Você está na página 1de 4

IEEE 2007 International

Symposium on Microwave, Antenna, Propagation, and EMC Technologies For Wireless Communications

An Improved Routing Protocol in Mobile Ad Hoc Networks


Ting Liu Kai Liu (School of Electronics and Information Engineering, Beijing University ofAeronautics and Astronautics, Beijing 100083, China)
Abstract: A novel routing protocol is proposed to improve the performance of destination-sequenced distance vector (DSDV) routing protocol in mobile ad hoc networks with high mobility. In the improved routing protocol, dissemination of invalid route information is restricted to local area. Meanwhile, a new route reconstruction scheme is used to create a new loop-free route to destination. Simulation results show that compared with DSDV, it can reduce the number of dropped data packets with little increased overhead in mobile ad hoc networks with high mobility. Keywords: mobile ad hoc networks, routing, DSDV

1 Introduction
Mobile ad hoc networks (MANETs) are comprised of mobile nodes that perform multi-hop data packet forwarding over wireless links. The mobility of nodes combined with the transient nature of wireless links results in a rapidly changing network topology. Since no infrastructure exists, it is a great challenge to de sign an effective and adaptive routing protocol. Many restrictions should be well considered, such as limited power and bandwidth. Destination-Sequenced Distance Vector (DSDV) [1] is a typical table-driven routing protocol based on the Distributed Bellman-Ford algorithm for MANETs. In DSDV, each route is tagged with a sequence number originated by the destination to indicate how old the route is. When a route update with a higher sequence number is received, the old route is replaced. In case of different routes with the same sequence number, the route with better metric is used. In DSDV, broken link may be detected by the layer-2 protocol, or it may in stead be inferred if no broadcasts have been received for a while from a former neighboring node. When link breakage happens, any route through the broken

link is assigned an infinite metric and a sequence number one greater than the former one (call an odd sequence number). It is the only situation when se quence number is generated by any node other than the destination. To notify other node the link breakage fast enough, DSDV requires routes with an infinite metric to be advertised immediately, without delay. The major advantage of DSDV is that it provides loop-free routes at all time. However, simulation studies [2] [3] [4] show that its packet delivery ratio is low in MANETs with high mobility, a lot of data packets are dropped due to link breakage. In this paper, an improved protocol for DSDV is proposed to reduce the number of dropped data packets in MANETs with high mobility. When link breaks, it restricts invalid route information to spread in local area, in the mean time, it uses a new route reconstruction scheme to re build a new loop-free route to destination. Simulation results show that compared with DSDV, it has higher packet delivery ratio in MANETs with high mobility.

2 Improved Routing Protocol


In
at
a

node does not mean that it is unreachable for the destination due to latency of route update. For exam ple, when node S finds that its route to node D be comes invalid, maybe it can create a new loop-free route via one of its neighbor which has a valid route that does not pass through node S. To help neighboring node rebuild a loop-free route to destination, node that hears an invalid route update needs to know whether it has a valid route which does not pass through the broadcasting node. In DSDV, routing table maintains next hop to each destination and new route is updated only if either of following two cases happens: (1) new route contains a higher
748-

routing protocols for MANETs, an invalid route

1-4244-1044-4/07/S25.00 2007 IEEE.

IEEE 2007 International

Symposium on Microwave, Antenna, Propagation, and EMC Technologies For Wireless Communications
shorter
sequence number is same, but

Based on the two properties of DSDV above, we propose a new way to judge whether node has a valid route which does not pass through the broadcasting node when it receives an invalid route update. That is, comparing to the in valid route, if node has a valid route that meets rule (1) and either rule (2) or (3), it then can broadcast a route reconstruction message to help the broadcasting node rebuild a new loop-free route to destination: (1) next hop is not the broadcasting node; (2) the route has a higher sequence number; (3) the sequence number is same, and the metric is less than or equal to (Li + 1). Here, Li is the distance between the broadcasting node and the destination before the route becomes invalid. The reason is that, as rule (1) excludes the case that the broadcasting node is the next hop, if we assume that the new route which meets either rule (2) or (3) takes node B as the next hop to destination, then the route stored by B has a higher sequence number or a metric that less than or equal to Li. So it is impossible for B to take the broadcasting node as the next hop to the destination, i.e., the new route does not pass through the broadcasting node. Base on the analysis above, we propose a novel protocol to improve the performance of DSDV in mo bile ad hoc networks with high mobility. In the im proved routing protocol, invalid route is indicated by an additional element other than an odd sequence number and an infinite metric. When link breaks, any route through that broken link is tagged with an inva lid flag, preserving its former sequence number and metric. In the meantime, an invalid route update is triggered to disseminate the information of link breakage and look for a new loop-free route to desti nation in a restricted area. Specifically, the improved routing protocol uses the following approach to re build invalid route when link breaks: Step 1, When node S detects that the link to one of its neighbor D breaks, any route through that broken link is assigned with an invalid flag, and node S triggers a invalid route update which contains the former metric and sequence number of the invalid route. Before the invalid route is rebuilt, in-coming data is buffered. Step 2. Upon receiving an invalid route update, node
new a

