Вы находитесь на странице: 1из 5

JOURNAL OF COMPUTING, VOLUME 3, ISSUE 12, DECEMBER 2011, ISSN 2151-9617 https://sites.google.com/site/journalofcomputing WWW.JOURNALOFCOMPUTING.

ORG

38

SSMR: A scalable Multipath Routing Protocol for Mobile Ad Hoc Networks


MohammadHasan Farzin, Sam Jabbehdari and Alireza Bagheri
Abstract Mobile Ad Hoc networks (MANET) are networks which are formed by mobile nodes and do not rely on any predefined and fixed infrastructure. Due to high mobility of this kind of network, network connectivity graph changes continuously, and consequently some paths fail. Such path failure has a significant effect on quality of service. One method to increase reliability is to use multiple support paths so that in case any path fails, other paths can be used. One of such protocols is SMR, which increases reliability by creating several disjoint link between the source node and the destination node. However, since this protocol is based on the source routing protocol DSR, as the size of the network increases, its efficiency decreases. In this paper, a mechanism is proposed to reduce overhead and maintain efficiency of this protocol when the size of the network increases. Index Terms Reactive Routing, Multipath, SMR.

1 INTRODUCTION

obile Ad Hoc Networks (MANET) are collections of multi hop wireless mobile nodes that communicate with each other without centralized control or established infrastructure [1]. In such networks, nodes have not a fixed position and move freely; therefore, failures are very common. Such failures result in repetition of route discovery, and delay in sending data packets, which will affect the quality of service [2]. Therefore, Routing in MANET is a critical task due to highly dynamic environment. A proper method to solve this problem is to use multiple paths between source and destination. Most of conventional protocols create only one path between source and destination [3], and therefore, if this path fails, route discovery must be repeated. This results in loss of network's resources, and delay in sending of packets, because a new path must be established before a new connection is reestablished. By creating multiple paths between source and destination, if a path becomes unavailable, then can use one of the support paths promptly. These paths should overlap minimally, because all those paths will become invalid, if a link belonging to multiple paths is broken out. Paths which have no common link are called disjoint links. Multipath routing can be useful in improving the effective bandwidth of communication pairs, responding to congestion and burst traffic, and increasing delivery reliability [4]. One of such multipath routing protocols is SMR [5]. This protocol finds the maximum disjoint paths between source and desination nodes. However, since this proto-

col is based on the source routing protocol DSR [6], it cannot scale well to large networks because source routing requires large data packets to carry full path information to the destination. In large networks, size of data packets become prohibitively high due to the longer paths they carry [7]. Source Routing causes the source node put address of all middle nodes at the header of all data packets. When network grows up, the number of addresses existing in the head is also increased. The sizes of packets are kept small in MANETs, for reasons such as low power of nodes, reliability, keeping the probability of damage resulting from loss of packets low, etc [8]. So, large headers reduce efficiency of the protocol. In this paper, a mechanism is proposed to the SMR protocol, scaleable, and more efficient. SMORT protocol is another protocol that is also scalable. This protocol reduces the routing overhead incurred in recovering from route breaks, by using secondary paths. SMORT computes fail-safe multiple paths, which provide all the intermediate nodes on the primary path with multiple routes (if exists) to destination [7].

2 THE SMR PROTOCOL

Lee et al [5] proposed a routing scheme to find paths that have the least overlap. This protocol is proposed based on DSR protocol. Disjoint multipath protocol of SMR which falls within demand based routing protocols is in fact a real time protocol which has added option of detection of multiple paths to DSR protocol. The main goal of SMR is to build maximally disjoint multiple paths [5]. To detect such paths, all paths to destination must be reported to destination so that the destination can choose among MohammadHasan Farzin Department of Computer Engineering, North them the least overlapping paths. To this end, destination Tehran Branch, Islamic Azad University, Tehran, Iran must know all the parts of the available path, For this Sam Jabbehdari Department of Computer Engineering, North Tehran reason, source routing method is used so that all routing Branch, Islamic Azad University, Tehran, Iran Alireza Bagheri Department of Computer Engineering Faculty in Amirka- information is available in the request packets. In this bir University of Technology scheme, middle node cannot respond to path request packets even if they have information about path to des-

JOURNAL OF COMPUTING, VOLUME 3, ISSUE 12, DECEMBER 2011, ISSN 2151-9617 https://sites.google.com/site/journalofcomputing WWW.JOURNALOFCOMPUTING.ORG

