By Topic

High Performance Switching and Routing, 2005. HPSR. 2005 Workshop on

Date 12-14 May 2005

Filter Results

Displaying Results 1 - 25 of 120
  • A cost-effective implementation of fast GMPLS shared protection for IP over 10 Gigabit Ethernet networks

    Publication Year: 2005 , Page(s): 100 - 103
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (263 KB) |  | HTML iconHTML  

    An experimental evaluation of the main limitations that affect failure detection and recovery in routed Gigabit Ethernet (GbE), i.e. GbE point-to-point connection between current IP/MPLS routers, is presented. An implementation of fast GMPLS shared protection is then successfully demonstrated in an IP over 10 Gigabit Ethernet testbed. Experimental results show that recovery times in the millisecond range are achieved. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Blocking probability of a high-speed optical switching network in presence of component failures

    Publication Year: 2005 , Page(s): 437 - 441
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (366 KB) |  | HTML iconHTML  

    A vertically stacked optical banyan (VSOB) network is attractive for serving as the optical switch architecture due to its small depth and absolute signal loss uniformity. A plane fixed routing (PFR) algorithm with optimal time complexity was proposed recently for rapid connection setup in a rearrangeable VSOB architecture. In this paper, we conduct the performance analysis of a high-speed PFR-based VSOB network and introduce an analytical model on blocking probability of the network when component failures (link failure and switching element failure) are considered. Simulation is conducted to verify the model, in which the simulation results show that the proposed analytical model can accurately describe the blocking behavior of a high-speed PFR-based VSOB network. The model significantly contributes to the related areas by providing network developers a quantitative guidance to determine the effects of adopting high-speed PFR algorithm, network component failures and workload variation on the blocking behavior of a VSOB network. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Measuring conditional Web throughput

    Publication Year: 2005 , Page(s): 442 - 447
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (393 KB) |  | HTML iconHTML  

    We focus on the analysis of the performance of HTTP transaction over TCP connections as derived from a measured data set collected at the ingress/egress point of Politecnico di Torino. We argue that looking at global average performance figures may hide some interesting trends that researchers and operators may want to detect. Thus, we partition the original measured set by looking at conditional densities and averages; considered conditional parameters include, among others, TCP connections RTTs, length of data set, adopted loss recovery algorithm. We show that particular care must be taken when analyzing data, since the measured set whose size is reduced by the partitioning scheme could be not large enough to ensure significance. A simple method to evaluate the significance of the presented measures is used. Ingenuity is required to understand behaviors not in line with the classical intuition driven by TCP knowledge. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Time-efficient near-optimal wavelength assignment in dynamically-reconfigurable WDM networks

    Publication Year: 2005 , Page(s): 453 - 456
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (254 KB) |  | HTML iconHTML  

    In wavelength-routed optical networks (WRONs), wavelength converters are considered as one of the most critical resources because they can significantly reduce the blocking probability but still remain quite expensive. Therefore, the problem of dynamic wavelength assignment that can optimally utilize converters is strongly desired. Unfortunately, previously proposed algorithms are impractical due to their huge computational complexities. In this paper, perhaps for the first time, we formulate this problem as an integer linear program (ILP) and solve it by a novel heuristics, called first-longest lambda-run algorithm (FLR). Our algorithm establishes a lightpath by chaining a minimum number of wavelength-continuous segments, called lambda-runs. The simulations in the NSFNET with different conversion capabilities show that more than 98% of lightpaths determined by FLR are optimal. As a result, FLR can significantly improve the network blocking performance, compared to first-fit algorithm. Especially, our algorithm is very time-efficient and easy to implement. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • ROSE: a novel link state information update scheme for QoS routing

    Publication Year: 2005 , Page(s): 24 - 28
    Cited by:  Papers (3)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (297 KB) |  | HTML iconHTML  

    Distributing the link state information may introduce a significant protocol overhead on network resources. Hence, it is critical to investigate the issue of effectively and efficiently updating link state information. In this paper, by demonstrating that link state update schemes without considering the QoS requirements of connections cannot provide satisfactory solutions to real networks, we introduce a novel link information update scheme, routing-oriented update scheme (ROSE). Via theoretical analysis and extensive simulations, we show that ROSE greatly outperforms the state of the art in terms of both protocol overhead and the accuracy of link state information. Another contribution of this paper is the introduction of a method to numerically evaluate the staleness of link state information. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Limits on the traffic carrying capacity of all optical networks with ring topology

