![](https://csdnimg.cn/release/download_crawler_static/15151009/bg2.jpg)
fiber or distribution fiber) at the back-end may lead to huge data
loss (Reaz et al., 2011; Ghazisaidi and Maier, 2009; Guo et al., 2007;
Lim and Nirmalathas, 2010). Therefore, it is necessary to enhance
the survivability of FiWi, especially against the failures at the back-
end. According to the failure severity, we can divide the failures at
the back-end into two levels, i.e., ONU-level failure and OLT-level
failure. ONU-level failure is mainly caused by the single distribution
fiber cut. In the scenario of ONU-level failure, all traffic of the failed
ONU is interrupted. OLT-level failure is mainly caused by single OLT
break or single splitter break or single feeder fiber cut. In the
scenario of OLT-level failure, all traffic of the failed segment is
interrupted.
Recently, some works are carried out to investigate the
survivability of FiWi (Sarkar et al., 2007; Correia et al., 2009;
Schutz and Correia, 2009; Ghazisaidi et al., 2011). Sarkar et al.
(2007) proposed a Risk-and-Delay Aware Routing (RADAR) algo-
rithm to protect FiWi against various failures such as gateway
failure, ONU failure or OLT failure. According to RADAR, a risk list
table is maintained in each wireless router to follow the state of
the wireless paths. Once a wireless path fails, the risk list table
will be updated and the subsequent packets will be rerouted.
Correia et al. (2009) addressed the problem of fault-tolerant
multi-radio FiWi planning. A fault-tolerant approach was pro-
posed including two steps. In the first step, according to the
shortest path criteria, a backup wireless path is computed for
each primary wireless link to tolerate the potential failure of
wireless link due to co-channel interference. In the second step,
aiming to improve network scalability, an Integer Linear
Programming (ILP) model is proposed to optimize the assignment
of channels for the primary and backup wireless links. As an
efficient alternative solution for the complicated ILP model, a
two-phase heuristic algorithm was further proposed to solve the
joint routing and channel assignment problem (Schutz and
Correia, 2009). Ghazisaidi et al. (2011) proposed a probabilistic
model to analyze the survivability of FiWi. Based on the proposed
model, they have demonstrated that FiWi is a cost-efficient
solution for the survivability of next generation PON. They also
proposed a variety of ONU selection schemes to wirelessly update
a subset of ONUs, which have significant effect on the surviva-
bility of FiWi.
All the previous works above consider protecting the back-end
of FiWi by means of the wireless rerouting in the front-end. For
example, when the ONU-level failure occurs, the failed ONU can
reroute the traffic affected by this failure to other available ONUs
through the wireless multi-hop paths between them. However,
the previous works cannot guarantee that there always exist the
available wireless paths between the ONUs. Since the limited
bandwidth of wireless channels, it is probable that the wireless
channels will be blocked when rerouting a lot of traffic between
different ONUs. As a result, the recovery delay increases. Besides,
the previous works cannot tolerate the OLT-level failure well. This
is because different segments may be so far away from each other
that the wireless path is not available for the traffic rerouting
between different segments.
In this paper, we propose a novel protection scheme for the
survivability of FiWi against ONU-level failure and OLT-level
failure. To tolerate the ONU-level failure, we use the wireless
rerouting method to recover the traffic affected by the failure.
Each ONU in the network is allocated a partner ONU. We select
some of the wireless routers and deploy each one with a backup
radio, such that each ONU has an available wireless-backup-path
to its partner ONU. Here, the wireless-backup-path refers to a
wireless path in which each wireless router has a backup radio.
Once the ONU-level fails, the failed ONU can reroute its traffic to
the partner ONU through the wireless-backup-path between
them. Each wireless router in the wireless-backup-path will
transfer the traffic by using the backup radio. Thus, the wireless
channel blocking is mitigated to reduce the delay for traffic
recovery. To tolerate the OLT-level failure
, we first designate one
of the ONUs in each segment as the backup ONU. Then, we cluster
all segments in the network. In each cluster, we place the backup
fibers among the backup ONUs in different segments to establish
the protection ring. Any pair of segments in the same cluster can
backup for each other through the optical-backup-paths on the
protection ring. Here, the optical-backup-path refers to an optical
path composed of multiple backup fibers. When a segment suffers
from the OLT-level failure, the failed segment can transfer its
traffic to other available segments in the same cluster along two
optical-backup-paths on the protection ring simultaneously. Thus,
the delay for recovering traffic is significantly reduced.
Based on the failure scenarios mentioned above, the proposed
protection scheme needs to deal with four failure cases as follows,
Case 1: When the ONU-level failure occurs in single segment,
but the OLT-level failure does not occur, the traffic affected by
the failure will be recovered by means of the backup radios.
Case 2: When the OLT-level failure occurs in single segment,
but the ONU-level failure does not occur, the traffic affected by
the failure will be recovered by means of the backup fibers.
Case 3: When the ONU-level failure and the OLT-level failure
occur in the same segment simultaneously (e.g., the feeder
fiber and one of the distribution fibers in the same segment fail
simultaneously), all ONUs in the failed segment lose their
connections with the OLT. This refers to a special case of the
OLT-level failure. Accordingly, the traffic affected by the failure
will be recovered by means of the backup fibers.
Case 4: When the ONU-level failure and the OLT-level failure
occur in different segments, the traffic affected by the ONU-
level failure and the OLT-level failure will be recovered by
means of the backup radios and the backup fibers,
respectively.
In summary, there are two problems to be solved in the
proposed protection scheme, backup radios deployment and
backup fibers placement. With the objective of minimum backup
radios cost, we propose a heuristic algorithm called Sharing
Backup Radios (SBR) for the deployment of backup radios. In the
SBR algorithm, different wireless-backup-paths are encouraged to
traverse the same wireless router and share the same backup
radio. Therefore, the cost of backup radios is reduced significantly.
With the objective of minimum backup fibers cost, we also
propose another heuristic algorithm called Shortest Protection
Internet
……
splitter
…
…
OLT
ONU
user
end
wireless
router
distribution
fiber
distribution
fiber
feeder
fiber
feeder
fiber
segment-m
segment-n
……
Fig. 1. Illustration of PON-WMN architecture for FiWi network.
Y. Liu et al. / Journal of Network and Computer Applications 36 (2013) 1057–10691058