Performance Comparison of MANET Routing Protocols for ns-3 · •Challenges for MANET Routing...
1
II. RESEARCH GOALS • Characteristics of MANETs Mobile ad hoc networks (MANETs) are self-configuring networks with mobile nodes connected by wireless links to form an arbitrary topology without an infrastructure. In MANETs, nodes self-organise and act as both end and intermediate systems. Routing in MANETs are challenged by the mobility of the nodes. Maintaining connectivity with wireless channels and nodes moving out of range from one another is another challenge in MANETs. • Challenges for MANET Routing Simulation Routing protocols are essential to the performance of wireless networks especially in mobile ad-hoc scenarios. The development of new routing protocols requires comparing them against well-known protocols in various simulation environments. Discrete event network simulator ns-3 is proposed as successor of ns-2 and is still under development, providing greater flexibility, evolvability, modularity, and the support of heterogeneity including hybrid wired and wireless models. Despite its advantages, ns-3 is relatively new with limited number of protocol models incorporated into its release distribution; existing built-in MANET routing protocols are limited to the OLSR (optimised link state routing) and AODV (ad hoc on-demand distance vector). We have implemented DSDV (destination-sequenced distance vector) and DSR (dynamic source routing) protocols in ns-3. Additionally, we compare performance of DSDV and DSR against AODV and OLSR. • MANET Routing Protocol Types MANET routing protocols can be classified into two categories based on their update mechanisms: proactive and reactive. Proactive routing protocols maintain updated routing information of all the nodes in the network by periodically distributing routing tables among each other. The advantage is that the routes to any destination are ready to use when needed. However, routing table grows with the size and node density of the network, rather than the number of routes actually needed. The overhead of flooding route advertisements to maintain convergence is a major drawback of proactive protocols. DSDV and OLSR are two major proactive routing protocols. Unlike proactive routing protocols, reactive routing protocols construct routes only when needed for data transmission. When a route to a new destination is requested, the node initiates a route request and waits until the route is discovered. There is no need to distribute their routing information periodically or to maintain routes for all the nodes in the network. The disadvantage is the delay in finding routes to new destinations. DSR and AODV are two well-studied reactive routing protocols. I. MOTIVATION • Yufei Cheng, Egemen K. Çetinkaya, and James P.G. Sterbenz, “Dynamic Source Routing (DSR) Protocol Implementation in ns- 3,” ICST SIMUTools Workshop on ns-3 (WNS3) , Sirmione, Italy, March 2012. • Hemanth Narra, Yufei Cheng, Egemen K. Çetinkaya, Justin P. Rohrer, and James P.G. Sterbenz, “Destination-Sequenced Distance Vector (DSDV) Routing Protocol Implementation in ns- 3,” ICST SIMUTools Workshop on ns-3 (WNS3) , Barcelona, Spain, March 2011, pp. 439–446. • The ns-3 network simulator, http://www.nsnam.org V. REFERENCES III. IMPLEMENTATION of PROTOCOLS in ns-3 IV. RESULTS and ANALYSIS Performance Comparison of MANET Routing Protocols for ns-3 Yufei Cheng, Hemanth Narra, Egemen K. Çetinkaya, Justin P. Rohrer, and James P.G. Sterbenz Electrical Engineering and Computer Science – Information and Telecommunication Technology Center, The University of Kansas – http://www.ittc.ku.edu/resilinets Acknowledgements: This project is supported in part by the National Science Foundation FIND and IFT June 2012 • Implement DSDV and DSR and share these models with the networking research and education community • Carry out simulations to analyse the performance of different routing protocols • Provide baseline performance comparison cases for new protocol design • DSDV Implementation in ns-3 DSDV uses the Bellman-Ford algorithm to calculate paths. The cost metric used is the hop count, which is the number of hops it takes for a packet to reach its destination. DSDV is a table-driven proactive protocol, thus it maintains a routing table with entries for all the nodes in the network and not just the neighbors of a node. The changes are propagated through periodic and trigger update mechanisms used by DSDV. Due to these updates, there is a chance of having routing loops within the network. To eliminate routing loops, each update from the node is tagged with a sequence number, in which an even sequence number represents normal route update whereas an odd sequence number represents route error message. The sequence number from each node is independently chosen