Vishal Sharma

Limited wavelength translation in all-optical WDM mesh networks

We analyze limited wavelength translation in all-optical, wavelength division multiplexed (WDM) wrap-around mesh networks, where up to W wavelengths, each of which can carry one circuit, are multiplexed onto a network link. All-optical wavelength translators with a limited translation range permit an incoming wavelength to be switched only to a small subset of the outgoing wavelengths. Although more restrictive than full wavelength translation (which permits an incoming wavelength to be switched to any outgoing wavelength), limited wavelength translation is a topic of recent study, since current practical wavelength translators are capable only of limited translation. We consider the case where an incoming wavelength can be switched to one of k (k=2,3) outgoing wavelengths (called the feasible wavelength set), and we obtain the probability that a session arriving at a node at a random time successfully establishes a connection from its source node to its destination node. Our analysis captures the state of a feasible wavelength set at a network node, which allows us to obtain the probability of successfully establishing the circuit. Based on this probability, we quantify the benefits of limited wavelength translation by demonstrating that in mesh networks, it can obtain most of the performance advantages of full translation at a fraction of the cost. Our work is the first to analyze limited wavelength translation for mesh networks under a probabilistic model, and accurately predicts the network performance over a wider range of network loads than previous works.

The ready-to-go virtual circuit protocol: a loss-free protocol for multigigabit networks using FIFO buffers

The ready-to-go virtual circuit protocol (or RGVC) is an immediate transmission protocol, in which the source need not wait for an end-to-end roundtrip delay for reservations to be made before transmitting the data. The protocol is designed to handle the lossless transport of ABR traffic, and will be used in the 40 Gb/s Thunder and Lightning testbed being prototyped at the University of California at Santa Barbara (UCSB). An important advantage of the RGVC protocol over previous connection and flow control protocols is that it is suitable for networks in which the switches use FIFO buffers that are shared by multiple sessions. The RGVC protocol ensures lossless communication by coupling link capacity with buffer space, so that when a portion of a buffer at a node is occupied, a proportional fraction of the incoming capacity to that buffer is frozen. Given the constraints on the frozen capacity, an algorithm is executed at each node to allocate the transmission rate to each FIFO buffer so as to maximize capacity utilization. The requirement that the protocol operate with FIFO buffers at the network nodes poses some unique challenges in the design that are not present in rate- and credit-based schemes. Briefly, since several sessions share a common FIFO buffer, per-VC flow control is no longer possible so control over the rate of an individual session is lost. Also, since the contents of the buffers change dynamically, the buffer composition becomes difficult to determine. For the rate-allocation algorithm of the RGVC protocol to be executed, however, the contents of the FIFO buffers at a node must be known, To implement the bookkeeping required, we present two schemes: the measurement-based scheme, where the bookkeeping function is implemented via measurements, done essentially in hardware; and the estimation-based scheme, where the bookkeeping is done analytically via the exchange of control packets between nodes.

An analysis of limited wavelength translation in regular all-optical WDM networks

We analyze limited-wavelength translation in regular all-optical, wavelength division multiplexed (WDM) networks, where up to W wavelengths, each of which ran carry one circuit, are multiplexed onto a network link. All-optical wavelength translators with a limited translation range permit an incoming wavelength to be switched only to a small subset of the outgoing wavelengths. We focus on the wraparound mesh and hypercube WDM networks, and analyze the case where an incoming wavelength can he switched to one of k (k=2, 3) outgoing wavelengths (called the feasible wavelength set). Our analysis captures the state of a feasible wavelength set at a network node, which allows us to obtain the probability that a session arriving at a node at a random time successfully establishes a connection from its source node to its destination node in each of these topologies. Based on this probability, we quantify the throughput and blocking performance of limited wavelength translation, and compare it to that of no wavelength translation and full wavelength translation. We demonstrate that in regular networks it can obtain most of the performance advantages of full translation at a fraction of the cost, and we present a simple, economical switch architecture to effect limited wavelength translation at a cost that is effectively independent of the number of wavelengths W in the system.

Switching using parallel input-output queued switches with no speedup

We propose an efficient parallel switching architecture that requires no speedup and guarantees bounded delay. Our architecture consists of k input-output-queued switches with first-in-first-out queues, operating at the line speed in parallel under the control of a single scheduler, with k being independent of the number N of inputs and outputs. Arriving traffic is demultiplexed (spread) over the k identical switches, switched to the correct output, and multiplexed (combined) before departing from the parallel switch.We show that by using an appropriate demultiplexing strategy at the inputs and by applying the same matching at each of the k parallel switches during each cell slot, our scheme guarantees a way for cells of a flow to be read in order from the output queues of the switches, thus, eliminating the need for cell resequencing. Further, by allowing the scheduler to examine the state of only the first of the k parallel switches, our scheme also reduces considerably the amount of state information required by the scheduler. The switching algorithms that we develop are based on existing practical switching algorithms for input-queued switches, and have an additional communication complexity that is optimal up to a constant factor.

