Int. J. Communications, Network and System Sciences, 2011, 4, 219-226
doi:10.4236/ijcns.2011.44026 Published Online April 2011 (http://www.SciRP.org/journal/ijcns)
An Efficient Routing Tree Algorithm for Combined Uplink
and Downlink Subframes on Centralized Scheduling in
IEEE 802.16 Based Wireless Mesh Networks
Ali Behfarnia, Vahid Tabataba Vakili
Department of Electrical Engineering, Iran University of Scien ce and Technology, Tehran, Iran
Received January 24, 2011; revised March 18, 2011; accepted March 20, 2011
Abstract
IEEE 802.16 mesh mode defines routing tree for transmitting data in centralized scheduling but it does not
define any explicit proposal for combining uplink and downlink subframes. Deploying combined uplink and
downlink subframes on the centralized scheduling scheme can be more flexible and utilization is improved.
However, existing interferences among the transmission of neighboring nodes lead to performance reduction
of the network. In this paper, an efficient routing tree algorithm is proposed with combined uplink and
downlink slot allocation on the centralized scheduling scheme which can abate interferences in the network.
This algorithm allows more subscriber stations to transmit concurrently and so improves spatial reuse in the
network. Also, the algorithm uses multi-channel and single channel systems and considers relay model,
smoothing switching frequently between transmitting and receiving in successive time slots and fairness in
the network. Extensive simulation results demonstrate the effectiveness of the proposed method in terms of
scheduling length, link concurrency ratio, network throughput and Channel Utilization Ratio (CUR).
Keywords: IEEE 802.16, Wireless Mesh Network, Centralized Scheduling, Slot Allocation, Routing Tree
1. Introduction
Wireless mesh network (WMN) is one of the attractive
technologies to achieve better services in next generation
wireless networks. The WMNs can be dynamically
self-configured and self-organized, with the nodes within
the mesh network automatically establishing and main-
taining mesh connectivity among themselves. These fea-
tures lead to advantages such as easy network mainte-
nance, robustness, low up-front cost and reliable service
coverage [1]. Also, the WMNs can be integrated with
other networks such as IEEE 802.11, IEEE 802.16, ad
hoc networks, cellular networks, etc.
Medium access control (MAC) layer in the IEEE
802.16 protocol stack supports two operation modes:
PMP (point-to-multi-point) and mesh (multipoint-to-
multipoint) [2]. All subscriber stations (SSs) in the PMP
mode are managed by base station (BS) and each SS has
to communicate with the BS before transmitting data to
other SSs. In contrast, all mesh SSs (MSSs) within the
network usually communicate with each other without
coordinating with the mesh BS (MBS).
IEEE 802.16 mesh mode employs time division multi-
ple access (TDMA) as the channel access method, where
each radio channel includes multiple frames. A frame
can be divided into time slots which can be assigned to
each SS and the BS. There are two scheduling schemes
for allocating these time slots to all SSs within the mesh
network. One of them is centralized scheduling and the
other is distributed scheduling. In the centralized scheme,
all traffic is to or from the BS and so this scheme is pref-
erable for transmitting internet traffic. All traffic will be
transmitted between SSs in the distributed scheduling
scheme and so this scheme is preferable for transmitting
intranet traffic. In both centralized and distributed schemes,
spatial reuse can be exploited for improvement of wire-
less resource and network throughput. The spatial reuse
in TDMA means that multiple links or nodes can transmit
in a time slot simultaneously only if the transmissions do
not lead to collision. In what fallows, we will focus on
the centralized scheduling in IEEE 802.16 mesh mode.
Several works have been tried to improve network
throughput and spatial reuse. Han in [3,4] proposed
scheduling algorithm for uplink traffic by using relay

220 A. BEHFARNIA ET AL.
model and constructing routing tree in centralized mode
but did not consider downlink traffic and multi-channel
system. Liu in [5] designed uniform slot allocation algo-
rithms which exploit combined uplink and downlink slot
allocation based on the centralized scheduling but did not
consider an efficient routing algorithm and multi-channel
system. Authors in [6,7] proposed centralized scheduling
and channel assignment for IEEE 802.16 mesh networks
but did not focus on downlink traffic and efficient rout-
ing algorithm. Sheu in [8] proposed an interference-
aware routing tree algorithm for the centralized schedul-
ing but did not consider combined uplink and downlink
traffic of each node and multi-channel system. Al- He-
myari in [9,10] proposed a cross-layer design between
network layer and MAC layer but did not discuss about
the downlink parameters and combined uplink and down-
link slot allocation.