but it must be incremented each time a periodic update is made by a node. The sequence number of normal update must be an even number, since each time a periodic update is generated the node increments its sequence number by 2 and adds its update to the routing message it transmits. The node cannot change the sequence number of other nodes. If a node wants to send an update for an expired route to its neighbors, only then it increments the sequence number of the disconnected node by 1. The nodes receiving this update will then look at the sequence number and if it is odd, will remove the corresponding entry from the routing table. Mobility of the nodes in MANETs causes route fluctuations for which DSDV uses settling time to dampen. We have also implemented a buffering mechanism for DSDV to allow fairer comparisons with other MANET routing protocols. The DSDV message header is shown in Figure 1a and the encapsulated DSDV header with IP and UDP is shown in Figure 1b. • DSR Implementation in ns-3 We implemented the DSR routing protocol in ns-3 and the protocol contains fixed-size header and option headers, and they are essentially shim headers between the transport layer and network layer as shown in Figure 2. There are option headers after the fixed-size header and different options are needed for different route operations. Figure 3a shows the route request header and route reply header is shown in Figure 3b. The source route header is used for data packet transmission and is shown in Figure 3c. We have also declared route cache to store all the routes that have been discovered in previous route discovery process. Similarly, we have declared the send buffer class to store all unsent data packets and route request table to avoid duplicate route requests as well as control the rate of consecutive route requests for one destination. The maintenance buffer is used to store the data packet when sent out from the send buffer and waiting for delivery of acknowledgment from the next hop node. Route request table is used to save route request information and it keeps track of route request initiated by the node itself and control the frequency of sending new route request messages. • Comparison of MANET Routing Protocols We perform the simulations over an area of 1500×300 m 2 . All the simulations are averaged over 10 runs with each simulation running for 1000 s; some are averaged over 20 runs to increase confidence. Simulations are performed with 50 nodes, and the source-sink traffic pairs are chosen to be 10, 20 and 30. We perform simulations with a packet size of 64 bytes to exclude the potential network congestion caused by large packets. All the nodes are configured to send 4 packets/s. Using this lower packet rate, we can correctly evaluate the performance of the routing protocols. We use the ns-3 On-Off application to generate CBR (constant bit rate) traffic. The 802.11b MAC is the link layer over the Friis propagation loss model to limit the transmission ranges of nodes. The transmission range of the nodes is set at 250 m for evaluation. To achieve this transmission range, the transmit power is set to be 8.9048 dBm. The mobility model used is steady-state random waypoint with random velocities from 0.01~20.0 m/s and pause times of 0~900 s. Figure 1. DSDV Headers Figure 1b. DSDV Encapsulated with IP Figure 1a. DSDV Message Header DSDV DSDV DSDV UDP IP … sequence number hop count destination address DSR fixed portion DSR options transport protocol payload IP header Figure 2. DSR Header Encapsulated Figure 3a. DSR Route Request Figure 3b. DSR Route Reply Figure 3c. DSR Source Route Figure 3. DSR Headers ⋯ target address address [1] identification option type opt data len address [n] address [1] option data len option type address [n] opt data len L ⋯ address [1] option data len address [n] segs left option type rsv slv ⋯ F L • Performance Measures The performance metrics for the evaluation are packet delivery ratio (PDR), routing overhead, and delay. • Results We compare the packet delivery ratio of existing MANET routing protocols implemented in ns-3 with DSR. From Figure 4, we can see that OLSR outperforms DSR, DSDV, and AODV. While the pause time increases, the PDR for all the protocols increases linearly. The overhead for DSDV and OLSR stays around 200 kb/s and does not change significantly with the pause time increase as shown in Figure 5. As proactive routing protocols, both DSDV and OLSR need to initiate route discovery even when there is no need for the route. The overhead for AODV and DSR decreases as the pause time increases, because the route breakage decreases. Figure 4. PDR with Varying Pause Time Figure 5. Overhead with Varying Pause Time
Transcript of Performance Comparison of MANET Routing Protocols for ns-3 · •Challenges for MANET Routing...
