Burst Rescheduling with Wavelength and Last-hopFDLReassignment in WDM Optical Burst Switching Networks

[attachment=17999]

I. INTRODUCTION
Wavelength division multiplexed (WDM) optical networks
are a promising candidate for the next generation backbone
transport networks. Optical burst switching (OBS), as described
in [1], [2], [3], [4] is able to strike a balance between
optical circuit and packet switching [2], [5], [6], [7].
A burst is formed at the ingress node by assembling a
number of IP packets that are destined to the same egress node.
A burst consists of a burst header (control packet) and a burst
payload (data burst). The control packet and the data burst are
initially separated by an offset time depending on the number
of hops the burst has to traverse. The base offset time required
is given by
H, where
is the control packet processing time
and H is the number of hops to be traversed. By extending
multi-protocol label switching (MPLS) capabilities to OBS
networks, explicit routing can be used at the ingress nodes [8].




II. PROPOSED BR-WFR RESCHEDULING ALGORITHM
We consider a network where every node is equipped with
a limited FDL buffer of size F with FDLi (i = 1, 2, . . . , F)
capable of optically delaying data by i time units. Each link
is assumed to carry W wavelengths.


A. Wavelength Reassignment
Wavelength reassignment considers reassigning the last
burst from a wavelength to the wavelength assigned to the
newly scheduled burst. Figure 1 illustrates the benefit of
wavelength reassignment. Figure 1(a) shows that without
wavelength reassignment, burst 7 cannot be scheduled. As
burst 1 through burst 6 have been scheduled one by one
to wavelength channels C1, C2 or C3 based on the latest
available wavelength similar to LAUC, burst 7 arriving at time
ta cannot be scheduled to any wavelength. Figure 1(b) shows
that with wavelength reassignment, burst 7 can be scheduled
successfully. Here, upon successful scheduling of burst 6, burst
5 at t2 is reassigned from C2 to C3 as a shorter void (t2 −t1)
is formed by burst 5 with burst 6 at C3 than with burst 2 at
C2. When burst 7 arrives, it can be scheduled to C2.



III. PERFORMANCE STUDY
In this section, the performance of the proposed BR-WFR
algorithm is studied through simulation. Its performance is
compared with that of LAUC and LAUC-VF. We also study
the performance of the rescheduling algorithm BR-WR which
differs from BR-WFR in that it carries out only wavelength
reassignment but not FDL reassignment. For a 95% confidence
level, the confidence interval is below 5% of the reported
values. Two metrics, burst dropping probability and performance
improvement are used to evaluate the performance. We
analyze the dropping probability for two classes of requests,
i.e., class-1 and class-2 where class-2 has a higher priority
over class-1. In order to differentiate class-2 from class-1
requests, class-2 requests are given extra offset time (three
times the average burst length used by class-1 requests plus the
maximum FDL length) so that they can reserve resources in
advance than class-1 requests [4].


CONCLUSIONS
We have proposed burst rescheduling for fast and improved
scheduling in WDM optical burst switching networks. Burst
rescheduling uses two mechanisms known as wavelength reassignment
and last-hop FDL reassignment. We have addressed
the implementation feasibility of burst rescheduling. Based
on the above rescheduling mechanisms we have developed
a computationally simple algorithm called BR-WFR. Through
simulation experiments, we have demonstrated that the performance
of the proposed algorithm is good for a range of
traffic loads and is close to that of the complex LAUCVF
void-filling algorithm under light loads.