Investigation on OLSR Routing Protocol Efficiency JIRI HOSEK 1, KAROL MOLNAR 2 Department of Telecommunications Faculty of Electrical Engineering and Communication, Brno University of Technology Purkynova 118, 612 00 Brno CZECH REPUBLIC hosek@feec.vutbr.cz 1, molnar@feec.vutbr.cz 2 Abstract: Optimized Link State Routing (OLSR) is one of the most often used routing protocols for Mobile Ad hoc NETworks (MANET). MANET networks represent an up-to-date and perspective way of communication with extensive application area. Since the topology of wireless ad hoc networks can change dynamically, the accurate configuration of routing mechanisms is very important. In order to estimate proper settings of the routing protocol parameters a simulation tool can be used. The paper introduces an analytical evaluation of parameters affecting the efficiency of the OLSR routing protocol. The results of this analysis are then used to calculate optimal values for different scenarios. The impact of optimized configuration parameters have been evaluated in the OPNET Modeler simulation environment and the corresponding results are described in the paper. Key-Words: - MANET, OLSR, OPNET Modeler, proactive protocols, reactive protocols, routing, WLAN 1 Introduction Since the topology of Mobile Ad hoc Networks (MANET) can change dynamically, routing is the main challenge in these networks. In addition to route-look-up ad hoc routing protocols have to face to frequent changes in topology, requirements on low transmission power and usage of asymmetric links. In contrast with the infrastructured networks, the routing processes are integrated into each mobile node which will act as both clients and the servers, forwarding and receiving packets to or from other neighbour nodes. The main goal of an ad hoc network routing algorithm is to establish a route correctly and efficiently between a pair of network nodes and provide packet forwarding to deliver messages according to the expected routes [1], [2]. The route establishment should be done with minimum overhead and bandwidth consumption. There have been developed many ad hoc routing protocols which can be divided into two main classes: proactive protocols and reactive on-demand protocols, as discussed in the following text. The proactive routing algorithms, also known as table-driven, aim to keep consistent and up-to-date routing information between every pair of nodes in the network by proactively propagating route updates at fixed time intervals. Usually, each node maintains this information in its table, thus protocols of this class are also called table-driven algorithms. Examples of a proactive protocol are Destination-Sequenced Distance Vector (DSDV), Optimized Link-State Routing (OLSR), and Topology-Based Reverse Path Forwarding (TBRPF) Protocols [1], [3]. An approach different from the proactive routing is the source-initiated on-demand routing. Reactive routing algorithms establish the route to a given destination only when a node requests it by initiating a route discovery process. This process is completed once a route is found or all the possible route permutations have been examined. Once the route has been established, the node considers it active until the destination is no longer accessible, or the route expires. Examples of the reactive protocols are Dynamic Source Routing (DSR) and Ad Hoc On-Demand Distance Vector (AODV) [1], [3]. The main goal of this paper is to present a workflow for the optimization of routing protocol parameters to minimize the overhead traffic and improve time characteristics. Therefore the mathematical analysis and performance comparison of a frequently used proactive routing protocol OLSR is introduced. ISBN: 978-1-61804-030-5 147
2 Optimized Link-State Routing (OLSR) protocol OLSR is a proactive routing protocol for mobile ad hoc networks. The protocol inherits the stability of the link state algorithm and has the advantage of having routes immediately available when needed due to its proactive nature. OLSR minimizes the overhead caused by flooding of control traffic by using only selected nodes, called Multi-Point Relays (MPR), to retransmit control messages [4]. This technique significantly reduces the number of retransmissions required to flood a message to all nodes in the network. Upon receiving an update message, the node determines the routes (sequence of hops) toward its known nodes. Each node selects its MPRs from the set of its neighbours saved in the Neighbour list. The set covers nodes with a distance of two hops. The idea is that whenever the node broadcasts the message, only the nodes included in its MPR set are responsible for broadcasting the message. Furthermore, as OLSR uses the Topology Control (TC) messages for continuous maintain of the routes to all destinations in the network, the protocol is very efficient for traffic patterns where a large subset of nodes is communicating with another large subset of nodes, and where the [source, destination]pairs change over time. The protocol is particularly suited for large and dense networks, as the optimization is done by using MPRs which work well in this context. The larger and more dense a network, the more optimization can be achieved as compared to the classic link state algorithm. OLSR uses hop-by-hop routing, i.e., each node uses its local information to route packets. The more detailed description of the OLSR protocol principles can be found in [3], [4]. 