II. RESEARCH GOALS
•Characteristics of MANETs
Mobile ad hoc networks (MANETs) are self-configuring networks with mobile nodes connected by wireless links to form an arbitrary topology without an infrastructure. In MANETs, nodes self-organise and act as both end and intermediate systems. Routing in MANETs are challenged by the mobility of the nodes. Maintaining connectivity with wireless channels and nodes moving out of range from one another is another challenge in MANETs.
•Challenges for MANET Routing Simulation
Routing protocols are essential to the performance of wireless networks especially in mobile ad-hoc scenarios. The development of new routing protocols requires comparing them against well-known protocols in various simulation environments. Discrete event network simulator ns-3 is proposed as successor of ns-2 and is still under development, providing greater flexibility, evolvability, modularity, and the support of heterogeneity including hybrid wired and wireless models. Despite its advantages, ns-3 is relatively new with limited number of protocol models incorporated into its release distribution; existing built-in MANET routing protocols are limited to the OLSR (optimised link state routing) and AODV (ad hoc on-demand distance vector). We have implemented DSDV (destination-sequenced distance vector) and DSR (dynamic source routing) protocols in ns-3. Additionally, we compare performance of DSDV and DSR against AODV and OLSR.
•MANET Routing Protocol Types
MANET routing protocols can be classified into two categories based on their update mechanisms: proactive and reactive. Proactive routing protocols maintain updated routing information of all the nodes in the network by periodically distributing routing tables among each other. The advantage is that the routes to any destination are ready to use when needed. However, routing table grows with the size and node density of the network, rather than the number of routes actually needed. The overhead of flooding route advertisements to maintain convergence is a major drawback of proactive protocols. DSDV and OLSR are two major proactive routing protocols. Unlike proactive routing protocols, reactive routing protocols construct routes only when needed for data transmission. When a route to a new destination is requested, the node initiates a route request and waits until the route is discovered. There is no need to distribute their routing information periodically or to maintain routes for all the nodes in the network. The disadvantage is the delay in finding routes to new destinations. DSR and AODV are two well-studied reactive routing protocols.
I. MOTIVATION
• Yufei Cheng, Egemen K. Çetinkaya, and James P.G. Sterbenz, “Dynamic Source Routing (DSR) Protocol Implementation in ns-3,” ICST SIMUTools Workshop on ns-3 (WNS3) , Sirmione, Italy, March 2012.
•Hemanth Narra, Yufei Cheng, Egemen K. Çetinkaya, Justin P. Rohrer, and James P.G. Sterbenz, “Destination-Sequenced Distance Vector (DSDV) Routing Protocol Implementation in ns-3,” ICST SIMUTools Workshop on ns-3 (WNS3) , Barcelona, Spain, March 2011, pp. 439–446.
• The ns-3 network simulator, http://www.nsnam.org
V. REFERENCES
III. IMPLEMENTATION of PROTOCOLS in ns-3 IV. RESULTS and ANALYSIS
Performance Comparison of MANET Routing Protocols for ns-3 Yufei Cheng, Hemanth Narra, Egemen K. Çetinkaya, Justin P. Rohrer, and James P.G. Sterbenz
Electrical Engineering and Computer Science – Information and Telecommunication Technology Center, The University of Kansas – http://www.ittc.ku.edu/resilinets
Acknowledgements: This project is supported in part by the National Science Foundation FIND and IFT June 2012
• Implement DSDV and DSR and share these models with the networking research and education community
• Carry out simulations to analyse the performance of different routing protocols
• Provide baseline performance comparison cases for new protocol design
•DSDV Implementation in ns-3
DSDV uses the Bellman-Ford algorithm to calculate paths. The cost metric used is the hop count, which is the number of hops it takes for a packet to reach its destination. DSDV is a table-driven proactive protocol, thus it maintains a routing table with entries for all the nodes in the network and not just the neighbors of a node. The changes are propagated through periodic and trigger update mechanisms used by DSDV. Due to these updates, there is a chance of having routing loops within the network. To eliminate routing loops, each update from the node is tagged with a sequence number, in which an even sequence number represents normal route update whereas an odd sequence number represents route error message. The sequence number from each node is independently chosen but it must be incremented each time a periodic update is made by a node. The sequence number of normal update must be an even number, since each time a periodic update is generated the node increments its sequence number by 2 and adds its update to the routing message it transmits. The node cannot change the sequence number of other nodes. If a node wants to send an update for an expired route to its neighbors, only then it increments the sequence number of the disconnected node by 1. The nodes receiving this update will then look at the sequence number and if it is odd, will remove the corresponding entry from the routing table. Mobility of the nodes in MANETs causes route fluctuations for which DSDV uses settling time to dampen. We have also implemented a buffering mechanism for DSDV to allow fairer comparisons with other MANET routing protocols. The DSDV message header is shown in Figure 1a and the encapsulated DSDV header with IP and UDP is shown in Figure 1b.