In this paper, an efficient and interference-aware rout-
ing tree algorithm is proposed which exploits combined
uplink and downlink slot allocation on the centralized
scheduling. The proposed method considers interferences,
downlink packets, uplink packets and node identifier for
determining parent nodes in routing tree construction.
This algorithm can improve spatial reuse and network
throughput. Multi-channel system is used to abate inter-
ference from transmission of all neighboring nodes
within the mesh network. In addition, this scheduling
algorithm considers fairness among the nodes within the
network and moderates switching frequently between
transmitting and receiving in successive time slots. This
algorithm is evaluated through some important metrics
such as scheduling length, channel utilization ratio, net-
work throughput and link concurrency ratio. To the best
of authors’ knowledge, no research work has been pro-
posed to address the routing tree method with consider-
ing uplink and downlink traffic in combined uplink and
downlink slot allocation on the centralized scheduling.
The rest of this paper is organized as follows. In sec-
tion 2, we will introduce the system model in IEEE
802.16 mesh mode. In section 3, we will explain the
proposed algorithm in detail. Section 4 shows the simu-
lation results. Finally, the paper is concluded in section 5.
2. Basic Scheduling Scheme in IEEE 802.16
Mesh Mode
In this section, overview on frame structure of the IEEE
802.16 mesh mode and routing tree are introduced.
2.1. Frame Structure in IEEE 802.16 Mesh Mode
As mentioned above, each radio channel can be divided
into frames. A frame consists of control subframe and
data subframe, as shown in Figure 1. Each control sub-
frame is further divided into network control subframe
and schedule control subframe. The network control
subframe creates and maintains the network by mesh
network entry (MSH-NENT) message and mesh network
configuration (MSH-NCFG) message. These messages
are transmitted periodically in the mesh network. If the
frame has not network control subframe, it has schedule
control subframe. The schedule control subframe con-
sists of centralized scheduling and distributed scheduling
for sharing nodes within the network in a common radio
resource. In centralized scheduling scheme, all SSs send
their recourse requests to the BS by mesh centralized
scheduling (MSH-CSCH) message. The BS receives
these resource requests and determines flow assignments
of all SSs. The SSs are informed about these assignments
through MSH-CSCH. Another message is called mesh
centralized scheduling configuration (MSH-CSCF). This
message is broadcasted by the BS to disseminate infor-
mation about routing tree and whole centralized sched-
uling configuration among all SSs within the mesh net-
work. The distributed scheduling scheme can be classi-
fied into coordinated and uncoordinated. Each SS in the
coordinated distributed scheduling must coordinate itself
by its two hops neighbor for a collision free data trans-
mission. But, schedules are determined by directed re-
quests and grants between each two SS in uncoordinated
distributed (like ad-hoc networks). Both coordinated and
uncoordinated schemes employ mesh distributed sched-
uling (MSH-DSCH) message to convey resource re-
quests and grants to the neighbors [2].
2.2. Routing Tree
To construct the routing tree, if a new station wants to
join the mesh network, it must scan the MSH-NCFG
message to synchronize with the network. Then, the new
station selects a parent node among all possible neigh-
bors according to specific policy (which will be dis-
cussed in next section). After that, the new station trans-
mits a MSH-NENT message with registration informa-
tion to the MBS via its parent node. After receiving the
registration message of new station, the MBS adds the
new station to the network as a child of determined parent
Figure 1. Frame structure for IEEE 802.16 based WMNs.

A. BEHFARNIA ET AL.
221
node.
