is considered. Two kinds of terminals are assumed, namely, bursty terminals for handling bursty traffic and heavily loaded terminals for long-holdingtime message traffic. In this method, the channel frame is divided into two subframes: one is for bursty terminals, and the other is for heavily loaded terminals. The subframe for heavily loaded terminals is further divided into two subchannels, a reservation subchannel (consisting of small slots) and a message subchannel. The bursty terminals transmit their packets in their dedicated subframes on the slotted ALOHA protocol. The heavily loaded terminal having a message transmits, first of all, a reservation packet in a randomly selected small slot of the reservation subchannel to reserve slots in the coming message subchannels. One slot in the same position of each of the succeeding message subchannels is reserved for the terminal until the end-of-use flag, transmitted from the terminal, is received by the satellite. Mean transmission delays for both kinds of traffic in this method are analytically obtained. We show that there exists an optimal frame length which minimizes mean transmission delay for one kind of traffic while keeping mean transmission delay for the other kind under some permissible value.

This paper considers a satellite communication system with the slotted ALOHA access mode where the satellite is assumed to have on-board regeneration ability and the finite buffer capacity for storing noncollided packets from the uplink. The system is analyzed with the downlink bandwidth restricted to1/Lof the uplink bandwidth, and with the conventional slotted ALOHA system whereL = 1and the buffer size at the satellite is equal to zero. Maximum throughput and throughput-delay characteristics of this system are analytically obtained. The satellite is given an optimal buffer size to achieve maximum throughput and minimum packet transmission delay. It is shown that the constraints that lead to the downlink bandwidth being lower than the uplink bandwidth do not necessarily result in throughput penalties in a buffered satellite system. Furthermore, when the total bandwidth which can be divided between up- and downlinks is given, if two or three times that of the downlink is assigned to the uplink and if the satellite has buffer storage space only for a few packets, we show that maximum throughput of the system will be improved about 40 percent with little change in delay characteristics in comparison with that of the conventional slotted ALOHA system where the given total bandwidth is equally divided between up- and downlinks.

Message, packet and line switching in computer communication networks are analyzed by a queueing model. Message transmission delay time and network throughput between a source-destination node pair are obtained as a function of various parameters including message length, traffic arriving at the network, and the number of switching nodes existing between the nodes. A criterion to determine the most suitable switching method under the given conditions is offered. Also, the maximum length of a packet in packet switching is discussed.