Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

An Integrated metric routing protocol for

multi-channel wireless networks


Yu Xutao He Wenbo
State Key Laboratory of Millimeter Waves, School of information Science & Engineering,
School of Information Science and Engineering Dalian Polytechnic University
Southeast University Dalian, P.R. China
Nanjing, P. R. China e-mail: hewb@dlpu.edu.cn
e-mail: yuxutao@seu.edu.cn

Abstract—Wireless protocols, for IEEE802.11a, provide


multi-channel technology can significantly increase
multiple none-overlapping channels. In this paper, we present network capacity, but also bring new problems, such as
an integrated metric routing protocol for multi-channel
network, called IMR. The metric of this protocol composes channel selection, multi-channel hidden node problems.
the channel status and node status. Simulation results show The traditional the routing protocols for multi-hop wireless
that this protocol improves network average throughput and
decrease end to end delay efficiently. network, such as Ad hoc On-Demand Distance Vector
(AODV) routing protocol and Dynamic Source Routing
Index Terms—multi-channel, routing protocol, wireless
networks
(DSR) protocol, can be used in the multi-channel networks,

Ⅰ. INTRODUCTION but not get optimal performance[3]. These routing

Currently, in most multi-hop wireless networks nodes protocols typically select shortest-hop routes, in which the

use single interface card with single-channel to channel diversity, interface switching cost, and interference

communicate. Physical layer and MAC layer of these along a path is ignored. Thus, the routing protocol for

networks generally use the IEEE 802.11 protocols. IEEE multi-channel is needed.

802.11 offers multiple non-overlapping physical channels The studies on [3-5] designed the routing protocols for

which are separated in frequency. IEEE 802.11b [1] multi-channel wireless networks. Draves et al. have

standard offers 14 options for the physical channel in the proposed a Multi-Radio Link-Quality Source Routing

2.4 GHz, Taking into accounting non-overlapping of (MR-LQSR) protocol that use a new routing metric named

channel, the frequency spacing must be at least 30MHz, the Weighted Cumulative Expected Transmission Time

802.11b can offer 3channles available for use in same (WCETT) for routing. This protocol only adapts to static

region. IEEE 802.11a [2] provides 12 channels, 8 in the networks and the number of interface card must the same

low part of the band for indoor use and 4 in the upper part as number of channel for each node. A scheme joints

for outdoor use. Previous studies indicate that employing channel assignment and routing problem using Integer

the multi-channel technologies in wireless network Linear Programming (ILP) has proposed in [4], and an

significantly improves the network capacity. To efficiently optimal solution can be obtained by LIP. But this scheme

exploit the performance of wireless network with incurs a higher Computational complexity.

multi-channel, the exiting protocols for single-channel In this paper, we propose a traffic-balanced routing

need to be redesigned. protocol for multi-channel wireless network. This protocol


applies to multi-channel, multi-card environment. Analysis
Compared with the single-channel networks, using

978-1-4244-9003-5/10/$26.00 ©2010 IEEE


and simulation results show that the proposed protocol
increases the average network throughput and decreases (2)
the end-to-end delay. In the following, Section Ⅱ
introduces the routing protocol. Section Ⅲ gives analysis Where N is the number of node sending packets (including
and simulation results. Finally, Section Ⅴ concludes this retransmission) during time T, Nf is the number of node
paper. sending data failed during time T.
Ⅱ.ROUTING STRATEGY
(3)
In this section, we describe the details of the routing
protocol named Integrated Metric Routing (IMR) protocol
for multi-channel. Before the detailed description on (4)
routing is given, we introduce the channel section strategy
Where Qin(i) (1≤i≤c) is the length (bytes) of data saved in
simply. In this strategy, each channel has a separate queue
channel i, Qin_sum is the total length (bytes) of all the data
and the switchable interface stays on a channel for at most t
saved in queues.
second, as showed in figure 1. The maximum length for
To determine a route for flow f, a route discovery
each queue is Qmax. The Packet arrived the node will select
process is initiated by the source node. The node will
a queue. If the queue with an interface’s working channel is
broadcasts a Route Request (RREQ) packet with a special
shortest, the packet will select this queue. Else taking into
sequence number over control channel. When an
account the switching cost and queuing cost, the node will
intermediate node ni receives an RREQ, it performs the
select an appropriate queue. All nodes in the network use
routing table update and compute the cost of RREQ. First it
the same channel as control channel.
checks the routing table and establishes the route to source
node if there is no routing to source node. It also checks if
the sequence number of the current RREQ appears in the
RREQs received earlier. If this is the case, it discards this
RREQ to prevent a loop. Then it adds the M of itself to the
path cost in the RREQ. Finally, in routing table, if there
exists a useable routing to destination node, it will reply
with a Route Reply (RREP) to source node and discard the
RREQ. Otherwise it rebroadcasts the RREQ in control
channel. The destination node sends a route reply (RREP)
Fig. 1. Example for channel queues
for every RREQ that it receives. The source node often
This protocol is designed to be an on-demand routing
receives several RREPs and selects the path with the least
protocol, similar to AODV. We designed a special routing
cost.
metric to decrease the average delay of network. Every
time T, nodes update the routing cost M. Ⅲ. PERFORMANCE EVALUATION