In the routing tree construction procedure, the hop-
count of the MBS is assumed to be zero and all neigh-
bors of MBS are assigned to be one. A station selects its
parent node based on smallest hop-count and then
hop-count of it is assigned to be its parent node plus one.
Figure 2(a) gives an example of mesh topology and
routing tree, where the dashed lines show the mesh to-
pology and the solid lines represent the routing tree.
3. Proposed Algorithm
3.1. Motivation and Problem Overview
Several issues can be considered concerning the central-
ized scheduling scheme. The issues can be categorized in
three parts:
• Routing tree: The policy of routing tree construc-
tion influences on system performance.
• Resource management: resource requests from all
SSs must be managed by the BS. Also, the BS
must determine the uplink and downlink flow as-
signments in such a way that overload does not
occur in the network.
• Scheduling algorithm: all SSs should be scheduled
without interferences to achieve higher spatial re-
use and network throughput in the mesh network.
In this work, we will focus on the routing tree and
scheduling algorithm. Resource management is assumed
to be known in the proposed method. In the other words,
we first design a routing algorithm and then schedule the
minislots of data subframe according to the routing tree.
This work can be called the routing and packet sched-
uling (RPS), which has become an important issue in the
centralized scheduling. To illustrate the issue more
clearly, let us consider the example of routing tree in
Figure 2(a). A single channel time division duplex model
is assumed to be adopted in this network and each SS is
assumed has only one packet for uplink transmission and
one packet for downlink transmission (homogeneous).
Figures 2(b)-(c) represents the scheduling matrix (SM)
which is based on routing tree shown in Figure 2(a). The
terms 1♦ and 1● which are shown in the SM are used for
transmitting downlink and uplink traffic, respectively.
Among several criteria to select the nodes for sending
packets, we select the least hop-count to the BS for up-
link transmissions and largest hop-count from the BS for
downlink transmission, because these selections are
demonstrated to be the best in [4,5]. Also, we can find
from the [5] that for combined uplink and downlink slot
allocation (CSLA), the downlink slot allocation is im-
plemented first and then followed by uplink slot alloca-
tion. Thus, it is adopted in our example. As we see from
the Figures 2(b)-(c), the node SS10 which with the largest
hop-count from the BS and largest identifier is transmit-
ted first by the BS. In this condition, nodes SS1and SS3
cannot transmit and receive. Also SS4 and SS6 cannot
transmit with considering interference model. The SS8
can transmit its uplink traffic to SS4 without interference
at the first time slot. If all nodes that are not interfered
with each other have ended their transmissions at a time
slot, the scheduling of that time slot is finished. Thus,
(a)
(b)
(c)
Figure 2. (a) Example of mesh topology and routing tree,
(b)-(c) scheduling matrix based on the routing tree in (a).

222 A. BEHFARNIA ET AL.
scheduling the first time slot is finished. In the next time
slot, node SS2 forwards the downlink traffic of SS10 to
the SS6. In this condition, with considering the interfer-
ence model, nodes SS1, SS3 and SS4 cannot receive the
traffic from other nodes and SS3, SS4, SS7, SS8 and
SS10 cannot transmit to other nodes. Thus only link SS2-
SS6 is scheduled in this time slot. With the same proce-
dure, scheduling matrix is constructed in the Figures
2(b)-(c), where 30 time slot is required to transmit the
traffic of all nodes to their destinations.
Constructing efficient routing tree algorithm for al-
lowing more nodes to transmit simultaneously at a time
slot is the goal of this paper. With considering the con-
struction of routing tree in the network layer and sched-
uling algorithm in the MAC layer, the cross layer con-
cept can be formed in our mind. In the network layer, all
information about routes and interferences of the nodes
within the network are acquired from BS. In the MAC
layer, all centralized scheduling information is sent to the
routing tree algorithm in network layer. After establish-
ing the routes of the nodes, the paths of routing tree are
fed to the MAC layer. With considering the channel as-
signment of multi-channel system in the MAC layer and
implementing centralized scheduling, we can achieve the
reduction of interferences in the network.