An efficient reservation connection control protocol for gigabit networks

Abstract The Efficient Reservation Virtual Circuit protocol (or ERVC) is a novel connection control protocol designed for constant-rate, delay-insensitive traffic in gigabit networks. We explain the operation of the protocol, discuss its features and advantages, and present its performance characteristics. The ERVC protocol is appropriate for sessions that require an explicit reservation of capacity and can tolerate the round-trip delay associated with the reservations. In the ERVC protocol, the durations of the sessions are recorded, and every node keeps track of the utilization profile of each outgoing link, which describes the amount of residual capacity available on the link as a function of time. This feature allows capacity to be reserved only for the duration of the session, starting at the time it is actually needed. Therefore, the protocol utilizes capacity considerably more efficiently than regular reservation schemes do and results in markedly lower blocking probability for new sessions. The ERVC protocol also has the “reservation ahead” feature, which allows a node to calculate the time at which the requested capacity will be available and reserve it in advance, avoiding in this way the wasteful repetition of the call setup phase.

An analysis of oblivious and adaptive routing in optical networks with wavelength translation

We present an analysis for both oblivious and adaptive routing in regular, all-optical networks with wavelength translation. Our approach is simple, computationally inexpensive, accurate for both low and high network loads, and the first to analyze adaptive routing with wavelength translation in wavelength division multiplexed (WDM) networks while also providing a simpler formulation of oblivious routing with wavelength translation. Unlike some previous analyses which use the link independence blocking assumption and the call dropping (loss) model (where blocked calls are cleared), we account for the dependence between the acquisition of wavelengths on successive links of a sessions path and use a lossless model (where blocked calls are retried at a later time). We show that the throughput per wavelength increases superlinearly (as expected) as we increase the number of wavelengths per link, due both to additional capacity and more efficient use of this capacity; however, the extent of this superlinear increase in throughput saturates rather quickly to a linear increase. We also examine the effect that adaptive routing can have on performance. The analytical methodology that we develop can be applied to any vertex and edge symmetric topology, and with modifications, to any vertex symmetric (but not necessarily edge symmetric) topology. We find that, for the topologies we examine, providing at most one alternate link at every hop gives a per-wavelength throughput that is close to that achieved by oblivious routing with twice the number of wavelengths per link. This suggests some interesting possibilities for network provisioning in an all-optical network. We verify the accuracy of our analysis for both oblivious and adaptive routing via simulations for the torus and hypercube networks.

Circuit switching with input queuing: an analysis for the d-dimensional wraparound mesh and the hypercube

We analyze circuit switching in a multiprocessor network, where connection requests (or sessions) arrive at each node of the network according to a Poisson process with rate /spl lambda/. Each session joins the appropriate input-queue at its source node, and, upon advancing to the head of the queue, transmits a setup packet to establish a connection. If the setup packet is successful, it reserves the links on the path for the duration of the session, and the session is served without interruptions. Otherwise, the connection request remains queued at the source, and subsequent attempts are made to establish the circuit. We analyze the queue of connection requests at the input-buffer of a network link, and obtain analytic expressions for the stability region, the average queuing delay, the average connection time, the average waiting time, and the average total delay, which show how these parameters depend on system variables, such as network dimension and session arrival rate. The queuing analysis focuses on the input-queue of a particular link, and accounts for the interactions with queues of other links through the retrial attempts and the associated probability of success. The queuing analysis is independent of the particular network topology under consideration, as long as the probability that a session arriving at a random time successfully establishes a connection can be calculated for that network. Simulations demonstrate the close agreement between the observed network behavior and that predicted by the analysis.

A new analysis for wavelength translation in regular WDM networks

We present a new analysis of wavelength translation in regular all-optical WDM networks, that is simple, computationally inexpensive, and accurate for both low and high network loads. In a network with k wavelengths per link, we model the output link by an auxiliary M/M/k/k queueing system. We then obtain a closed-form expression for the probability P/sub succ/ that a session arriving at a node at a random time successfully establishes a connection from its source node to its destination node. Unlike previous analyses, which use the link independence blocking assumption, we account for the dependence between the acquisition of wavelengths on successive links of the sessions path. Based on the success probability, we show that the throughput per wavelength increases superlinearly (as expected) as we increase the number of wavelengths per link; however the extent of this superlinear increase in throughput saturates rather quickly. This suggests some interesting possibilities for network provisioning in an all-optical network. We verify the accuracy of our analysis via simulations for the torus and hypercube networks.

Some Closed Form Results for Circuit Switching in a Hypercube Network

We consider circuit switching in a hypercube network where each session has to establish a dedicated connection (or circuit) to a destination node for a random length of time. We first obtain an expression for the steady-state probability that a session successfully establishes a circuit, and then obtain analytic expressions for various delay parameters, such as the time between the arrival of a session and the time it is completed (including input queueing delays). The analytic expressions allow the delay parameters to be calculated without performing costly fixed-point iterations, and show how these parameters depend on the session arrival rate and the hypercube dimension.

Loss-free communication in high-speed networks

We introduce two novel connection control protocols for high-speed networks, which we call the efficient reservation virtual circuit (or ERVC) protocol and the ready-to-go virtual circuit (or RGVC) protocol. Both protocols have been developed for the 40 Gbit/s fiber-optic ATM-based Thunder and Lightning network, currently being designed and built at UCSB. The ERVC protocol, which is designed for constant-rate, delay-insensitive sessions, uses reservations and requires little buffering at intermediate nodes, while the RGVC protocol, which is designed for variable-rate and delay-sensitive sessions, uses back-pressure to control the source transmission rate and requires buffering at intermediate nodes. In this paper, we present the main features of both protocols and indicate the advantages of each.