21-05-2012, 03:13 PM
Auto Tracking System for Free Space Optical Communications
[attachment=22587]
Abstract
This paper proposes a transceiver system for
automatic tracking and dynamic routing for free space optical
(FSO) communication. The proposed transceiver architecture
has M transmitters and M receivers in a M x M configuration
that can dynamically orient themselves at different angles to
establish a link between two transceivers. A request to setup a
lightpath is blocked if a free wavelength is not available at a
transmitter or a receiver in a node along the route. The number
of resources remaining the same, in comparison to a fixed FSO,
the methods we propose for routing and recovery that exploit the
M x M transceiver architecture achieve reduced blocking
probability as well as increased percentage of recovery of
affected traffic after a link failure .
Index Terms—Free space optical (FSO) communication;
GMPLS; optical network; survivability; dynamic routing .
I. INTRODUCTION
ITH the tremendous growth in bandwidth intensive
dynamic real time traffic, the network architecture is
shifting toward the model that consists of high speed routers
and optical fibers. The requirement of high capital cost for the
fiber to the home service and the strict RF regulations create
the gap to connect these infrastructures to the real users. Free
space optics (FSO) technology that establishes point to point
communication links is a good candidate to exploit the
tremendous capacity of optical fibers for last-mile access,
metro network extensions, and enterprise connectivity. Due to
unregulated bandwidth, FSO provides low cost, low power,
high security and high rates [1]. Generally, the FSO
transceivers remain in static location to avoid any
misalignments to maintain continuous line of sight. The point
to point FSO links provide higher data rate but with limited
coverage. Consequently, the blocking probability increases. In
wireless environments, the channel is highly dynamic, and the
system will likely experience a considerable degradation in
performance. Survivability is also an issue that must be
addressed. The main challenge in FSO network design is to
exploit the available resources efficiently when operating in
wireless environment.
Most of the papers, for example [5, 6], studythe fixed FSO
system. In this paper, we have modified the static architecture
of FSO transceiver and introduced transceiver capabilities that
provide automatic tracking and dynamic routing features. The
blocking performance and the survivability after a single link
failure are studies via simulation.
The rest of the paper is organized as follows: Section II
provides an overview of the system architecture. Section III
presents simulation results and section IV presents
conclusions from our study.
II. SYSTEM ARCHITECTURE
Figure 1 shows three different configurations of tracking
FSO that adapt to dynamic environment to find available links
from source to destination. The architecture uses M x M
transceivers, each with M transmitters and M receivers that
can rotate at different angles to establish a connection.
Assuming the central node to be the source node or the
transmitting station, we consider three different scenarios.
A. Scenario I
As shown in Figure 1a, the nodes are able to track and
switch the traffic to adjacent nodes that are at an angle of 0,
90, 180 or 270. It has capability to switch the traffic at full
rate.
B. Scenario II
As shown in Figure 1b, the nodes can track and switch the
traffic to adjacent nodes that are at an angle of 0, 45, 90, 135,
180, 225, 270, 315 and has capability to switch the traffic at
full rate.
C. Scenario III
Figure 1c shows the architecture in which each node can
switch the traffic to other nodes that are at an angle of 0, 45,
90, 135, 180, 225, 270 and 315 as well as to nodes that are
two hops away. In this case, nodes transmit data at full rate to
the adjacent nodes and at half the full rate to nodes that are
two hops away.
Generalized multi-protocol label switching
(GMPLS),developed in the Internet Engineering Task Force
Auto Tracking System for Free Space Optical
Communications
Rabindra Ghimire and Seshadri Mohan
W
(a) (b) ©
Fig. 1. Architecture for tracking FSO
ICTON 2011 Tu.D5.4
978-1-4577-0882-4/11/$26.00 ©2011 IEEE 1
(IETF) is used as the control plane protocol [2]. Signaling is
assumed to be out of band. GMPLS can be deployed in FSO
where wavelengths are used to establish low level point-topoint
links for the transmission of packets between high
performance routers. In optical internet, GMPLS routers
translate label assignments into corresponding wavelength
assignments and different network nodes communicate with
each other via end-to-end all optical connections known as
lightpaths. GMPLS employs open shortest path first with
traffic engineering (OSPF-TE) as the routing protocol and
resource reservation protocol with traffic engineering (RSVPTE)
as the signaling protocol. OSPF-TE floods the nodes with
network topology information, which each node uses to
compute paths to other nodes in the network, whereas RSVPTE
utilizes the paths computed by nodes to send signaling
messages and establish lightpaths. This paper adapts the
destination initiated routing scheme proposed in [3]. In this
case, OSPF-TE floods the network with summarized
information of network topology. It advertises all links from a
particular node as alive as long as single free transmitter is
available at that node.