3.2. Interference Model and Channel
Assignment
There are two types of interferences in the wireless net-
works in which the nodes use a shared medium to com-
municate with each other. The first type is called primary
interference which occurs when a station has to transmit
or receive at one time. The other type is called secondary
interference which occurs when a station M who intends
to receive from station N is within the range of another
transmitter whose transmissions, though not intended for
M, interfere with the transmissions of N.
For the single channel single transceiver mode, the
following interference model is assumed in the mesh
network:
• Each node’s radio is assumed half duplex, and so
nodes cannot receive and transmit in one time. Al-
so, all nodes cannot receive from multiple nodes or
transmit to multiple nodes in one time.
• All nodes in the range of the transmitter except re-
ceiver cannot receive from other nodes and all
nodes in the range of receiver except transmitter
cannot transmit, because interference occurs in these
conditions.
In respect to the use of single transceiver for each
node in the network, the primary interference cannot be
eliminated. However, by using multi-channel single trans-
ceiver system each node can employ one channel at each
time slot, and also can employ other channel in the other
time slots. Thus, different stations can use different chan-
nel in a time slot and so a lot of nodes can transmit/receive
simultaneously. Hence, with exploiting multi-channel
system the secondary interferences can be eliminated.
For illustrating the channel assignment of the multi-
channel system more lucidly, we can return to the sched-
uling matrix and routing tree which are shown in Figure
2. According to use of the single channel system and
single transceiver, in the second time slot of the schedul-
ing matrix, only SS2 can transmit its packets to SS6.
However, if the multi-channel system was used in this
network with the same condition, the link BS-SS3 would
be activated for downlink transmission and also link
SS5-SS1 and SS9-SS4 would be activated for uplink
transmission.
3.3. Proposed Routing Tree Algorithm Common
The goal of our routing algorithm is to find the best paths
between the static SSs and the BS in the routing tree to-
pology. In this algorithm four parameters are considered;
number of hop-count from the BS, node identification
(i.e. ID node), number of interfering nodes and the num-
ber of uplink and downlink traffic per each node. The
proposed algorithm is implemented based on the follow-
ing steps:
1) Select the parent node based on the least hop-count
from the BS.
2) If there is one node in the transmission range,
choose it as the parent node, else go to step 3.
3) Calculate the number of neighboring nodes that
lead to interference from the selected nodes and then
choose the parent nodes with the least number of
interference.
4) If there is one node with the least number of inter-
ference, choose it as the parent node, else go to step.
5) Calculate the load of the routes which include the
candidate parent nodes. This parameter can be ob-
tained as summation of all uplink packets and down-
link packets of the nodes along the route to the BS.
After that, choose the parent node in the path with
the minimum load.
6) If there is one path, choose corresponding parent
node, else go to step 7.
7) Select the parent node based on minimum ID node.
The ambiguity of the routing tree construction with
this algorithm can be abolished by a simple example.
Let’s return to mesh topology in Figure 2(a) and assume
each node has one packet for uplink transmission and
one packet for downlink transmission. In this condition,
choosing parent nodes in accordance with the proposed

A. BEHFARNIA ET AL.
223
algorithm can be implemented as follows. SS1, SS2 and
SS3 have one upper layer station and so they choose it as
their parent node. SS4 can select SS1 and SS2 as its par-
ent node and so the algorithm will enter to the third step.
With considering the interference number of candidate
nodes, SS4 selects SS1 as its parent node. SS5 has only
one upper level node and so it will select SS1 as the par-
ent node. SS6 and SS7 select SS3 as their parent node
with the similar reasons of SS4 and SS5. SS8 has SS4
and SS6 in its upper layer. These two stations have same
condition in interference number and so the proposed
algorithm will enter to the step 5. But, these two parent
nodes have similar value in their load of routing paths
which are equal to 4 (2 uplink packets plus 2 downlink
packets). In this condition, the algorithm will enter to the
step 7 and so SS8 select SS4 as its parent node. SS9 se-
lect SS5 between SS5 and SS4 as the parent node be-
cause the interference number of SS5 is lower than SS4.
With the same reason SS10 chooses SS7 as its parent
node. Figure 3 shows the proposed routing tree by the
solid lines.