sequence

route offers

number; (2)

path.

that equal to or less than (Li + 1), then it broadcasts a route reconstruction message immediately. Otherwise, it does nothing. Here, Li is the metric that the invalid route update contains. Step 4. If node S receives a route reconstruction mes sage, then it reconstructs the invalid route to destina tion via the node which broadcasts the route recon struction message; meanwhile, it broadcasts this mes sage to notify its other neighbors. Step 5. If the route between node S and the destina tion is not repaired, that means, all the neighbors of node S do not hear a new route update from node S, when either of these nodes has incoming data packet through the route which is invalid, then it broadcast this invalid route information to notify its own neighbors and look for a new route to the node D within its one-hop area. By this way, the area of inva lid route information dissemination and route recon struction is enlarged gradually on demand. When link breakage happens, the approach above tries to rebuild a new loop-free route by using a new route reconstruction scheme in restricted area. It takes good use of the existing information in network and prevents invalid route information from affecting the entire network.

compares it to the existing information in its routing table. If its own route takes use of node S as the next hop to destination, then it assigns its route with an in valid flag. Otherwise, it does the step 3. Step 3. Comparing with the invalid route, if node has a valid route with higher sequence number, or a valid route with the same sequence number and a metric

3 Simulation Results
We performed simulation experiments using NS [5J. The network was formed by 100 nodes, which move about over a rectangular (lOOOmx 1000m) flat area. The IEEE 802.11 for wireless LAN, with a channel bandwidth of 2 Mbps, is used as the MAC layer pro tocol. The transmission range of each node is 250m. Each simulation is run for 400 seconds. Random-waypoint was selected as the mobility model. We run our simulations with movement pat-

749-

IEEE 2007 International

Symposium on Microwave, Antenna, Propagation, and EMC Technologies For Wireless Communications
it drops less data packets.
3.2 Average End-to-end Delay

terns

Traffic sources are CBR (Constant Bit Rate). The size of CBR packet is 512 bytes. We fixed the sending rate at 1 packet per second, and the number of sources at 10. All connections were started at times uniformly distributed between 0 and 5 seconds. We chose to evaluate the improved routing protocol (to be simple, call I-DSDV) together with DSDV and DSDV-SQ (another interpretation of DSDV in NS, in which receipt of a new sequence number for a destinalion causes a triggered update). Performance com parison for the three routing protocol is performed according to the following four metrics:
J.l Packet Delivery Ratio

speed is fixed at 20m/s.

100, 150, 200, 250, 300, 350, and 400. The maximum

generated for 9 different pause times (s): 0, 50,

The average end-to-end delay increases with pause time for the three protocols are shown in Figure 2. DSDV and DSDV-SQ exhibit a lower delay than I-DSDV at all time regardless of pause time, and they have almost the same result. As pause time increases, their difference becomes smaller.

Figure 1 shows packet delivery ratio of the three routing protocol, as a function of node pause time.

Fig 2 Comparison of End-to-end Delay


The explanation is that, as I-DSDV uses a new route reconstruction scheme to create a new loop-free route to destination when link breaks, and buffered data packets are retransmitted through the new route, the average end-to-end delay of I-DSDV becomes lower due to the longer transmission time of buffered data packets. As pause time increases, more data packets are buffered due to increased link breakage.
3.3 Routing Overhead

DSDV-SQ
150 00 50 Pause Time(s)

DSDV

I-DSDV

Fig 1 Comparison of Packet Delivery Ratio


It is obvious that I-DSDV has a better performance on packet delivery ratio than DSDV and DSDV-SQ,

especially at higher rate of node mobility (i.e., at large pause time). It delivers a greater percentage of data packets regardless of node mobility rate. Even as the pause time is Os, which represents constant mobility, it delivers almost 88% data packets. On the contrary, DSDV and DSDV-SQ perform poorly. Their packet delivery ratio descends significantly as pause time de creases, dropping to below 63% at pause time Os. In addition, DSDV-SQ has higher packet delivery ratio
than DSDV, which is the same as the result in [2J. The reason is that, as I-DSDV takes good use of the existing valid route information to reconstruct a new loop-free route to destination when link breaks, it can adapt more quickly to the frequent topology changes in MANETs. So comparing to DSDV and DSDV-SQ,

