Tony T. Lee

Nonblocking copy networks for multicast packet switching

In addition to handling point-to-point connections, a broadband packet network should be able to provide multipoint communications that are required by a wide range of applications. The essential component to enhance the connection capability of a packet network is a multicast packet switch, capable of packet replications and switching, which is usually a serial combinations of a copy network and a point-to-point switch. The copy network replicates input packets from various sources simultaneously, after which copies of broadcast packets are routed to their final destination by the switch. A nonblocking, self-routing copy network with constant latency is proposed. Packet replications are accomplished by an encoding process and a decoding process. The encoding process transforms the set of copy numbers, specified in the headers of incoming packets, into a set of monotone address intervals which form new packet headers. The decoding process performs the packet replication according to the Boolean interval splitting algorithm through the broadcast banyan network, the decision making is based on a two-bit header information. This yields minimum complexity in the switch nodes. >

A modular architecture for very large packet switches

Switch modules, the building blocks of this system, are independently operated packet switches. Each module consists of a Batcher sorting network, a stack of binary trees, and a bundle of banyan networks. The modular architecture is a unification of the Batcher-banyan switch and the knockout switch, and can be physically realized as an array of three-dimensional parallel processors. Switch modules are interconnected only at the outputs by multiplexers. The partitioned switch fabric provides a flexible distributed architecture, which is the key to simplify the operation and maintenance of the whole switching system. The modularity implies less stringent synchronization requirements and makes higher-speed implementation possible. The proposed modular switch is intended to meet the needs of broadband telephone offices of all sizes. It is estimated that a modular switch with terabit capacity can be built using current VLSI technologies. >

M/G/1/N Queue with Vacation Time and Exhaustive Service Discipline

This paper studies an M/G/1 queueing system with a finite waiting room and with server vacation times consisting of periods of time that the server is away from the queue doing additional work. This model has been used in conjunction with a related model to analyze the performance of a processor with a cyclic scheduling algorithm and where, due to finite queueing capacities, losses are a primary concern. Service at the queue is exhaustive, in that a busy period at the queue ends only when the queue is empty. At each termination of a busy period, the server takes an independent vacation. The queue length process is studied using the embedded Markov chain. Using a combination of the supplementary variable and sample biasing techniques, we derive the general queue length distribution of the time continuous process, as well as the blocking probability of the system, due to the finite waiting room in the queue. We also obtain the busy period and waiting time distributions.

M/G/1/N queue with vacation time and limited service discipline

Abstract A variant of an M/G/1 queuing model with finite waiting room is studied in this paper. In this system, every busy period of the server at the queue is followed by the execution of additional tasks. The time spent by the server to perform these tasks is called a vacation time away from the queue. The server will begin a vacation from the queue if either the queue has been emptied or M customers have been served during the visit. The embedded Markov chain approach is used to obtain the steady state queue length distribution. The Laplace-Stieltjes transforms of the busy period and cycle time distributions are given. Using a combination of the supplementary variables and sample biasing techniques, the waiting time distribution, blocking probability and general queue length distribution are derived. Finally the results for the case of infinite waiting places are provided.

Path switching-a quasi-static routing scheme for large-scale ATM packet switches

A quasi-static routing scheme called path switching for large-scale ATM packet switch systems is proposed. Previously the Clos network has been used as the model for many large-scale ATM switch architectures, in which the most difficult issue is path and bandwidth assignment for each connection request. The static routing scheme, such as multirate circuit switching, does not fully exploit the statistical multiplexing gain. In contrast, the dynamic routing scheme, such as straight matching, requires slot-by-slot computation of route assignment. Path switching is a compromise of these two routing schemes. It uses a predetermined periodical connection pattern in the central stage, look-ahead selection in the input stage, and output queueing in the last stage. The scheduling of path switching consists of capacity assignment and route assignment. The capacity assignment is constrained by the quality of service of connection requests. The route assignment is based on the timespace interleaving of the coloring of bipartite multigraphs. We show that path switching can handle multirate and multimedia traffic effectively in the Clos network.

A broadband optical multicast switch

We propose a new optical muhicast switching system based on a two-phase contention resolution algorithm This architec ture may simultaneously support packet switching, and circuat channel emulation. It can handle time-multiplexed variable bit rate random access packets and reserved access packets in a single framework that incorporates unicast and multicast switching. We present the overall switch architecture, its optical device requirements, and possible implementation schemes. Performance enhancements through the addition of multiple tracks are discussed which indicate the flexibility inherent in this design. Results of an analysis of the throughput and switch performance are discussed.