3.4. Centralized Scheduling Scheme
From the resource management, all SSs and the BS are
assigned service token (ST) based on their uplink and
downlink traffic demands. The ST which is assigned to
each SS for uplink or downlink (at the BS) demands can
be obtained as its traffic demands divided by greatest
common divisor of all SSs traffic demands. For example,
if traffic demands of SSs are 2 Mbps, 4 Mbps, 6 Mbps
and 10 Mbps, the STs which are assigned to SSs will be
1, 2, 3 and 5, respectively. With this assignment, the
fairness can be guaranteed because the nodes within the
network will not be starved. Downlink service token ma-
trix (DSTM) and uplink service token matrix (USTM)
can be constructed from downlink and uplink ST de-
mands of all nodes.
Figure 3. The proposed routing tree algorithm.
The link selection criterion for implementing the slot
allocation algorithm is adapted to the CSLA method
which is discussed in [5]. Therefore, we select the least
hop-count to BS for uplink transmissions and largest
hop-count from the BS for downlink transmission. Also,
downlink slot allocation has higher priority than uplink
slot allocation in each time slot. In each link selection
process, transmit node (T_n), receive node (R_n) and
source node (S_n) of each link are obtained. The S_n
refers to the source node of ST and the T_n and R_n are
relay nodes in the route between S_n and its destination.
After each slot allocation, the corresponding matrixes
will be adjusted and SM will be set. Setting the SM is as
follows:
SM (T_n, k) = S_n (1)
Also, adjusting the DSTM and USTM are as follow:
DSTM (T_n, S_n) = DSTM (T_n, S_n) – 1 (2)
DSTM (R_n, S_n) = DSTM (R_n, S_n) + 1 (3)
USTM (T_n, S_n) = USTM (T_n, S_n) – 1 (4)
USTM (R_n, S_n) = USTM (R_n, S_n) + 1 (5)
Relation (1) means that T_n is scheduled at time slot k
in the SM. The service token of corresponding S_n at the
transmitter is diminished by one, and the service token of
corresponding S_n at the receiver is increased by one, as
in (2) and (3). Likewise, the USTM will be adjusted sim-
ilar to DSTM, as in (4) and (5). In this algorithm, the first
input first output (FIFO) is assumed as the buffer of each
node. Also, each receiving node forwards the traffic after
receiving all the traffic from its transmitter node. In the
other words, each receiving node collects all ST of its
transmitter and then transmits the traffic. This procedure
leads to reduce switching frequently between transmit-
ting and receiving nodes in successive time slots. For
more information about switching frequently between
receiving and transmitting within two adjacent time slots,
we refer interested readers to [5].
Figure 4 gives the SM in accordance with the pro-
posed routing tree algorithm which is shown in Figure 3.
Figure 4(a) gives the SM with using single channel sys-
tem and Figure 4(b) gives the SM with using multi-
channel system. In these SMs, we assume that the value
of ST for each node in uplink and downlink direction is
to be one (homogeneous). As we see from these figures,
we can find that 23 and 20 time slots are required to con-
vey the traffic of all nodes to their destinations which is
less than traditional structure which is shown in Figure 2.
4. Simulation
4.1. Performance Metrics
The performance of the proposed algorithm is evaluated

224 A. BEHFARNIA ET AL.
(a)
(b)
Figure 4. Scheduling matrixes of the proposed algorithm
with using (a) single channel system and (b) multi-channel
system.
through scheduling length, normalized network
throughput, channel utilization ratio (CUR) and link
concurrency ratio (LCR). The scheduling length is the
number of time slots which is required to transmit the
traffic of all SSs and the BS to their destinations in a
cycle. The cycle of a link scheduling is the time needed
to transmit all the data traffic under certain traffic model
in the WMN. The scheduling length is one of the preva-
lent criteria in the scheduling algorithm which is consid-
ered in most of existing literature. Normalized network
throughput is employed for evaluating the network per-
formance which can be calculated as the total number of
STs of all nodes divided by scheduling length. The CUR
can be obtained as the number of occupied time slots in
SM divided by number of all possible time slots. All
possible time slots can be obtained as the scheduling
length multiplied by the number of nodes. Transport ca-
pacity of the mesh network can be improved via higher
CUR. The LCR is the average number of nodes that can
transmit data simultaneously at a time slot. This parame-
ter is employed for evaluating the spatial reuse efficiency
in the mesh network.
