1,721,014 research outputs found

    ANALYTICAL EVALUATION OF IMPROVED TRANSMISSION TECHNIQUES IN DEFLECTION ROUTING NETWORKS

    No full text
    This paper presents improvements of existing analytical techniques to evaluate the performance of non-priority deflection routing networks in uniform traffic. Pure ghot-potato in a ShuffleNet topology will be presented for three polite-access techniques: Transmit-no-hold, Transmit-Hold and No-Head-of-Line-Blocking

    New structures of the optical node in transparent optical multihop networks using deflection routing

    No full text
    Novel single-receiver/single-transmitter/single-buffer node structures for ultrafast two-connected multihop transparent optical packet-switching networks with deflection routing are introduced. A Shufflenet topology in uniform traffic is used as a benchmark to compare several shared optical memory schemes and their control algorithms. These simple structures minimize the number of crossbar switches needed at each optical node and have moderate control complexity, while still yielding large throughput and small delay. The minimization of the number of crossbar switches results in an improved optical power budget. An analytical model, obtained by an extension of the existing theory, provides a design tool to search for efficient node control strategies that minimize the number of deflections and packet misses for each constrained node structure

    NEW STRUCTURES OF THE OPTICAL NODE IN TRANSPARENT OPTICAL MULTIHOP NETWORKS USING DEFLECTION ROUTING

    No full text
    New single-receiver/single- transmitter/single- buffer node structures for two-connected multihop transparent optical packet-switching network8 with deflection routing are introduced. A ShuBeNet topology an uniform trafic is adopted to compare various shared optical memory schemes and their control algorithms. These simple structures minimize the number of crossbar switches needed at the node and have moderate control complexity, while still yielding high throughput and low delay. Analytical results are obtained by an estension of the existing theory and verified by simulation

    Analysis and comparison of hot-potato and single-buffer deflection routing in very high bit rate optical mesh networks

    No full text
    The steady state behavior of regular two-connected multihoP networks in uniform traffic under hot-Pobto and a simple single-buffer deflection routing technique is analyzed for very high bit rate optical applications. Manhattan Street Network and ShuffleNet are compared in terms of throughput, delay, deflection probability, and hop distribution both analytically and by simulation. It is analytically verified that this single-buffer deflection routing technique recovers in both networks more than 6O% of the throughput loss of hot-pohto with respect to store-and-forward when packets are generated with independent the average message length exceeds 20 packets. destinations. This gain, however, decreases to below 40% when the average message length exceeds 20 packets

    Novel packet architecture for all-optical ultra-fast packet switching networks

    No full text
    Header recognition and packet detection in all-optical networks using on/otT optical ultrafast signalling at a fixed wavelength can be implemented by means of recently demonstrated optical sampling AND gates. A novel packet structure in which the header is spread in a TDM fashion over the optical packet allows the number of such AND gates to be minimised in the routing and receiving blocks thereby best exploiting the required electronic

    CHANNEL LIMITATIONS IN ULTRA-FAST MULTIHOP ALL-OPTICAL SOLITON NETWORKS

    No full text
    Non-regenerative ultra-fast packet switching networks are severely constrained by the optical channel. Limitations on the maximum achievable bit rate and network physical size are presented when soliton pulses are used in a ShuffleNet network with deflection routing

    MINIMUM LOSS NODE STRUCTURES FOR DEFLECTION ROUTING TRANSPARENT OPTICAL NETWORKS

    No full text
    New node structures with a single transmitterheceiver (TW RX) using single-buffer deflection routing' are proposed here for two-connected slotted multihop networks

    Design and channel constraint analysis of ultra-fast multihop all-optical networks with deflection routing employing solitons

    No full text
    Regular two-connected multihop transparent optical networks using ultrahigh bit-rate single-wavelength on/off keying are addressed. A novel solution for packet and node architecture is introduced to take full advantage of the recently demonstrated optical samplers and of the electronic processing capability. Channel transmission error arguments show how the size of these nonregenerative networks employing deflection routing is limited for a given optical bit rate. These limits are quantified for the Manhattan Street Network and ShuffleNet employing solitons. An upper bound on network performance in terms of maximum achievable bit rate and throughput for a given packet error rate is evaluated by taking into account the soliton self-frequency shift due to Raman scattering, the jitter due to amplified spontaneous emission noise, the short-range interaction, and their interplay. Results show that if the packet error rate is to be bounded below 10^-6, the node-to-node fiber span cannot exceed a few kilometers for network sizes greater than 64 nodes when the optical bit rate is as high as 100 Gb/s

    Self-clocking scheme for bit synchronization in ultrafast packet switching transparent optical networks

    No full text
    A challenging issue in ultrafast packet switching transparent optical networks is fast acquisition of synchronisation at the bit level for packet demultiplexing. In the Letter a novel scheme is proposed, in which a timing reference is transmitted along with each data packet in the form of a clock comb. This comb is extracted at each node reached by the packet Tor header recognition and, at the destination, for reading oi that particular packet. The synchronisation problem IS solved with no need for a phase locked loop, at the expense of an increased effective packet length
    corecore