2009 5th International Colloquium on Signal Processing & Its Applications (CSPA)
High Throughput Routing Algorithm Metric for OLSR Routing Protocol in Wireless Mesh Networks Mohamed ELshaikh1, Nidal Kamel2, Azlan Awang3 UNIVERSITI TEKNOLOGI PETRONAS, Dept. of Electrical & Electronics Engineering Seri Iskandar, 31750 Tronoh, Perak, Malaysia E-mail:
[email protected],
[email protected],
[email protected] metric. Due to the static nature of nodes and unpredictable behavior of wireless link, hop-count based routing protocols are expected to decrease the overall network throughput, and increase the End-to-End delay which results in a low performance network. Moreover, hop count-based protocols prefers path with less number of hop, and does not account to link condition. Thus, it may select a poor link, while better path is available with additional hops. All this can cause a degrading in the network performance. In this paper we aim at improving the performance of WMN by modifying an existing routing protocol. Three commonly used routing protocols are studied and their performances are compared. The main goal of this stage is to study the influence of different routing protocols on WMNs. The comparison is conducted with two network scenarios; a high mobility network and a low mobility network. A new routing technique for WMNs based-on SNR as a new metric for OLSR routing protocol is suggested. The new metric has been implemented with the OLSR routing protocol module using OPNET simulator. The following metrics are used as a performance metric in this simulation.
Abstract- Wireless Mesh Networks (WMN); have been attracting an increasingly intensive research over the last few years. WMNs consist of end clients and mesh routers, communicate wirelessly in a multi-hop fashion. The commonly used routing protocols depend on the hop-count metric to determine the preferred route between source and destination. Due to the unpredictable behavior of the wireless medium, the need for a routing protocol that aware to the medium condition becomes a necessity. In this paper we propose a signal-to-noise ratio (SNR) as a metric to determine the optimum route between source and destination. Optimized Link State Routing Protocol (OLSR) is selected for the new metric. A comparison has been made between the conventional hop-count metric and SNR for the OLSR routing protocol. The network throughput, End-toEnd delay and routing protocol overhead are used as a performance metrics for the comparison. The obtained results show that using SNR as a routing algorithm metric is performing better than the conventional hop-count in WMN environments.
I. INTRODUCTION The major advantages of WMNs over the other wireless networks are the low-cost, self organization, self configuration, last mile internet solution, scalability, and reliability. These advantages have attracted the researcher over the last five years. Nowadays, WMNs is gaining an increased attention from the IEEE community. This led the IEEE organization to emerge a special working group (IEEE 802.11s) in charge of the issues deriving from a completely wireless distribution system used to interconnect different Basic Service Sets (BSSs) through secure and performing links. In a multi-hop networks, like WMN, one of the main factors that influences the performance is the routing protocol. Generally speaking, routing protocols can be classified basedon the routing metric to 1) hop count-based routing protocols, like AODV where the optimum path is defined as the path that goes through the minimum number of nodes, 2) the link quality-based routing protocols, like OLSR where some metrics such as the bandwidth and the packet error rate are considered to define the optimum path to the destination. Routing protocols developed for Ad-hoc networks are not necessary suitable for WMNs because the design of a protocol needs to take advantage of some WMN’s characteristics. Protocols developed so far do not completely make use of the characteristics of WMNs, such as node mobility factor. Most of these protocols are using hop-count as a routing protocol
978-1-4244-4152-5/09/$25.00 ©2009 IEEE
Network throughput: is the amount of successful received data in the destination. End-to-end delay: is the average time to transfer a data packet from a source node to a destination. Protocol overhead: The total number of routing packets transmitted per data packet delivered at destination. Each transmission (hop-wise) of a routing packet was counted as one transmission. Mobility: to measure the affect of different amounts of mobility in a network, two scenarios are implemented with different amounts of mobility. The affect of mobility is measured in terms of network throughput and Endto-End delay. The rest of the paper is organized as follow. Section II goes through the related work. Section III demonstrates the simulation design parameters. Section IV presents the simulation results. Section V concludes the paper. II. RELATED WORK Most of the current research in routing protocols for wireless Mesh networks [4, 5, 6] has focused on finding the suitable routing metric. In [4] the Expected transmission count metric (the average count of transmission is necessary to transfer a packet successfully between two nodes) is proposed. Each node estimates the number of transmissions needed to
445
send a packet through a certain link; by calculating the average of the retransmissions that occurs. Not like SNR, the proposed metric in [4] does not cope well with the high topological change environments, because it uses the mean lose ratio in defining the optimum route, which means it logically depends on the previous retransmission that occurs to estimating the current one. In SNR the link quality is defined with no need for any previous states, which makes it to cope well with short-term channel variations. In [6] they propose a metric that estimates the per-link frame delivery ratios and uses the end-to-end path loss probability as a path cost. This metric does not account for the total bandwidth consumed, because it will prefer two links of low frame loss ratio over a single link with higher loss rate; when link-layer retransmissions are used, the single higher loss link may be able to deliver the packet without as many total transmissions as the two-hop path (ETX is motivated by this observation) [8].