4.2. Simulation Setup
We compare the proposed algorithm with the CSLA me-
thod in [5] which is proposed for combining uplink and
downlink slot allocation in the centralized scheduling.
Also, the results compare with uplink and downlink slot
allocation which are executed separately.
We use C programming language to evaluate the per-
formance of the proposed algorithm. A square area of
size 50 by 50 units is used for distributing all SSs in the
mesh network randomly and uniformly. The BS is lo-
cated on the center of the square area. A same transmis-
sion range is assumed for all SSs and the BS. Two sta-
tions will be neighbors if the distance between them is
smaller than their transmission range. The transmission
range of each node is assumed to be 10 units. The service
token number of the nodes is randomly selected between
0 and 5 (heterogeneous) and 1 (homogeneous). Simula-
tions are implemented for the range of nodes from 20 to
100 with increment step of 10. Results presented in this
section are averaged among 500 connected graphs.
4.3. Simulation Results
Figures 5-8 gives the scheduling length, channel utiliza-
tion ratio, link concurrency ratio and normalized network
throughput. The meaning of proposed-sig and pro-
posed-mul that are used in these figures refer to use of
single channel and multi-channel systems with the pro-
posed algorithm.
As we see from the Figure 5, the scheduling length
increases with the number of nodes because with in-
creasing the nodes in the network, the traffic demand is
also increased. Moreover, we can find that with use of
multi-channel system the scheduling length is signifi-
cantly improve, and this is because with employing mul-
ti-channel system more subscriber stations can transmit
simultaneously.
From the Figure 6, we can observe that increasing the
number of nodes lead to decrease the CUR because with
increasing the nodes, interfering nodes is also increased.
Also, with exploiting multi-channel system in the algo-
rithm the CUR can significantly improve in compare to
use of single channel system. The reason is due to use of
same channel at the same time slots which lead to incre-
ment the interferences which are sorely affected on si-
multaneous transmission. Moreover, we can find that the
CUR is usually below 10% for the number of nodes
more than 20 because the nodes which are at the first
hop-count from the BS act as bottlenecks in compare to
the nodes with higher number of hop-count which idle in
most of the time.
Figure 7 presents the simulation results of the link
concurrency ratio. We notice that the LCR in the pro-
posed algorithm using multi-channel system does not
ascendant in this figure. This fact stems from the primary
interference. In the other words, with using multi-channel
system and increasing the nodes in the network, the pri-
mary interference occurred more than previously in the
network.

A. BEHFARNIA ET AL.
225
(a)
(b)
Figure 5. Scheduling length. (a) Service token of all nodes


0,5; (b) Service token of all nodes = 1.
(a)
(b)
Figure 6. Channel utilization ratio. (a) Service token of all
nodes


0,5; (b) Service token of all nodes = 1.
(a)
(b)
Figure 7. Link concurrency ratio. (a) Service token of all
nodes


0,5; (b) Service token of all nodes = 1.
(a)
(b)
Figure 8. Normalized network throughp ut. Service token of
all nodes


0,5; (b) Service token of all nodes = 1.

A. BEHFARNIA ET AL.
226
Figure 8 shows the normalized network throughput.
The figure represents that the proposed algorithm with
using multi-channel system is higher throughput than
single channel and the proposed algorithm is better than
CSLA algorithm which is discussed in [5]. The uplink
plus downlink is the worst performance among the
schemes because this scheme deprives of exploiting
combined uplink and downlink transmissions and so the
spatial reuse is lower than other schemes. As expected in
Figure 6, Figure 7 and Figure 8, ST does not influence
on the CUR, LCR and normalized network throughput so
much, because ST cannot independently influence on
these metrics.