39

tination. Path detection process in SMR is the same as it is in DSR, except that some changes have been made to it in order to provide the least overlapping paths. In SMR, path detection process starts by sending path request packet. Request packets contain addresses of source and destination, and a unique row number. Whenever a node other than destination node receives the packet, it will check it out, and in case it is new, that node will add its nodenumber to the packet and redistributes it. Otherwise, if the received request packet is repeated, a method different from DSR will be used, and in this method, the middle node will not dispose of repeated path request packets, and if the said packet is received from a difference link, and the number of its skips is not larger of the number of skips of the first path request packet, then, it will be redistributed in the network, and by doing so, it raises the probability of finding paths with the least overlap. In SMR protocol, destination chooses two paths, which are the least overlapping. One of these paths is the path with the least delay. The reason behind such choice is to minimize delay in real time protocols. When the destination node receives the first path request packet, it records the entire path available in the packet in its memory. Then, it responds by putting this path in it path response packet. This path has the least hop among available paths between source and destination. Next middle node also receive path request packet based on the path enclosed in that packets and then send it to the next node. After this process, node waits for a while before other path request packets are received, and then, it chooses paths that least overlap with the first path. The procedure is as follows: First, the paths with least overlapping with the first path are chosen. If there are two or more such paths, the one with the least hops will be chosen. If also two or more such paths are detected at this stage, the one that has arrived soone to the destination node will be chosen. Because the given paths pass through different nodes, failare of one node may affect only one path only. In this case, when a node has problem with delivering data packets to the next node, it will consider this link as failed, and send the path error packet upwardly. Upon receiving this message, destination will deem the path as invalid, and if there is another path to the destination, it starts to send data via that path. There are two schemes to specify detection time. In first scheme, path detection process happens after any path becomes invalid, and in second scheme, path detection process happens after all paths have become invalid. Simulations show that the second scheme is more efficient.

with this protocol is that it is not scalable. This is because this protocol is based on DSR protocol, and DSR protocol is not scalable because of source routing. Fig 1 shows Format of sent data packets in SMR protocol.
The Source node The Middle nodes The Destination node

Data

Fig. 1. the Format of data packets in SMR protocol

3 PROPOSED METHOD
The SMR protocol finds node-disjoint paths. The problem

For example, if we assume the size of sent packets to be 512 bytes, and the size of each node address 4 bytes, then, if we have 10 middle nodes, 40 bytes will be added to each packet for addresses of the middle nodes. It is clear that as network grows up, the overhead increases and consequently network efficiency decreases. Here, a mechanism is proposed for making SMR protocol scalable, by which efficiency of networks with higher scalability is maintained. In fact, the main cause of nonscalability of smr protocol is that the addresses of the middle nodes are put at the header of all sent packets. If this overhead could somehow be removed, the problem of non-scalability would be resolved to some extent. Here, we call our protocol SSMR (Scalable SMR). SSMR works based on SMR, but it maintains its efficiency in networks with a higher scalability. To this end, in SSMR protocol, a current path table is considered for each node. This table is considered for the purpose of memorizing paths which are currently using this node as a middle node. First, source node broadcast path requect packet throughout the network. The node receiving this packet distributes these source path request packets and repeated path request packets qualified according to the source SMR protocol for these packets to reach the destination. Then, the destination chooses the path of the first received path request packet, and selects the second path with the minimum overlap with this path. Having received the first path request packet, and having chosen the second path, path response packets are sent to the source. Finally, two paths between source and destination are detected. But at the time of sending the path response packet to source node, each middle node records this path in its current path table upon receiving this path response packet. This record contain node numbers of source and destination (which sends response packet), the next middle node (from which the path response packet is received). This record is used as header of the packets when data are sent from the source to the destination. After process of path detection, sending of data from source starts, with the exception that each data packet only contains address of the source and the destination. Middle nodes identify the destination and the next step to destination by considering this header of the packet, and comparing it with their current paths tables, and so they deliver the packet to the next node. The difference between this protocol and source SMR is that, header of this

JOURNAL OF COMPUTING, VOLUME 3, ISSUE 12, DECEMBER 2011, ISSN 2151-9617 https://sites.google.com/site/journalofcomputing WWW.JOURNALOFCOMPUTING.ORG