If a node receives a request to set up a unidirectional
lightpath, a route is calculated using the information provided
by routing protocol. The signaling session is triggered and it
will initiate RSVP-TE protocol to send path message along the
found route. All signaling are assumed to be out of band.
RSVP-TE forwards a path message towards the destination.
The path message carries Explicit Route (ER) object and
Label Set (LS) object [4]. ER carries the route that path
message follows and LS carries a set of available transceivers
that can be selected to establish a lightpath. The possibility of
optical to electrical and electrical to optical conversion is
assumed if required. If a lightpath cannot be established due to
the unavailability of transceivers, then blocking occurs, and
path error message is sent back to source node. If the path
message arrives at the destination node, it will initiate a resv
message to reserve resources along the reverse of the route
traversed by the path message. If the resv message is unable to
reserve resources at a node as other requests might have
reserved some selected wavelengths in the mean time, then the
call will be blocked due to unavailability of resources. If
blocking occurs, a resv error message is sent towards the
source and a resv tear message towards the destination. If no
blocking occurs, then resv message arrives at the node
indicating that resources have been reserved and a lightpath
has been established between the source and the destination
nodes. In case of link failure, the node close to the failed link
is responsible for detecting the failure and sending notify
message to all other nodes. After receiving notify message, the
source node will look for an available route by initiating
RSVP-TE protocol. Note that the occupied resources affected
by failure are not released to reestablish the affected lightpaths
as that complicates the procedure. The source node attempts
only once to recover the affected lightpath.
III. SIMULATION SCENARIO
We simulate using OPNET the performance of a 8 by 8
mesh tracking FSO network shown in Figure 2. The traffic is
generated uniformly. Any node can be selected as source and
destination node. Shortest path is given higher priority. Any
available transmitter can be selected randomly at the
destination to establish a light path. The arrival of requests for
lightpaths follows a Poisson process with exponential holding
time. Traffic in the network is varied by varying the ratio of
holding time to inter-arrival time. The traffic is uniformly
distributed between every pair of nodes in the network. The
traffic is converted into wavelength level end-to-end lightpath
requests.
A. Case I
In this case, only 2 x 2 transceivers with tracking capability
are assumed in each node. Figure 3 compares the blocking
performance for scenarios I, II and III. Blocking in scenario I
is relatively more compared to scenario II and scenario III. In
scenario I, the tracking is at angles of 0, 90, 180 and 270,
there by limiting the available route towards the destination
compared to scenarios II and III. In scenario II and scenario
III, tracking occurs at angles of 0, 45, 90, 135, 180, 225, 270
and 315. Scenario III can further switch to nodes that are two
hops away but with transmission rate of one half the full rate.
Similar performance is observed for scenario II and scenario
III. In scenario III, since transmission of data is at half the rate
for the links that are two hops away, any route using two hops
long link along a lightpath will be occupied for twice as long
as those lightpaths using only single links. The simulation
parameters are set in such a way that two hop links are used
only if single hop links are not available. We observe that two
hops are rarely used to establish a lightpath.
Figure 4 illustrates the recovery performance of affected
traffic after single link failure for all three scenarios. The
recovery of affected traffic in case of scenario I is less
compared to scenario II and scenario III. Though blocking
performance in scenario II and scenario III are similar, the
recovery of affected traffic in case of scenario III is better than
scenario II and scenario I as more flexible routes are available
given the same number of resources in all the three cases.
Fig. 2. 8 by 8 example mesh network used for simulation
ICTON 2011 Tu.D5.4
2
B. Case II
In this case 4 x 4 transceivers with tracking capability are
assumed in each node. Figure 5 illustrates the blocking
performance. Similar results as that of case I can be observed.
Scenarios II and III show better blocking performance
compared to scenario I. Similarly, Figure 6 illustrates the
recovery performance of the affected traffic. The recovery of
affected traffic in scenario III is better compared to scenario I
and scenario II.
IV. CONCLUSION
This paper presents a novel scheme that uses tranceivers
capable of tracking along different directions to reduce the
blocking and increase the recovery of failed traffic. The
performance of the different schemes are evaluated and
compared using different scenarios. OPNET-based
simulations reveal that, if we introduce tracking along
different directions with the number of transmitters and
receivers remainig the same in all the scenarios, the blocking
probability can be reduced considerably and a larger
percentage of failed traffic can be recovered.