5. Conclusions
An efficient routing tree algorithm is proposed for com-
bined uplink and downlink slot allocation on the central-
ized scheduling scheme in IEEE 802.16 based wireless
mesh networks. This algorithm considers interferences,
downlink packets, uplink packets and node identifier for
determining parent node in the routing tree construction.
This algorithm can reduce interference among the nodes
and improve spatial reuse in the network. Also, some
related issues such as fairness, relay model, slot reuse
and moderating switching frequently between transmit-
ting and receiving in the successive time slots have been
considered. This work is compared with combined up-
link and downlink slot allocation (CSLA) that is pro-
posed in [5]. Results confirm that the proposed algorithm
can significantly increase the link concurrency ratio,
network throughput and channel utilization ratio as well
as reducing the scheduling length. In conclusion, the
presented algorithm will provide a simple routing algo-
rithm for combined uplink and downlink slot allocation
on the centralized scheduling in IEEE 802.16 mesh
mode.
6. References
[1] I. F. Akyildiz, X. Wang and W. Wang, “Wireless Mesh
Networks: A Survey,” Elsevier Journal of Computer Net-
works, Vol. 47, No. 4, March 2005, pp. 445-487.
[2] IEEE Standard 802.16-2004, “Local and Metropolitan
Area Networks—Part 16: Air Interface for Fixed Broad-
band Wireless Access Systems,” October 2004.
[3] B. Han, F. P. Tso, L. Lin and W. Jia, “Performance Eva-
luation of Scheduling in IEEE 802.16 Based Wireless
Mesh Networks,” Proceedings of International Conference
on Mobile Adhoc and Sensor Systems, Vancouver, 9-12
October 2006, pp. 789-794.
doi:10.1109/MOBHOC.2006.278652
[4] B. Han, W. J. Jia and L. D. Lin, “Performance Evaluation
of Scheduling in IEEE 802.16 Based Wireless Mesh
Networks,” Elsevier Journal of Computer Communications,
Vol. 30, No. 4, February 2007, pp. 782-792.
[5] S. Liu, S. Feng, W. Ye and H. Zhuang, “Slot Allocation
Algorithms in Centralized Scheduling Scheme for IEEE
802.16 Based Wireless Mesh Networks,” Elsevier Jour-
nal of Computer Communications, Vol. 32, No. 5, March
2009, pp. 943-953.
[6] Y. Tang, Y. Yao and J. Yu, “A Novel Joint Centralized
Scheduling and Channel Assignment Scheme for IEEE
802.16 Mesh Networks,” Proceedings of 4th Interna-
tional Conference on Computer Science & Education,
Nanning, 25-28 July 2009, pp. 289-293.
[7] P. Du, W. Jia, L. Huang and W. Lu, “Centralized Sche-
duling and Channel Assignment in Multi-Channel Single-
Transceiver WiMax Mesh Network,” Proceedings of IEEE
Wireless Com- munications and Networking Conference,
Hong Kong, 11-15 March 2007, pp. 1734-1739.
[8] P. R. Sheu, C. F. Hu, C. C. Liou, F. C. Chuang and Y. C.
Chen, “An Efficient and Interference-Aware Centralized
Routing Tree Algorithm for the Routing and Packet
Scheduling Problem in IEEE 802.16 Mesh Networks,”
Proceedings of International Conference on Communica-
tion and Mobile Computing, Shenzhen, 12-14 April 2010,
pp. 496-503. doi:10.1109/CMC.2010.340
[9] A. Al-Hemyari, C. K. Ng, N. K. Noordin, A. Ismail and S.
Khatun, “Cross Layer Design in 802.16d,” Australian
Journal of Basic and Applied Sciences, Vol. 3, No. 3,
2009, pp. 1591-1600.
[10] A. Al-Hemyari, C. K. Ng, N. K. Noordin, A. Ismail and S.
Khatun, “Constructing Routing Tree for Centralized Sche-
duling Using Multi-Channel Single Transceiver System
in 802.16 Mesh Mode,” Proceedings of IEEE Internatinal
Conference on RF and Microwave, Kuala Lumpur, 2-4
December 2008, pp. 192-196.