40

protocol only contains the addresses of source and destination, and doesnt contain addresses of the middle nodes. This reduces the overhead of sent data and so increases routing capability. Fig 2 shows header of sent data packets in the proposed method.
the Source Node the Destination Node Data

THE SIMULATION AND THE EXPERIMENTS RESULTS

Fig. 2. the Format of data packets in the proposed SSMR protocol

The following pseudo-code explain the function of source node and destination nodes, and the middle nodes. The pseudo-code of the source node in the proposed SSMR protocol: When a path to a particular destination is needed, distribute the path request packet among all Wait for returning of path request packets from the destination Upon receiving the first respond packet from the destination, start sending data to the destination via the respective path Upon failure of one of paths, use the remaining path alone Upon failure of both paths, go to stage 1 The pseudo-code of the destination node in the proposed SSMR protocol: Wait for path request packets. Upon receiving path request packets, first reverse its address, and save the reverse address in your memory, then, send the path response packet, and wait for the rest of packets Upon receiving the rest of packets, choose a path with the least overlap with the first path according to your protocol, and send the second response packet The pseudo-code of the middle nodes in the proposed SSMR protocol when receiving a path request packet: If you found your address at the header of the packet, dispose the packet Otherwise, if you have already received the packet, and number of new closed step is lower than the number of previous packet, or you received it via the same path as you had received the previous packet, dispose the packet. Otherwise, add your address to the packet and forward it to your neighbors. The pseudo-code of the middle nodes in the proposed SSMR protocol when received a path response packet: Record the node numbers of the source, the destination and the node number of the node from which you received the packet as the next step in your temporary table. Considering header of the packet, send it to the source

In this section, to conduct simulation tests and analysis of our proposed protocol, we used the simulation environment of NS2. In this simulation, we used nodes with range of 250 m, which are randomly placed at difference points. In an environment with dimensions 1500500 m to 3000500 m. Also, node use IEEE 802.11 protocol in control layer of access to network interface. It should be noted that node randomly move in this environment. the of each simulation was chosen to be 300 sec. Also, CBR traffic was chosen, and the size of packets was chosen to be 256 bytes.

4.1 The Header Size of Packets


Here, we show how increase in the number of middle nodes results in increase in the length of the header of data packets in SMR protocol, while the size is constant in our proposed SSMR protocol. According to the results of simulation, when NS2 simulator uses SSMR protocol, a constant 20 byte space is added to every data packets. But when SMR protocol is used, some space in addition to the above-mentioned 20 byte space is also added to the source size, which increases as the number of middle node increases. In NS2, each middle node is by default 4 byts long.

Fig. 3. Comparison of size of the header of packets of the proposed SSMR protocol with the SMR protocol

According to Fig 3, size of the header of data packets remains unaffected by increasing in the number of middle nodes at 20 bytes. But when SMR method is used, as the number of middle node increases, the size of data packets also increases. As a result of such increase, when size of the network increases, efficiency of the network decreases.

4.2 The End-To-End Delay


Here, the amount of End-To-End delay is calculated. The Fig 4 shows the end-to-end delay of packets.

JOURNAL OF COMPUTING, VOLUME 3, ISSUE 12, DECEMBER 2011, ISSN 2151-9617 https://sites.google.com/site/journalofcomputing WWW.JOURNALOFCOMPUTING.ORG

41

available. But, because SMR protocol is based on DRS protocol, and DSR protocol uses source routing, therefore, as is the case with DRS protocol, the efficiency of SMR protocol reduces as size of networks are increasedy. This is explained by the source routing, which puts the addresse of the middle nodes at the header of data packets. In our proposed mechanism, capability of scalability is added to SMR protocol by removing addresses of the middle nodes from the header of data packets. The results of simulations show SSMR out performs smr in, the endto-end delay and routing overhead and provides scalability for MANETs.

Fig. 4. Comparison of the routing end-to-end delay of the proposed SSMR protocol with SMR protocol

6 FUTURE WORKS
New works can be conducted on the proposed protocol and multipath routing. For example, one can make changes to manner of choice of disjoint links, for example, choice of the link that meet some criteria such as link stability, etc, rather than choice of the first link that has delivered the path request message to the destination. Also, the proposed protocol can be improved by making corrections to path maintenance method, and repairing errors through path failure detector node rather than prompt sending of it to node of source.