[attachment=22587]
Abstract
This paper proposes a transceiver system for
automatic tracking and dynamic routing for free space optical
(FSO) communication. The proposed transceiver architecture
has M transmitters and M receivers in a M x M configuration
that can dynamically orient themselves at different angles to
establish a link between two transceivers. A request to setup a
lightpath is blocked if a free wavelength is not available at a
transmitter or a receiver in a node along the route. The number
of resources remaining the same, in comparison to a fixed FSO,
the methods we propose for routing and recovery that exploit the
M x M transceiver architecture achieve reduced blocking
probability as well as increased percentage of recovery of
affected traffic after a link failure .
Index Terms—Free space optical (FSO) communication;
GMPLS; optical network; survivability; dynamic routing .
I. INTRODUCTION
ITH the tremendous growth in bandwidth intensive
dynamic real time traffic, the network architecture is
shifting toward the model that consists of high speed routers
and optical fibers. The requirement of high capital cost for the
fiber to the home service and the strict RF regulations create
the gap to connect these infrastructures to the real users. Free
space optics (FSO) technology that establishes point to point
communication links is a good candidate to exploit the
tremendous capacity of optical fibers for last-mile access,
metro network extensions, and enterprise connectivity. Due to
unregulated bandwidth, FSO provides low cost, low power,
high security and high rates [1]. Generally, the FSO
transceivers remain in static location to avoid any
misalignments to maintain continuous line of sight. The point
to point FSO links provide higher data rate but with limited
coverage. Consequently, the blocking probability increases. In
wireless environments, the channel is highly dynamic, and the
system will likely experience a considerable degradation in
performance. Survivability is also an issue that must be
addressed. The main challenge in FSO network design is to
exploit the available resources efficiently when operating in
wireless environment.
Most of the papers, for example [5, 6], studythe fixed FSO
system. In this paper, we have modified the static architecture
of FSO transceiver and introduced transceiver capabilities that
provide automatic tracking and dynamic routing features. The
blocking performance and the survivability after a single link
failure are studies via simulation.
The rest of the paper is organized as follows: Section II
provides an overview of the system architecture. Section III
presents simulation results and section IV presents
conclusions from our study.
II. SYSTEM ARCHITECTURE
Figure 1 shows three different configurations of tracking
FSO that adapt to dynamic environment to find available links
from source to destination. The architecture uses M x M
transceivers, each with M transmitters and M receivers that
can rotate at different angles to establish a connection.
Assuming the central node to be the source node or the
transmitting station, we consider three different scenarios.
A. Scenario I
As shown in Figure 1a, the nodes are able to track and
switch the traffic to adjacent nodes that are at an angle of 0,
90, 180 or 270. It has capability to switch the traffic at full
rate.
B. Scenario II
As shown in Figure 1b, the nodes can track and switch the
traffic to adjacent nodes that are at an angle of 0, 45, 90, 135,
180, 225, 270, 315 and has capability to switch the traffic at
full rate.
C. Scenario III
Figure 1c shows the architecture in which each node can
switch the traffic to other nodes that are at an angle of 0, 45,
90, 135, 180, 225, 270 and 315 as well as to nodes that are
two hops away. In this case, nodes transmit data at full rate to
the adjacent nodes and at half the full rate to nodes that are
two hops away.
Generalized multi-protocol label switching
(GMPLS),developed in the Internet Engineering Task Force
Auto Tracking System for Free Space Optical
Communications
Rabindra Ghimire and Seshadri Mohan
W
(a) (b) ©
Fig. 1. Architecture for tracking FSO
ICTON 2011 Tu.D5.4
978-1-4577-0882-4/11/$26.00 ©2011 IEEE 1
(IETF) is used as the control plane protocol [2]. Signaling is
assumed to be out of band. GMPLS can be deployed in FSO
where wavelengths are used to establish low level point-topoint
links for the transmission of packets between high
performance routers. In optical internet, GMPLS routers
translate label assignments into corresponding wavelength
assignments and different network nodes communicate with
each other via end-to-end all optical connections known as
lightpaths. GMPLS employs open shortest path first with
traffic engineering (OSPF-TE) as the routing protocol and
resource reservation protocol with traffic engineering (RSVPTE)
as the signaling protocol. OSPF-TE floods the nodes with
network topology information, which each node uses to
compute paths to other nodes in the network, whereas RSVPTE
utilizes the paths computed by nodes to send signaling
messages and establish lightpaths. This paper adapts the
destination initiated routing scheme proposed in [3]. In this
case, OSPF-TE floods the network with summarized
information of network topology. It advertises all links from a
particular node as alive as long as single free transmitter is
available at that node.