Figure 3 shows the total number of routing packets transmitted during the simulation for the three routing protocols as a function of pause time.
DSDV-SQ
I-DSDV DSDV

150 00 50 Pause Time (s)

Fig 3 Comparison of Routing Overhead


Both of DSDV and DSDV-SQ have almost constant overhead. Similar to the simulation result in literature [2], DSDV-SQ needs much more routing packets than DSDV. On the contrary, I-DSDV exhibits a flexible property on routing overhead. Its routing overhead

750-

IEEE 2007 International

Symposium on Microwave, Antenna, Propagation, and EMC Technologies For Wireless Communications
data packets take a longer way to reach the destination duo to latency of route update.

decreases as pause time increases. It needs about 7000 routing packets (about 28%) more than DSDV when pause time Os, but about 8000 (about 20%) less than DSDV-SQ. It has almost the same overhead with DSDV when there is little node mobility. The constant behavior of DSDV and DSDV-SQ arises as DSDV is a largely periodic routing protocol. Routes are established and reconstructed in periodic updates. On the other hand, I-DSDV tries to rebuild the invalid routes by spreading invalid route in local area when link breaks. So it decreases the frequency of periodic routing update, and reduces the number of routing packets generated in the simulation.
3.4 Path Optimality

4 Conclusion
the DSDV protocol, an improved routing protocol is proposed to reduce the number of dropped data packets in MANETs with high mobility. When link breaks, it tries to create a new loop-free route to destination by using a novel route reconstruction scheme in restricted area. Simulation results show that it has better packet delivery ratio than DSDV in MANETs with high mobility Based
on

difference on path optimality, we compare the difference between the hops that a packet takes to reach its destination and the shortest path length for DSDV-SQ and I-DSDV. For space reasons, only the data when pause time is 200s (at which DSDV-SQ and I-DSDV have almost the same packet delivery ratio 92%) is aggregated. The result is shown in Figure 4. A difference of 0 means the packet takes the shortest path and a differ ence greater than 0 indicates the number of extra hops the packet takes.
no

Since DSDV and

DSDV-SQ has

Acknowledgements
This work was supported by the National Nature Science Foundation of China (No. 10577005, 60532030 and 60502017), the Innovation Foundation of Aerospace Science and Technology of China, and the National Defense Pre-Research Project of "Eleventh Five-Year-Plan" of China ((No.513250403).
Reference

[1] CE. Perkins and P. Bhagwat. Highly dynamic destination sequenced distance vector routing (DSDV) for mobile com puters, in Proc. ACM SIGCOMM 94, London, UK, Oct. 1994,
pp. 234-244. [2] J. Broch, D. A. Maltz, D. B. Johnson, Y. Hu, and J. Jetcheva. A Performance Comparison of Multi-Hop Wireless Ad Hoc Network for mobile ad hoc networks, in Proc. 4th annual

I
Fig 4 Comparison of Path Optimality
I-DSDV in finding the shortest path. There is about 3000 out of 3600 data packets take the shortest path in DSDV-SQ, while about 2800 in I-DSDV. It can be explained that, when old route becomes invalid, I-DSDV tries to create a new loop-free route to destination. When the route is reconstructed, pack ets buffered before was retransmitted through the new route to destination. That means, the retransmitted

DSDV-SQ has a better performance than

IEEE/ACM international conference on Mobile Computing and Networking, Dallas, Texas, U.S. Oct. 1998, pp. 85-97. [3] P. Karavetsios and A.A. Economides. Performance com parison of distributed routing algorithms in ad hoc mobile net works. WSEAS Transactions on Communications, Vol. 3, Issue 1,2004, pp. 317-321. [4] Yi Lu, Weichao Wang, Yuhui Zhong, Bharat Bhargava. Study of Distance Vector Routing Protocols for Mobile Ad Hoc Networks, in Proc. the first IEEE International Conference on Pervasive Computing and Communications (PerCom'03), Fort Worth, Texas, U.S. March 2003, pp. 187-194. [5] The Network Simulator, http://www.isi.edu/nsnam/ns.

751-

Você também pode gostar