3 Analysis of MANET Routing Protocol Parameters The following mathematical description and analysis help us to optimize the configuration parameters of routing protocols in order to reduce the corresponding control traffic and improve transmission quality. 3.1 Initial Premises In order to provide the mathematical analysis of OLSR and AODV configuration parameters, we determined three initial premises that help us with better description of the behaviour of these routing protocols: maximal speed of node's movement: s = 10ms -1, transmitting range with radius: r = 200m, bit error rate: BER = 10-3. The transmitting range was chosen as an average value in outdoor environment which refers to the 1Mbps data rate for 802.11b technology [5]. For the purpose of simplification, we suppose that the transmitting range has circular shape with the same transmitting power in each direction, see Fig. 1. The bit error rate 10-3 is a typical value for outdoor wireless environment. Fig. 1 Transmitting range of mobile node The necessary prerequisite of a successful communication between two mobile nodes is that both of them are situated in the transmitting range of the second node. In the worst-case scenario, shown in Fig. 2, when the nodes n 1 and n 2 start (t = 0s) to communicate at the same location C (see Fig. 2) and then they are ISBN: 978-1-61804-030-5 148
simultaneously moving straight from each other, we can calculate the maximum duration of communication according to equation (1): r t, (1) s where r is the maximum distance between nodes and s is the speed of their movement. In other words t is the time when a node leaves the transmitting range of the second one and the communication will be terminated. With respect to our initial premises, the value of the time interval t is according to (1): 2r 400 t 20 s. (2) 2s 20 This value indicates that the communication between two moving mobile nodes, see Fig. 2, will not be terminated earlier than after 20s. From a practical point of view time interval t can be considered to be the initial value for the determination of Hello interval and other routing protocol parameters. Fig. 2 Mobile nodes simultaneously moving straight from each other 3.2 Optimization of OLSR parameters Since more intensive packet transmission leads to higher power consumption and quicker depletion of the battery, we focused, during the OLSR optimization, especially on the following four attributes and the effects of their length on the amount of routing traffic. 3.2.1 Hello interval This attribute specifies the time interval between two Hello packets. Hello packets are necessary to maintain adjacencies between nodes. Hello packets carry 1-hop neighbour and 2-hops neighbour information. The main goal of our optimization is to reduce the amount of routing traffic and also minimize a time period t m measured as a time between leaving the transmitting range of the current neighbour node and being registered in the neighbour list of another mobile node. The default value defined in [4] is 2 seconds. According to our assumptions the value of 3 seconds should give the same results with less control traffic because during t = 20s the mobile node sends 6 Hello messages. The seventh Hello message will be received by a new neighbour node in new transmitting range. In the scenario shown in Fig. 2, the time interval t m is 1 second which is acceptable value. 3.2.2 Topology Control interval The Topology Control (TC) interval specifies the time period between Topology Control messages. TC messages are generated by MPR nodes to carry topology/connectivity information. These messages populate the topology table of each node. This information is used for routing table calculations [4], [7]. The default value defined in [4] is 5 seconds. According to our assumptions TC interval should be twice of the Hello interval, which is 6s. This value was defined in terms of the computations similar to the Hello interval. During t = 20s the mobile node sends 3 TC messages. The fourth TC message will be received by a new neighbour node in new transmitting range. ISBN: 978-1-61804-030-5 149