•DSR Implementation in ns-3
We implemented the DSR routing protocol in ns-3 and the protocol contains fixed-size header and option headers, and they are essentially shim headers between the transport layer and network layer as shown in Figure 2. There are option headers after the fixed-size header and different options are needed for different route operations. Figure 3a shows the route request header and route reply header is shown in Figure 3b. The source route header is used for data packet transmission and is shown in Figure 3c. We have also declared route cache to store all the routes that have been discovered in previous route discovery process. Similarly, we have declared the send buffer class to store all unsent data packets and route request table to avoid duplicate route requests as well as control the rate of consecutive route requests for one destination. The maintenance buffer is used to store the data packet when sent out from the send buffer and waiting for delivery of acknowledgment from the next hop node. Route request table is used to save route request information and it keeps track of route request initiated by the node itself and control the frequency of sending new route request messages.
•Comparison of MANET Routing Protocols
We perform the simulations over an area of 1500×300 m2. All the simulations are averaged over 10 runs with each simulation running for 1000 s; some are averaged over 20 runs to increase confidence. Simulations are performed with 50 nodes, and the source-sink traffic pairs are chosen to be 10, 20 and 30. We perform simulations with a packet size of 64 bytes to exclude the potential network congestion caused by large packets. All the nodes are configured to send 4 packets/s. Using this lower packet rate, we can correctly evaluate the performance of the routing protocols. We use the ns-3 On-Off application to generate CBR (constant bit rate) traffic. The 802.11b MAC is the link layer over the Friis propagation loss model to limit the transmission ranges of nodes. The transmission range of the nodes is set at 250 m for evaluation. To achieve this transmission range, the transmit power is set to be 8.9048 dBm. The mobility model used is steady-state random waypoint with random velocities from 0.01~20.0 m/s and pause times of 0~900 s.
Figure 1. DSDV Headers
Figure 1b. DSDV Encapsulated with IP Figure 1a. DSDV Message Header
DSDV DSDV DSDV UDP IP …
sequence number
hop count
destination address
DSR fixed portion
DSR options
transport protocol
payload
IP header
Figure 2. DSR Header Encapsulated
Figure 3a. DSR Route Request Figure 3b. DSR Route Reply Figure 3c. DSR Source Route
Figure 3. DSR Headers
⋯
target address
address [1]
identification option type opt data len
address [n]
address [1]
option data len option type
address [n]
opt data len L
⋯
address [1]
option data len
address [n]
segs left option type rsv slv
⋯
F L
•Performance Measures
The performance metrics for the evaluation are packet delivery ratio (PDR), routing overhead, and delay.
•Results
We compare the packet delivery ratio of existing MANET routing protocols implemented in ns-3 with DSR. From Figure 4, we can see that OLSR outperforms DSR, DSDV, and AODV. While the pause time increases, the PDR for all the protocols increases linearly.
The overhead for DSDV and OLSR stays around 200 kb/s and does not change significantly with the pause time increase as shown in Figure 5. As proactive routing protocols, both DSDV and OLSR need to initiate route discovery even when there is no need for the route. The overhead for AODV and DSR decreases as the pause time increases, because the route breakage decreases.
Figure 4. PDR with Varying Pause Time
Figure 5. Overhead with Varying Pause Time