Here, the difference between th routing end-to-end delay of the two protocols increases with increase in the number of nodes, which indicates better performance of the proposed protocol in networks with higher scalability.

4.3 The Routing Overhead


Here, the amount of overhead is calculated. This amount of overhead is the ratio of delay to the number of safe packets received by the destination. The Fig 5 shows ration of end-to-end delay of packet to number of difference node.

REFERENCES
[1] S. Taneja, A. Kush, A Survey of Routing Protocols in Mobile Ad Hoc Networks, International Journal of Innovation, Management and Technology, Vol. 1, No. 3, August ISSN: 20100248 , 2010 X. Li, Multipath Routing and QoS Provisioning in Mobile Ad hoc Networks, Submitted for the degree of Doctor of Philosophy, Department of Electronic Engineering Queen Mary, University of London, April 2006 C. E. Perkins, E. M. Royer, Ad hoc On-demand Distance Vetor Routing, Mobile Comp Sys and Apps, -02.pp:85-100, 1999. L. Liyan, W. Muqing, C. Ziqing, S. Jingfang, Analysis and Optimization of Multipath Routing Protocols Based on SMR, 2nd International Conference on Signal Processing Systems, 2010. S. Lee , M. Gerla, Split Multipath Routing with Maximally Disjoint Paths in Ad hoc Networks, IEEE International Conference on Helsinki , Finland , ISBN: 0-7803-7097-1, 2001. H. Qi, Using ACK Replay to Improve Performance of Dynamic Source Routing Protocol for Mobile Ad Hoc Networks, Dissertation Submitted In Fulfillment Of The Requrements For The Degree Of Master Of Computer Science University Malaya, 2010. L. Reddeppa, S.V. Raghavan, SMORT: Scalable multipath ondemand routing for mobile ad hoc networks, Network Systems Laboratory, Department of Computer Science and engineering , Indian Institute of Technology Madras, Chennai 00036, India , October 2005 C. Vuran, F. Akyildiz, Cross-Layer Packet Size Optimization for Wireless Terrestrial, Underwater, and Underground Sensor Networks, Computer Science and Engineering, Department of CSE Conference and Workshop Papers, University of Nebraska Lincoln, 2008

[2]

[3] [4]

[5]

[6] Fig. 5. Comparison of the routing overhead of the proposed SSMR protocol with SMR protocol

Here, the difference the rouing between overheads of the two protocols increase with increase in number of node, which indicates better performance of the proposed protocol in networks with higher scalability.

[7]

[8]

5 CONCLUSION
Today, many multipath routing protocols for MANETs are available. Among them, SMR protocol has been studied more due to its ability to detect more node-disjoint paths. One of the obvious advantages of this protocol is that fact that node-disjoint paths increase reliability, because when a node fails, only one path may become un-

JOURNAL OF COMPUTING, VOLUME 3, ISSUE 12, DECEMBER 2011, ISSN 2151-9617 https://sites.google.com/site/journalofcomputing WWW.JOURNALOFCOMPUTING.ORG

42

MohammadHaan Farzin is now M.Sc student in computer engineering at North Tehran Branch, Islamic Azad University, Iran. Sam Jabbehdari currently working as an assistant professor at the department of Computer Engineering in IAU (Islamic Azad University), North Tehran Branch, in Tehran, since 1993. He received his both B.Sc. and M.S. degrees in Electrical Engineering Telecommunication from K.N.T (Khajeh Nasir Toosi) University of Technology, and IAU, South Tehran branch in Tehran, Iran, in 1988, through 1991 respectively. He was honored Ph.D. degree in Computer Engineering from IAU, Science and Research Branch, Tehran, Iran in 2005. He was Head of Postgraduate Computer Engineering Department IAU-North Tehran Branch during 2008-2010. He also has written Advanced Topics in Computer Networks book in Persian Language (Tehran, Classic, 2009). His current research interests are Scheduling, QoS, MANETs, Wireless Sensor Networks and Grid Computing Systems. Alireza Bagheri received the B.S. and M.S. degrees in computer engineering from Sharif University of Technology (SUT) at Tehran, the Ph.D degree in computer science from Amirkabir University of Technology (AUT) at Tehran. Currently he is an assistant professor in the computer engineering and IT department at Amirkabir University of Technology at Tehran. His research interests include computational geometry, graph drawing and graph algorithms.

Вам также может понравиться