    Publication Year: 2005 , Page(s): 44 - 48
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (244 KB)  

    This paper presents limits on the traffic earning capacity of a DWDM-based ring network. We develop limits for single-hop traffic, multi-hop traffic, as well as the total traffic carried by the network under the assumption that the incident traffic among all node-pairs is symmetric. We present some interesting results illustrating the total earned traffic by a ring network as a function of available wavelengths with fixed number of nodes, and as a function of nodes for a fixed number of wavelengths. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • An embedded solution to IBGP oscillations

    Publication Year: 2005 , Page(s): 312 - 316
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (344 KB) |  | HTML iconHTML  

    The occurrence of oscillations in the border gateway protocol (BGP [Rekhter, Y et al., 1995]) has been discussed in many papers. Autonomous system (AS) configurations using route reflection [Bates, T et al., 2000] are considered to be especially sensitive to oscillations ([Mc Pherson et al., 2002], [Griffin, T et al., 2002] and [Basu, A et al., 2002]). Moreover, such configurations are commonly used by the transit ASes where oscillations may have a significant impact on BGP stability and on associated traffic. This paper proposes a novel approach to avoid instability due to oscillations. It presents a lightweight embedded module added to route reflector(s) that detects internal BGP (IBGP) oscillations and provides assistance to solve them. This article gives motivation, architecture and experimentation results with the embedded module. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Reducing external speedup requirements for input-queued crossbars

    Publication Year: 2005 , Page(s): 222 - 225
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (258 KB) |  | HTML iconHTML  

    This paper presents a modified architecture for an input queued switch that reduces external speedup. Maximal size scheduling algorithms for input-buffered crossbars requires a speedup between port card and switch card. The speedup is typically in the range of 2, to compensate for the scheduler performance degradation. This implies, that the required bandwidth between port card and switch card is 2 times the actual port speed, adding to cost and complexity. To reduce this bandwidth, a modified architecture is proposed that introduces a small amount of input and output memory on the switch card chip. This architecture allows for internal speedup in the switch card and the external speedup between port card and switch card can be reduced significantly. A simulation study is used for buffer dimensioning and demonstrates the feasibility of the proposed architecture. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Self-tuning multiloop PI rate controller for an MIMO AQM router with interval gain margin assignment

    Publication Year: 2005 , Page(s): 401 - 405
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (342 KB) |  | HTML iconHTML  

    We design a self-tuning multiloop PI (proportional-integral) rate controller for the MIMO (multiple-input multiple-output) AQM (active queue management) router to support best-effort service traffic in the Internet. Unlike the manual trial and error approach, our controller design employs the DNA (direct Nyquist array) method by shaping Gershgorin band analytically, thus allowing users to achieve the good stability robustness of the MIMO AQM control system through assigning proper gain margin interval. The self-tuning multiloop PI rate controller is located in every output port of the MIMO router and can calculate the respective advertised source transmission rates based on the instantaneous queue length of the buffer. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A novel optical storage area network implemented in a metro WDM setting

    Publication Year: 2005 , Page(s): 49 - 52
    Cited by:  Papers (2)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (279 KB) |  | HTML iconHTML  

    This paper presents a review of various optical storage area network (SAN) related technologies. It reviews recent developments in SAN technologies. It also reviews the recent developments in metro wavelength division multiplexing (WDM) networks. The paper proposes metro WDM networks technologies for storage area networking. It presents analysis and simulation results of a proposed architecture of SANs based on metro WDM technologies. Throughput, delay and packet dropping probability results are presented under Poisson and self-similar traffic. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Distributed clustering method for large-scaled wavelength routed networks

    Publication Year: 2005 , Page(s): 416 - 420
    Cited by:  Papers (5)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (379 KB) |  | HTML iconHTML  

    The scalability of routing protocol has been considered as a key issue in large-scaled wavelength routed networks. Hierarchical routing scales well by yielding enormous reductions in routing table length, but it also increases path length. This increased path length in wavelength-routed networks leads to increased blocking probability because longer paths tend to have less free wavelength channels. However, if the routes assigned to longer paths have greater wavelength resources, we can expect that the blocking probability does not increase. In this paper, we propose a distributed node-clustering method that maximizes the number of lightpaths between nodes. The key idea behind our method is to construct node-clusters that have much greater wavelength resources from the ingress border nodes to the egress border nodes, which increases the wavelength resources on the routes of lightpaths between nodes. We evaluate the blocking probability for lightpath requests and the maximum table length in simulation experiments. We find that the method we propose significantly reduces the table length, while the blocking probability is almost the same as that without clustering. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Performance improvement of turbo coded multi-route multi-hop networks using parity check codes