3.2.3 Neighbour hold time interval This parameter specifies the link expiry time. This is typically set to 3 times the value of the Hello interval [4]. With each Hello message arriving, the link expiry timer is reset. If a Hello message is not received during this time, then the link is considered to be lost. If all links to a neighbour node have lost, the neighbour is declared to be unreachable. The packet loss or packet received with an error are the most often problems in the wireless networks. The packet loss is in most cases caused by the mobile node leaving the transmitting range of the second node. Therefore, we made the investigation which event (the packet loss or packet error) comes with higher probability. If we suppose that the average size of a Hello message is about 250 bits and BER is 10-3 then statistically speaking an error can occur in every fourth message. In addition, common Forward Error Correction (FEC) algorithms implemented in mobile network technologies are able to correct at least one bit in the message. Based on our initial premises statistically two consecutive messages with error will not appear more frequently than once in 16 messages. Therefore the probability of leaving the transmission range of the neighbour node and terminating the connection is significantly larger, then the loss of two consecutive messages. According to this statement we set the OLSR neighbour hold time interval to 2 times the Hello interval which is 6s. This setting is preferable to networks with high mobility where it is important to update the adjacencies between nodes frequently. 3.2.4 Topology hold time interval This attribute specifies the expiry time for entries in topology table. This attribute is typically set to 3 times the value of the default TC interval. Topology table entries are refreshed by TC messages based on their originator address and sequence number. During the optimization of this parameter we used the same assumptions as in the case of OLSR Neighbour hold time interval. Therefore we determined that the Topology hold time can be 2 times the value of the optimized TC interval which is 12s. 4 Configuration of Simulation Model The modelling and evaluation of the OLSR routing protocol were realized in the OPNET Modeler simulation environment which is a well-known software tool for design, simulation and analysis of different type of network technologies, architectures and protocols [7]. In OPNET Modeler we created a model of the mobile ad hoc wireless network composed of 72 mobile nodes. The nodes were randomly placed in a 5000 x 5000m campus environment. In order to investigate the effects of the optimized routing parameters, we developed several simulation scenarios, which differ in the configured value of the parameters investigated. 4.1 Wireless Mobility Configuration The mobility and wireless network parameters were identical for all nodes in each scenario. All nodes were configured to move randomly within the defined wireless domain. The speed of each mobile node was defined by the uniform distribution between 0 and 10ms -1. We have set the Distance Threshold, which also defines the transmission range of each mobile node, to 200 meters. The average node pause time was set to 5s. The OPNET Modeler s default wireless network parameters were used for the simulation except the data rate that was decreased to 1Mbps to correspond better to transfer rates in the real MANET environment. 4.2 Traffic Configuration The network traffic was defined between 5 randomly selected pairs of nodes through the IP G711 Voice demand flow model component. This traffic model generates interactive voice flow with bit rate of 120kb/s and average packet size of 120B. The transmission started after 120 seconds from the beginning of the simulation. ISBN: 978-1-61804-030-5 150
4.3 OLSR Simulation Parameters For the OLSR routing protocol, two scenarios were created. The first scenario used the default parameters of OPNET Modeler and the second one with parameters obtained by our analytical approach described in the previous section. Table 1 shows default and optimized simulation parameters for the OLSR routing protocol. Table 1 OLSR simulation parameters OLSR parameters Default value Optimized value Hello interval [s] 2 3 Topology Control interval [s] 5 6 Neighbour hold time [s] 6 6 Topology hold time [s] 15 12 5 Simulation Results and Discussion According to our theoretical assumptions the simulation of both scenarios with duration of 30 minutes was performed. We focused especially on the contribution of the optimized OLSR parameters and also evaluated the end-to-end delay and jitter for a selected pair of mobile nodes. Further, the most important results of our research are described. Fig. 3 shows the total traffic generated by Hello messages in bits per second for the entire network. It is evident that the amount of Hello traffic is lower in the scenario with optimized parameters due to extended Hello interval. The one second extension of the Hello interval reduces the corresponding traffic by 35%. The simulation results also showed that both the amount of Hello traffic and its optimization depend on the density of the mobile ad hoc network. As well as in the case of the Hello interval, we also extended the TC interval by 1 second. The number of TC messages sent in the scenario with optimized parameters was reduced by the 23%. Fig. 3 Hello traffic sent Fig. 4 shows the total amount of routing traffic sent by all nodes in the MANET which includes the Hello messages generated and TC messages sent and forwarded. The initial peak immediately at the beginning of the ISBN: 978-1-61804-030-5 151