Mobility: Trajectory modules are used to get the node’s mobility. Different trajectories are used. Each trajectory defines the path and the velocity of the moving node. The influence of mobility variation is achieved by designing two scenarios (high and low mobility scenario). IV. RESULTS AND DISCUSSION The results are presented in respect to the performance metrics as follow: Protocol overhead: The overhead of three routing protocols is presented here. Figures (2a,2b,and 2c), show the AODV, DSR, OLSR routing traffic that are generated by the routing protocols respectively; which is considered as the routing protocol traffic overhead. The obtained results show that the routing traffic that generated by the OLSR protocol is less than the DSR and AODV overhead in the simulated scenario. Thus OLSR has the highest performance in WMNs in terms of routing traffic generation (It produces less protocol overhead).
III. SIMULATION DESIGN Network: Four sub networks are created. Each network consists of 25 nodes. The nodes can transmit and receive data from all the nodes, as long as the distance between the two nodes is less than 300m, for each sub-network there is a node with a higher data rate than other nodes. Each sub-network is created in area of 300X300m2. Nodes in the network are spread randomly in the dedicated area.
(a)
Node: Nodes’ physical layer and MAC layer characteristics are designed according to IEEE 802.11 standard nodes. Each node generates two packets per second. Packets size used here are 512 and 1024 bits. The traffic generation model parameters are presented in figure (1).
(b)
(c)
Fig. 2. (a), (b), and (c): AODV, DSR, and OLSR routing traffic in pits/sec respectively.
Network throughput: A comparison between the throughput of three routing protocols (AODV, DSR, and OLSR) is done. Figure (3) demonstrates the obtained results of the collected throughput in the non-mobility scenario. The same comparison is also done for the mobility scenario. The throughput is collected and demonstrated in figure (4). From the obtained result, it appears clearly that DSR has the lowest throughput among the other routing protocols, and OLSR has the highest throughput in this scenario. The low performance of DSR is due to the big number of nodes in the simulated network. However, when the mobility scenario is applied, the DSR has the highest throughput among the other protocols. This demonstrates that, AODV and OLSR routing protocols acts poor in the high mobility networks.
Fig. 1. Traffic generation model parameters.
Protocol: three routing protocols (OLSR, DSR, and AODV) are implemented to the routing layer of each node. The protocols are designed in respect to the RFC standards []. The modified OLSR: SNR value is calculated in the receiver frontend of each node. The OPNET receiver module uses a series of modules (Pipeline stage) to simulate the parameter that affected the received signal. SNR is calculated in the pipeline stage according to two modules of noise the background noise and the accumulated noise modules. The calculated SNR is sent to the OLSR module in respect to received data. OLSR module uses the received SNR value to maintain its routing table accordingly.
446
Fig. 3. AODV,DSR,and OLSR network throughput in (Mb/s) (non mobility scenario), X-axis represents the real time and the Y-axis represents the throughput in (Mb/second).
Fig. 5. OLSR, AODV, DSR, and SNR-OLSR End-to-End delay in Seconds, X-axis represents the routing protocol and the Y-axis represents the ETEdelay in seconds.
SNR-OLSR: In Figure 6, in this graph it appears clearly that the SNR OLSR protocol improves the network throughput. In the original OLSR scenario the network throughput is decreased with both time and mobility. In contrast, the network is recording higher performance when the SNR OLSR protocol is used.