    Publication Year: 2005 , Page(s): 274 - 277
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (214 KB) |  | HTML iconHTML  

    Wireless multi-hop networks are promising techniques for next generation wireless communications. Multi-hop networks have large channel capacity and can achieve multi-route transmissions. By using multi-route transmissions, we can obtain a route diversity effect at the destination nodes. In order to enhance the diversity effect, we usually adopt forward error correction techniques. Forward error correction techniques and turbo codes effectively work against severe wireless environment including fading channels. The iterative turbo decoder requires accurate channel information to achieve good decoding performance. In multi-hop transmissions, however, the receiver usually measures only the channel information of the last hop and it does not know the channel information of the whole route. In this paper, we propose the turbo coded multi-route multi-hop networks using parity check bits. By using parity bits, we can easily estimate the whole channel information in multi-route multi-hop networks. Using numerical examples, we show that the proposed scheme improves the performance in the fast fading channel conditions. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Building packet buffers using interleaved memories

    Publication Year: 2005 , Page(s): 1 - 5
    Cited by:  Papers (12)  |  Patents (8)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (309 KB) |  | HTML iconHTML  

    High end routers need to store a large amount of data. Dynamic random access memories (DRAMs) are typically used for this purpose. However, DRAM memory devices don't match the bandwidth requirements, especially in terms of random access speeds. In this paper, we analyze a generalized memory interleaving scheme. This scheme implements a large, fast memory using multiple, slower DRAMs. In the presence of small amount of speed-up, we show that reasonable statistical guarantees (i.e., low drop probabilities) can be provided by using small SRAM buffers that queue read/write requests to DRAMs. We then relate drop probabilities to SRAM buffer size for a wide range of statistical arrival patterns. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Expelling policies for shared memory fast packet switches with variable size packets of multiple priority

    Publication Year: 2005 , Page(s): 332 - 335
    Cited by:  Papers (2)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (222 KB) |  | HTML iconHTML  

    In this paper we study the efficient design of buffer management policies for QOS efficient shared memory based fast packet switches. There are two aspects that are of interest. First is the packet size: whether all packets have same or different sizes. Second aspect is the value or space priority of the packets, do all packets have the same space priority or different packets have different space priorities. We present two types of policies to achieve QOS goals for packets with different priorities: the push out scheme and the expelling scheme. For this paper we consider the case of packets of variable length with two space priorities and our goal is to minimize the total weighted packet loss. Simulation studies show that, expelling policies can outperform the push out policies when it comes to offering variable QOS for packets of two different priorities and expelling policies also help improve the amount of admissible load. Some other comparisons of push out and expelling policies are also presented using simulations. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Path-coupled signaling for NAT/firewall traversal

    Publication Year: 2005 , Page(s): 231 - 235
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (294 KB) |  | HTML iconHTML  

    Many complex protocols negotiate secondary flows on the application layer. Examples of such protocols include many peer-to-peer applications, SIP, H.323, etc. In general, this prevents firewalls from allowing them and through NATs public/private address space mapping from routing them. In this paper, we describe the requirements and design of an end-application triggered, path-coupled signaling protocol for NAT/firewall traversal. Finally, we show a prototypical implementation and discuss preliminary performance evaluation. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • TrueWay: a highly scalable multi-plane multi-stage buffered packet switch

    Publication Year: 2005 , Page(s): 246 - 253
    Cited by:  Papers (9)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (636 KB) |  | HTML iconHTML  