(1) In this section, we evaluate the performance of routing


protocol. Nodes are randomly placed in a 500m×500m
Where p is the probability of the node to send data failed, d
square area. We assume nodes in the network are equipped
is the coefficient of queue delay, and α is the weighting
with IEEE 802.11a interface card and the bandwidth of
factor(0 ≤ α ≤ 1). In section Ⅲ,α=0.5 is used.
each orthogonal channel is the same. 12 non-overlapping
channels is available, including 11 data channels and one significantly better than AODV and IMR-2. For example,
control channel. with 40 nodes, the average throughput of IMR-4 is about
We set the data rate of all channels to 54 Mbps, and 12.4Mbps, the AODV is about 11.6 Mbps.
the maximum transmission range is 100m. In simulations, When there are 30 nodes, figure 4 and figure 5 show
FTP flows are setup between randomly selected source the simulation results with different number of flows.
node and destination node. All FTP flows carry 500 Compared with AODV, IMR gets better performance. As
packets/s with packet size is fixed at 1000 bytes that is flow number increases, IMR obtains more significant
4Mpbs for every flow. The other parameters are improvements. This is because more flows lead to more
summarized in table 1. data competition, and IMR offers more channels to avoid
TABLE 1. Parameters Used in Simulations congestion.
Parameter Value The simulation results also show that with more
SlotTime 9μs interface cards per node can obtain better performance.
SIFSTime 16μs
13.5
DIFSTime 34μs
AODV
CWmin 15 IMR-2
13.0 IMR-3

A verage Throughput (M bps)


CWmax 1023 IMR-4

T 1s 12.5

α 0.5
12.0
Qmax 10000bytes
11.5

Metrics used in the simulations are average throughput


11.0
and average end to end delay. Average throughput is 10 20 30 40 50 60
Node number
defined as the aggregate of flows’ average throughputs in
Fig. 2. Average throughput versus node number
the steady state. The average end to end delay is defined as
the time between the packet send from the source node and
received by the destination node successfully. The duration 0.30
of the entire process mainly includes the time required for AODV
IMR-2
0.28
packet queuing time in the cache, link layer retransmission IMR-3
IMR-4
0.26
time, as well as the propagation time.
End-to-End Delay (s)

0.24
Figure 2 and figure 3 show that the average throughput
0.22
and end-to-end delay of AODV and IMR with different
number of interfaces per node. When there are 4 FTP flows, 0.20

the performances of IMR are better than AODV and the 0.18

IMR with greater number of wireless interface get better 0.16

performance. The performance of IMR with 2 interface 10 20 30 40 50 60

card is slighter better than AODV. The reason is that in Node Number

IMR protocol one channel is selected as control channel so Fig. 3. End to End delay versus node number
that the IMR with 2 interface card is similar with have one
data. The performances IMR with 3 or 4 interface card are
[2] IEEE 802.11 Working Group, Part 11:Wireless LAN Medium
Access Control (MAC) and Physical Layer (PHY) specifications:
High-speed Physical Layer in the5Ghz band , IEEE Standard
16 AODV 802.11a, 1999.
IMR-2
IMR-3 [3] R. Draves, J. Padhye, B. Zill, “Routing in multi-radio multi-hop
Average Throughput (Mpbs)

14
IMR-4 wireless mesh networks,” in Proc. of ACM Mobicom 2004,
pp.114-128.
12
[4] A. H. Mohsenian-Rad, W. S. Wong, “Joint logical topology design,
10 interface assignment, channel allocation, and routing for
multi-channel wireless mesh networks,” IEEE Transactions on
8 Wireless Communications, vol. 6, no. 12, 2007, pp. 4432-4440.
[5] Hon Sun Chiu, Kwan Lawrence Yeung, King-Shan Lui, “J-CAR:
6 An efficient joint channel assignment and routing protocol for
IEEE 802.11-based multi-channel multi-interface mobile Ad Hoc
4 networks,” IEEE Transactions on Wireless Communications
vol.8,no.4, 2009,pp.1706-1715
0 2 4 6
Number of Flows

Fig. 4. Average throughput versus number of flows

0.24 AODV
IMR-2
0.23
IMR-3
0.22 IMR-4
End-to-End Delay (s)

0.21

0.20

0.19

0.18

0.17

0.16

0.15
0 1 2 3 4 5
Number of Flows

Fig. 5. End to End delay versus node number

Ⅴ. CONCLUSION
In this paper, we have proposed channel section
strategy and the routing protocol called IMR for
multi-channel wireless networks. The channel section
strategy and routing metric of IMR not only take into
account the cost of switching delay but also the channel
quality and traffic balance. Simulation results showed that
IMR protocol increases average throughput and reduces
end to end delay.

REFERENCES
[1] IEEE Working Group, Part 11:Wireless LAN medium access
control (MAC) and Physical layer (PHY) specifications :
High-speed physical layer extension in the 2.4 GHz band, IEEE
Standard 802.11b , 1999.

You might also like