Toward profit-seeking virtual network embedding algorithm via global resource capacity

In this paper, after proposing a novel metric, i.e., global resource capacity (GRC), to quantify the embedding potential of each substrate node, we propose an efficient heuristic virtual network embedding (VNE) algorithm, called as GRC-VNE. The proposed algorithm aims to maximize the revenue and to minimize the cost of the infrastructure provider (InP). Based on GRC, the proposed algorithm applies a greedy load-balance manner to embed each virtual node sequentially, and then adopts the shortest path routing to embed each virtual link. Simulation results demonstrate that our proposed GRC-VNE algorithm achieves lower request blocking probability and higher revenue due to the more appropriate consideration of the resource distribution of the entire network, when compared to the two lastest VNE algorithms that also consider the resources of entire substrate network. Then, we introduce a classical reserved cloud revenue model, which consists of fixed revenue and variable one. Based on this revenue model, we design a novel admission control policy selectively accepting the VNR with high revenue-to-cost ratio to maximize the InPs profit based on an empirical threshold. Through extensive simulations, we observe that the optimal empirical threshold is proportional to the ratio of variable revenue to the fixed one.

The design and analysis of an ATM multicast switch with adaptive traffic controller

The paper describes several improvements to a nonblocking copy network proposed previously for multicast packet switching. The improvements provide a complete solution to some system problems inherent in multicasting. The input fairness problem caused by overflow is solved by a cyclic running adder network (CRAN), which can calculate running sums of copy requests starting from any input port. The starting point can change adaptively in every time slot based on the overflow condition of the previous time slot. The CRAN also serves as a multicast traffic controller to regulate the overall copy requests. The throughput of a multicast switch can be improved substantially if partial service of copy request is implemented when overflow occurs. Call-splitting can also be implemented by the CRAN in a straightforward manner. Nonuniform distribution of replicated packets at outputs of the copy network may affect the performance of the following routing network. This output fairness problem due to underflow is solved by cyclically shifting the copy packets in every time slot. An approximate queueing model is developed to analyze the performance of this improved copy network. It shows that if the loading on each output of the copy network is maintained below 80%, the average packet delay in an input buffer would be less than two time slots. >

Broadband packet switches based on dilated interconnection networks

A theoretical foundation for the evaluation and comparison of a very broad spectrum of fast packet-switching techniques is developed. Based on this framework, the authors investigate the complexity of various packet switch designs, and demonstrate the advantage of dilation as a switch-design technique. Packet switches are classified either as loss systems or waiting systems, according to whether packets losing contention are dropped or queued. In a loss system, the packet loss probability can be made arbitrary small by providing enough paths between inputs and outputs. The authors focus on the question: how does the switch complexity grow as a function of switch size for a given loss probability requirement? A uniform approach to this problem is developed. It is shown that for an N/spl times/N switch, the required number of switch elements for both the parallel-banyan network and the tandem-banyan network is of order N(log N)/sup 2/, whereas the complexity of a dilated-banyan network is of order N log N(log log N). Within the class of waiting systems, it is shown that the parallel banyan networks in a Batcher-parallel-banyan network can be replaced by a dilated-banyan network without sacrificing the nonblocking property. Thus, as with parallelization, dilation can also be used to increase the throughput of a waiting system. In addition, the authors also explore the application of dilation in a large modular switch design which is realized by an interconnection structure consisting of Batcher-dilated-banyan networks and statistical multiplexers. >

Call scheduling algorithms in a multicast switch

Multicast switching is emerging as a new switching technology that can provide efficient transport in a broadband network for video and other multipoint communication services. The authors develop and analyze call scheduling algorithms for a multicast switch. In particular, they examine two general classes of scheduling algorithms: call packing algorithms and call splitting algorithms. The performance improvement by the call packing algorithms examined is shown to be negligible. In contrast, the call splitting algorithms can provide significantly lower blocking by reducing the level of output port contention. However, excessive call splitting could degrade performance because of the additional load introduced to the input ports. The authors present a simple call splitting algorithm called greedy splitting which achieves near-optimal performance. >