    Packet switches have been extensively studied during the last two decades. Most commercially available packet switches have a single path between input and output ports, which limits the scalability of the switch. In order to accommodate the exponentially increasing demand of Internet traffic, we propose an ultra scalable multi-path switch architecture, called TrueWay, a multi-plane multi-stage buffered switch. Packets are delivered between the stages of switch modules with link-to-link flow control to avoid overflowing the next-stage's buffers. Schemes such as back-pressure, credit-based and our proposed DQ scheme are discussed. One of the challenging issues of multi-path buffered switch is the maintenance of packet orders that can be resolved by appropriate port-to-port flow control. Schemes such as static hashing, time-stamp-based re-sequencing, dynamic hashing, and window-based re-sequencing, are considered. We show by simulation that the TrueWay switch with a speed up of 1.6 is able to perform nearly as well as the output buffered switch under most interested traffic distributions. A small-scale prototyped switch fabric has been built on a 16-card chassis with high-speed SerDes interconnections at the backplane (with 640 Gbps capacity), and with FPGA chips on each card to reconfigure the switch to test various stage-to-stage and port-to-port flow control schemes. With today's ASIC technology, e.g., 64×64 switch chip with SerDes interfaces and VCSEL (vertical cavity surface emitting laser) optical interconnections, the TrueWay switch can scale up to 40 Tbps. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Performance analysis of burst segmentation schemes supporting multiple traffic classes

    Publication Year: 2005 , Page(s): 495 - 499
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (302 KB) |  | HTML iconHTML  

    Optical burst switching (OBS) with the support of multiple traffic classes is an important topic. Assuming traffic classes are maintained, via choosing different offset times, a new OBS scheme called optical burst switching with burst splining (OBSS) is proposed in this paper. OBSS provides high system throughput by taking advantages of the idle gaps on output wavelength channels when the just-enough-time (JET) reservation scheme is used. Then an analytical model that supports N traffic classes is constructed and verified by simulations. The proposed model can be applied to the conventional OBS schemes as well as the OBSS scheme we proposed. Numerical results show that in our OBSS scheme, traffic classes can be effectively maintained via different offset times, and the packet loss probability is also much smaller than the conventional OBS. It is interesting to find that the choice of offset time for a specific traffic class affects the packet loss probabilities of classes higher than it, while the impact to classes below it is much smaller. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Survivable routing and wavelength assignment (RWA) in optical virtual private networks (O-VPNs)

    Publication Year: 2005 , Page(s): 426 - 432
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (385 KB)  

    This paper tackles the resource allocation problem for wavelength division multiplexing (WDM) networks supporting virtual private networks (O-VPNs), in which working and spare capacity are allocated in the networks for satisfying a series of traffic matrices corresponding to a group of O-VPNs. Based on the (M:N)n protection architecture where multiple protection groups (PGs) are supported in a single network domain, we propose two novel integer linear programming (ILP) models, namely ILP-I and ILP-II, aiming to initiate a graceful compromise between the capacity efficiency and computation complexity without losing the ability of addressing the QoS requirements in each O-VPN. ILP-I optimizes the task of resource allocation by taking each O-VPN as a PG, while the ILP-II breaks down each O-VPN into multiple small PGs where all the working paths in each PG are mutually link-disjointedly routed. Experiment results show that in terms of capacity efficiency, a significant improvement can be achieved by ILP-I compared to that by ILP-II at the expense of much longer computation time. Although ILP-II is outperformed by ILP-I, it can handle the situation with an arbitrary size of O-VPNs. We conclude that the proposed ILP-II model yields a scalable solution for the capacity planning in the survivable optical networks supporting O-VPNs based on the (M:N)n protection architecture. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Bandwidth management and delay control over EPONs

    Publication Year: 2005 , Page(s): 457 - 461
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (322 KB) |  | HTML iconHTML  

    Ethernet passive optical networks (EPONs) address the first mile of the communication infrastructure between the service provider central offices and the customer sites. As a low-cost, high-speed technology, EPONs are deemed as the solution to the bottleneck problem of the broadband access network. This paper proposes a dynamic bandwidth allocation scheme to control the data delay over the EPON upstream channel. The theoretical analysis of the average delay and the delay reduction, in conjunction with the extensive simulations, demonstrates that our scheme provides effective data delay control over EPONs. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Performance analysis of WDM optical buffers for asynchronous variable length packets

    Publication Year: 2005 , Page(s): 302 - 305
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (230 KB) |  | HTML iconHTML  

    Optical buffers realized with fiber delay line (FDL) have been used in optical packet switching networks for contention resolution. The dimension of delay granularity is the key to employ FDL efficiently. In this paper, we develop an approximate model for evaluating the performance of wavelength division multiplexing (WDM) based optical buffers for asynchronous variable length packets. The theoretical model is verified through simulation and the results show that the delay granularity of FDL is slightly dependent on the buffer length and the number of wavelength in a fiber for a given traffic load. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A method of detecting performance degradation at TCP flow level from sampled packet streams

