1,721,162 research outputs found
Transmission analysis of a space-division optical star network with deflection routing
The combined traffic-transmission analysis of a single-wavelength
optical star local area network (LAN) is presented. The star is a
multistage space-division photonic switch which uses deflection
routing. Deflected packets delivered to the wrong user are
transparently re-routed to the switch. It is found that the network
works best when lightly loaded at the optical level. Using fast
optical transmitters and receivers to deplete the optical layer has
the positive effect of reducing both deflections and crosstalk
Weakly vs strongly multihop space-division optical networks
Transparent multihop optical networks suffer from
the accumulation from node to node of crosstalk and amplified
spontaneous emission noise, which may severely degrade the quality
of received signals. It is thus important to keep the number of
intermediate hops as low as possible. This paper compares two
single-wavelength cell-switching space-division optical networks
that employ deflection routing. The first has a well-known Manhattan
Street (MS) distributed topology. The mean internodal
distance of this network is approximately the square root of the
number of nodes.We term this network as strongly multihop. The
second has a centralized star topology: the star is a multistage
space-division photonic switch with limited buffers. Deflected
cells delivered to the wrong user are transparently rerouted to
the star. This network is intrinsically single-hop and gradually
becomes multihop because of deflections. We term this network
as weakly multihop. As the carried traffic increases, the link load
increases much more rapidly in the strongly multihop topology,
and so do both the crosstalk level per hop and the number of
hops caused by deflections. For the same carried traffic, the
accumulated crosstalk and spontaneous emission levels in a welldesigned
star-based network are much lower than in a strongly
multihop network. Hence, lower packet error rates and lower
delay jitter are expected for the centralized network. Moreover,
for both networks, a simple frequency sweeping technique is
shown to substantially reduce the dominant signal-crosstalk beat,
thus allowing network operation with switch crosstalk factors as
low as -20 dB
COMBINED TELETRAFFIC/TRANSMISSION ANALYSIS OF A TRANSPARENT SPACE-DIVISION OPTICAL STAR NETWORK
The combined t ele t r affic- tr ansmission analysis of
a single-wavelength cell-switching all-optical local area network
(LAN) with star topology is presented. The star is a multistage
space-division photonic switch that uses deflection routing. Deflected
packets delivered to the wrong user are transparently rerouted
to the switch. Each time a packet crosses the central
switch, it collects amplifier noise and crosstalk due to imperfect
optical switching. As the network load increases, the crosstalk
level per crossing increases, as well as the number of crossings
caused by deflections. The traffic statistics hence strongly affect
the quality of the received signals. It is found that such networks
work well only when lightly loaded. Using fast optical transmitters
and receivers to deplete the optical layer has the positive
effect of reducing both deflections and crosstalk, while allowing
simpler architectures at the optical level
SPACE-DIVISION OPTICAL STAR NETWORKS WITH DEFLECTION ROUTING
An optical star network is implemented bgy an MxM space-division cell switch to whichg M nodes are connected by dedicated fibers. Each node is equipped with an optical transmitter and an optical receiver. Deflection routng is used to simplify the structure of the central interconnect and the routing of packets within it. Deflected packets delivered to the wrong user are re-routed to the switchg. A multi-stage optical implementation of the central interconnect based on Sfuffle-Exchange stages made of 2x2 crossbar directional couplers is considered. The elementary beta swiching elements within the interconnect fabric contain either zero or one optical buffer. The number of stages n is varied from 1 to logM. The network topology thus evolves from a pure Shuffle Exchange network, which is multihop, to a potentially single-hop star network when the interconnect is complete (n=logM). It is shown how the drastic reduction of the hop count in the complete interconnect case allows both larger throughput and substantially lower loss as conpared to the multihop approach, making such topology an attractive candidate for a transparent optical implementation
DEFLECTION ROUTING IN MULTIHOP SPACE-DIVISION OPTICAL NETWORKS
Transparent multihop optical networks suffer from the accumulation from node to node of crosstalk and amplified spontaneous emission noise, which may severely degrade the quality of received signals. It is thus important to keep the number of intermediate hops as low as possible. This paper compares two single-wavelength cell-switching space-division optical networks that employ deflection routing. The first has a well-known Manhattan Street (MS) distributed topology. The second has a centralized star topology: the star is a multistage space-division photonic switch with limited buffers. Deflected packets delivered to the wrong user are transparently re-routed to the star. In both networks, as the network load increases, the crosstalk level per hop increases, as well as the number of crossings caused by deflections. Thge traffic statistics hence strongly affect the quality of the received signals. A simple frequency sweeping technique is shown to effectively reduce the signal-crosstalk beat, thus allowing network operation with switch crosstalk factors as low as -20 dB. It is found that a distributed topology like MS is not scalable in terms of both Throughput/delay and transmission quality, and the centralized topology should be preferred
Sensitivity penalty distribution in fibers with PMD: a novel semi-analytical technique
We proposed a novel semi-analytical technique for the evaluation of system outage probabilities in the presence of polarisation mode dispersion (PMD) at all-orders. Such technique relies on the rotational model for the fiber Muller matrix, whose rotational parameters were statistically characterized. Simulations of propagation can be performed either on a few hundred selected fibers or on a set of synthesized rotational models, to obtain reliable outage probability (OP) vs. average differential group delay (DGD) curves, with significant gain in simulation time
Is there life beyond the principal states of polarization?
In this paper we show that it is possible to describe exactly the field intensity distorted by
polarization mode dispersion (PMD) at the output of a single-mode fiber, without resorting to the
principal states of polarization (PSPs). Such description is based on the eigenmodes of what we
call the extracted matrices, for which we establish relationships with the PSPs, e.g., by showing their
alignment with the PSPs when these are not depolarized over the signal bandwidth.We also show that
the eigenmodes of the extracted matrices are less depolarized than the PSPs, and are thus the most
convenient frame of reference to express the output field and intensity for high bit rate transmitted
signals. We thoroughly investigate the properties of a recently introduced fiber model, which we
refer to as the rotation model, based on the rotation of the extracted eigenmodes, which extends the
intuition of Bruyère’s PSPs rotation model (1996) to the most stable frame of reference, and hence
is expected to give the most accurate description of the PMD fiber for large bandwidth signals. We
finally provide a novel eye closure penalty (ECP) formula based on the extracted eigenmodes, and
show its relation with the currently available ECP formulas. The accuracy of the novel ECP formula
is tested against simulation results of transmission through a fiber synthesized by the rotation model
Extracting PMD statistics from single emulated fibre sample
To obtain reliable PMD statistics from frequency-swept
measurements of emulated fibres, criteria are provided to relate
the measurement bandwidth to the fibre mean diffwential group
delay and the number of frequency samples
- …