simulation indicates the phase when the OLSR builds up the routing tables. This function is provided regardless if a data transmission is required or not. As it is evident from the figure the amount of routing traffic can significantly be influenced by the modification of routing protocols parameters. The optimization factor depends on the number of mobile nodes in the network and their mobility. As a result of the optimization of OLSR parameters, used in the simulation scenario, it was possible to decrease the total routing traffic by 40. Fig. 4 Total routing traffic sent in the network The previous results proved our hypothesis that by extending the period between control messages we can reduce the amount of control traffic. Of course, we also had to examine the influence of the extended periods on the quality of the network services. For this purpose Voice over IP a typical real-time service was selected, which features with a robust quality measurement system. Fig. 5 shows the ETE (End-To-End) packet delay for IP G711 voice service running between two randomly selected nodes. It is evident that modified OLSR parameters did not reduce the ETE delay significantly. From the Fig. 5 it can be seen that the end-to-end delay is less than 35 milliseconds for both scenarios which is an acceptable value for VoIP applications. The reason of such a low value is that OLSR nodes know the route to the destination already before the arrival of the communication request. Similarly to the ETE delay, the improvement of the OLSR protocol has no effect on the jitter. 6 Conclusion The aim of our work was to identify the key factors influencing the efficiency of the proactive OLSR mobile ad hoc routing protocol and optimize the configuration parameters for OLSR with respect to these factors. In the next step we evaluated the performance of this protocol in the OPNET Modeler simulation environment where we created two MANET scenarios; one with default and one with optimized OLSR parameters. Since more intensive packet transmission leads to higher power consumption and quicker depletion of the battery, we focused especially on the Hello and Topology Control intervals and the effects of their length on the amount of routing traffic. Based on the simulation results we can conclude that the extension of the Hello and TC intervals brings significant reduction of total routing traffic generated in the entire mobile ad hoc network. The results also ISBN: 978-1-61804-030-5 152
show that optimized parameters slightly improve the packet end-to-end delay. In general we can conclude that the optimization of the OLSR parameters has no negative effect on the quality of a real-time end-user service. Finally, taking into account the results of our research we can say that the configuration of the routing protocol parameters should consider the network density and the type of network traffic operated. Fig. 5 End-to-end delay for IP G711 voice flow between two randomly selected nodes Acknowledgement: This paper has been supported by the Grant Agency of the Czech Republic (Grant No. GA102/09/1130) and the Ministry of Education, Youth and Sports of the Czech Republic (Project No. MSM0021630513). References: 1. Royer, M. E., Toh, C., A Review of Current Routing Protocols for Ad Hoc Mobile Wireless Networks, IEEE Personal Communications, Vol. 6, No. 1, 1999, pp. 46-55. 2. Illyas, M., The Handbook of Ad Hoc Wireless Networks, CRC Press, 2003. 3. Boukerche, Z., Algorithms and Protocols for Wireless and Mobile Ad Hoc Networks, Wiley, 2009. 4. Clausen, T., Jacquet, P., Optimized Link State Routing Protocol (OLSR) RFC 3626, IETF, 2003. 5. Santi, P., Topology Control in Wireless Ad Hoc and Sensor Networks, Wiley, 2005. 6. Perkins, C., Belding, E., Das, S., Ad hoc On-Demand Distance Vector (AODV) Routing RFC 3561, IETF, 2003. 7. OPNET Technologies, OPNET Modeler Product Documentation Release 16.0, OPNET Technologies Inc., 2010. 8. Qasim, N., Said, F., Aghvami, H., Mobile Ad Hoc Networking Protocols' Evaluation through Simulation for Quality of Service, IAENG International Journal of Computer Science, Vol. 36, No. 1, 2009, pp. 1-9. 9. Gomez, C., Garcia D., Paradells, J., Improving Performance of a Real Ad Hoc Network by Tuning OLSR Parameters, 10th IEEE Symposium on Computers and Communications (ISCC'05), 2005, pp. 16-21. ISBN: 978-1-61804-030-5 153