    Publication Year: 2005 , Page(s): 157 - 161
    Cited by:  Papers (3)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (346 KB) |  | HTML iconHTML  

    Managing the performance at the flow level through traffic measurement is crucial for effective network management. On the other hand, with the rapid rise in link speeds, collecting all packets has become difficult, so packet sampling has been attracting attention as a scalable means of measuring flow statistics. We have therefore established a method of detecting performance degradation at the TCP flow level from sampled flow behaviors. The proposed method is based on the following two flow characteristics: (i) sampled flows tend to have high flow-rates and (ii) when a link becomes congested, the performance of high-rate flows becomes degraded first. These characteristics indicate that sampled flows are sensitive to congestion, so we can detect performance degradation of flows that are sensitive to congestion by observing the rate of sampled flows. We also show the effectiveness of our method using measured data. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Fuzzy logic-based dynamic routing management policies for mobile ad hoc networks

    Publication Year: 2005 , Page(s): 341 - 345
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (240 KB)  

    A routing algorithm (fuzzy dynamic routing protocol, FDRP) based on fuzzy logic-based was proposed for mobile ad hoc networks (MANET). It is applied to manage routing policies and to enhance routing performance dynamically. The parameters in MANET were classified with fuzzy degree of membership. The different parameter values were selected according to the different networks environments. The algorithm was integrated to AODV protocol and simulated by NS2. The FDRP algorithm was compared with standard AODV algorithm. The results of the simulation showed that the performance of the FDRP algorithm proposed the paper were noticeably improved. So, it is feasible that the fuzzy logic algorithm is applied to optimize routing performance of MANET. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Serving dynamic groups in application-level multicast

    Publication Year: 2005 , Page(s): 432 - 436
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (316 KB) |  | HTML iconHTML  

    We propose an ALM framework for dynamic groups (such as stock-quote application) in this paper, where users may hop from one multicast group to another quite frequently even though the total pool of users in the system may remain quite stable. Our approach efficiently maintains multiple multicast trees for dynamic subsets of end-hosts, and hence is called subset-ALM (SALM). SALM first builds a relatively stable mesh consisting of all end-hosts for control messaging, which is used to efficiently guide the construction of dynamic overlay trees for data multicast. We choose Delaunay triangulation (DT) as an example for mesh formation, and study various tree construction mechanisms based on the degree of embedding of the tree branches in the mesh (embedded, bypass and intermediate trees). Using simulation on Internet-like topologies, we show that SALM achieves low costs in terms of relative delay penalty and physical link stress, even for large multicast groups (in excess of a thousand end-hosts). View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A regenerator pool dimensioning method to avoid early blocking

    Publication Year: 2005 , Page(s): 83 - 87
    Cited by:  Papers (2)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (306 KB) |  | HTML iconHTML  

    We propose a method to dimension pools of regenerators which are preconfigured in the network nodes. We aim to minimize the overall number of regenerators in the network, given that a specified number of first connections are not blocked by these regenerators. Furthermore, we propose extensions of the method to minimize the blocking probability after the initial operation phase. The approach uses a connection set-up algorithm as subalgorithm to create a matrix for dimensioning calculation. The effectiveness of the method is assessed for two case study networks. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • HCF: a starvation-free practical algorithm for maximizing throughput in input-queued switches

    Publication Year: 2005 , Page(s): 68 - 72
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (315 KB) |  | HTML iconHTML  

    Using virtual output queueing (VOQ), maximum matching scheduling algorithms have been shown to achieve 100% throughput in input-queued switches, but has high complexity such that implementation is infeasible for high-speed systems. Iterative maximal matching algorithms, proposed as an alternative, cannot run for more than a few iterations due to the hardware complexity involved, thus resulting in low throughput. In this paper, we introduce a starvation-free iterative maximal matching algorithm called highest count first (iHCF). The iHCF algorithm gives preferential service based on the approximate age of the head-of-line cell in a VOQ and maximizes the size of the matching using round-robin priority pointers. We show that iHCF can achieve 100% throughput under i.i.d and uniform traffic in a single iteration. We also show using simulations that it performs as well as other known practical algorithms and achieves 100% throughput when run for only a few iterations under different admissible traffic patterns. Compared to other algorithms, iHCF leads to low complexity architecture such that the scheduling does not become the bottleneck. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.