If a node receives a request to set up a unidirectional
lightpath, a route is calculated using the information provided
by routing protocol. The signaling session is triggered and it
will initiate RSVP-TE protocol to send path message along the
found route. All signaling are assumed to be out of band.
RSVP-TE forwards a path message towards the destination.
The path message carries Explicit Route (ER) object and
Label Set (LS) object [4]. ER carries the route that path
message follows and LS carries a set of available transceivers
that can be selected to establish a lightpath. The possibility of
optical to electrical and electrical to optical conversion is
assumed if required. If a lightpath cannot be established due to
the unavailability of transceivers, then blocking occurs, and
path error message is sent back to source node. If the path
message arrives at the destination node, it will initiate a resv
message to reserve resources along the reverse of the route
traversed by the path message. If the resv message is unable to
reserve resources at a node as other requests might have
reserved some selected wavelengths in the mean time, then the
call will be blocked due to unavailability of resources. If
blocking occurs, a resv error message is sent towards the
source and a resv tear message towards the destination. If no
blocking occurs, then resv message arrives at the node
indicating that resources have been reserved and a lightpath
has been established between the source and the destination
nodes. In case of link failure, the node close to the failed link
is responsible for detecting the failure and sending notify
message to all other nodes. After receiving notify message, the
source node will look for an available route by initiating
RSVP-TE protocol. Note that the occupied resources affected
by failure are not released to reestablish the affected lightpaths
as that complicates the procedure. The source node attempts
only once to recover the affected lightpath.
III. SIMULATION SCENARIO
We simulate using OPNET the performance of a 8 by 8
mesh tracking FSO network shown in Figure 2. The traffic is
generated uniformly. Any node can be selected as source and
destination node. Shortest path is given higher priority. Any
available transmitter can be selected randomly at the
destination to establish a light path. The arrival of requests for
lightpaths follows a Poisson process with exponential holding
time. Traffic in the network is varied by varying the ratio of
holding time to inter-arrival time. The traffic is uniformly
distributed between every pair of nodes in the network. The
traffic is converted into wavelength level end-to-end lightpath
requests.
A. Case I
In this case, only 2 x 2 transceivers with tracking capability
are assumed in each node. Figure 3 compares the blocking
performance for scenarios I, II and III. Blocking in scenario I
is relatively more compared to scenario II and scenario III. In
scenario I, the tracking is at angles of 0, 90, 180 and 270,
there by limiting the available route towards the destination
compared to scenarios II and III. In scenario II and scenario
III, tracking occurs at angles of 0, 45, 90, 135, 180, 225, 270
and 315. Scenario III can further switch to nodes that are two
hops away but with transmission rate of one half the full rate.
Similar performance is observed for scenario II and scenario
III. In scenario III, since transmission of data is at half the rate
for the links that are two hops away, any route using two hops
long link along a lightpath will be occupied for twice as long
as those lightpaths using only single links. The simulation
parameters are set in such a way that two hop links are used
only if single hop links are not available. We observe that two
hops are rarely used to establish a lightpath.
Figure 4 illustrates the recovery performance of affected
traffic after single link failure for all three scenarios. The
recovery of affected traffic in case of scenario I is less
compared to scenario II and scenario III. Though blocking
performance in scenario II and scenario III are similar, the
recovery of affected traffic in case of scenario III is better than
scenario II and scenario I as more flexible routes are available
given the same number of resources in all the three cases.
Fig. 2. 8 by 8 example mesh network used for simulation
ICTON 2011 Tu.D5.4
2
B. Case II
In this case 4 x 4 transceivers with tracking capability are
assumed in each node. Figure 5 illustrates the blocking
performance. Similar results as that of case I can be observed.
Scenarios II and III show better blocking performance
compared to scenario I. Similarly, Figure 6 illustrates the
recovery performance of the affected traffic. The recovery of
affected traffic in scenario III is better compared to scenario I
and scenario II.
IV. CONCLUSION
This paper presents a novel scheme that uses tranceivers
capable of tracking along different directions to reduce the
blocking and increase the recovery of failed traffic. The
performance of the different schemes are evaluated and
compared using different scenarios. OPNET-based
simulations reveal that, if we introduce tracking along
different directions with the number of transmitters and
receivers remainig the same in all the scenarios, the blocking
probability can be reduced considerably and a larger
percentage of failed traffic can be recovered.