Fig. 4. AODV,DSR,and OLSR network throughput in (Mb/s) (mobility scenario), X-axis represents the real time and the Y-axis represents the throughput in (Mb/second).
End-to-End Delay: The results (figure (5)) show that, by using DSR the packet may need more than one second to reach its destination. AODV can deliver a packet in hundreds of millisecond, but OLSR needs only few milliseconds to deliver its data packet. This is mainly because DSR and AODV are reactive routing protocols, and in reactive protocols there is no exiting route for the destination when it is needed. Thus reactive protocols need to create a request to find a path to the desired destination when there is a demand for it. Reactive protocols use this technique to reduce the generated routing protocol overhead, which might result in a bad delivery time. This technique shows a significant improvement in the high mobility networks (ad hoc, and MANET), but in WMNs where the nodes are stationary; it can be found that, reactive routing protocols are not well suited for WMNs.
Fig. 6. OLSR vs. SNR-OLSR throughput in (Mb/s), X-axis represents the real time and the Y-axis represents the throughput in (Mb/second).
The increase of mobility in the network causes the instability of the wireless link; this results in rapidly change in the transmission quality of nodes. Moreover, in the high mobility network a routing protocol that considers the link condition is expected to gain better performance than the traditional hop-count protocols. Figure 7, depicts the obtained end-to-End delay for both the original and the modified OLSR protocol in a high mobility scenario. Considering the collected results it can be said that, the SNR OLSR is suitable for the high mobility scenarios. Moreover, the OLSR performance is improved in both high and low mobility networks when SNR is used as a metric.
447
the OLSR protocol. SNR is usually calculated in the physical layer of wireless node with no need for additional protocol overhead. The performance of the modified protocol is examined in high and low mobility scenarios. The results show that, the modified protocol is outperforming the original one in terms of throughput and time-delay. REFERENCES [1]
E. Borgia, M. Conti, F. Delmastro, and E. Gregori," Experimental Comparison of Routing and Middleware Solutions for Mobile Ad Hoc Networks: Legacy vs CrossLayer Approach", Pervasive Computing & Networking Lab. (PerLab) IIT Institute CNR via G. Moruzzi,1 56124 Pisa, Italy 2005. [2] B. Aboba, “Architectural implications of link indications. Internetdraft (work in progress)”, The Internet Engineering Task Force, December 2005. [3] Charles E. Perkins, Elizabeth M. Royer, “Ad hoc On-Demand Distance Vector Routing”, Proceedings of the 2nd IEEE orkshop on Mobile Computing Systems and Applications, New Orleans, LA, February 1999, pp. 90-100. [4] D. B. Johnson and D. A. Maltz, “Dynamic Source Routing in AdHoc Wireless Networks”, In T. Imielinski and H. Korth, editors,Mobile Computing, Kluwer, 1996. [5] T. Clausen, P. Jacquet, “Optimized Link State Routing Protocol (OLSR)”, RFC3626, 2003. [6] Jiwei Chen, Yeng-Zhong Lee, Daniela Maniezzo, Mario Gerla;" Performance Comparison of AODV and OFLSR in Wireless Mesh Networks", University of California, Los Angeles, CA 900951596, U.S.A. [7] T. Huovila, P. Lassila, J. Manner, and A. Penttinen, " ABIState of the Art Analysis of Wireless Mesh Technologies 2006", University of Helsinki Helsinki University of Technology 7th November 2006. [8] Gaurav Dawra, " Terrain Based Routing Protocol For sparse ADHOC intermittent network (TRAIN)" , Montana State University Bozeman, Montana,November 2005.
Fig. 7. OLSR and SNR-OLSR End-to-End delay in Seconds, X-axis represents the simulation time in seconds and the Y-axis represents the ETEdelay in seconds.
V. CONCLUSION In this paper a comparison between the AODV, DSR, and OLSR as routing protocols has been conducted. The performances have been tested in two scenarios with different mobility. The OLSR protocol is showing better performance than the two others, in terms of higher throughput and the lower End-to-End delay. Based on that, we consider the OLSR as better suited protocol for the simulated scenarios for WMNs. However, the OLSR is still in need for a metric that consider the link condition. In the second part of this chapter the SNR is proposed as a suitable metric and implemented in
448
