Implementation of Epidemic Routing with IP Convergence Layer in ns-3 - - PowerPoint PPT Presentation
Implementation of Epidemic Routing with IP Convergence Layer in ns-3 Dr. Justin P. Rohrer and Lt. Andrew N. Mauldin jprohrer@nps.edu Naval Postgraduate School TaNCAD Lab ( https://tancad.net ) WNS3, June 13, 2018 Abstract Implementation of
jprohrer@nps.edu
Naval Postgraduate School TaNCAD Lab (https://tancad.net)
WNS3, June 13, 2018
Implementation of Epidemic Routing with IP Convergence Layer in ns-3
Paper Abstract [Rohrer and Mauldin, 2018]
We present the Epidemic routing protocol implementation in ns-3. It is a full-featured DTN protocol in that it supports the message abstraction and store-and-haul behavior. We compare the performance of our Epidemic routing ns-3 implementation with the existing implementation of Epidemic in the ONE simulator, and discuss the differences.
www.nps.edu 2 / 44
Introduction Background Prior Work Implementation Evaluation Validation Conclusion
www.nps.edu 3 / 44
A Word About Us...
Justin P. Rohrer:
◮ Assistant Professor, Network Group,
Computer Science Department
◮ Teaches: CS3502 (Networks I), CS4538 (Wireless Security),
CS4554 (Network Modeling), CS4558 (Traffic Analysis)
◮ Leads: Center for Tactical Networked Communications @NPS ◮ Pi/Co-PI on Disruption Tolerant Networking and Network
Measurement Projects
◮ Past contributor to various ns-3 routing and transport models
Lt Andrew Mauldin:
◮ USN, Master’s Student, Computer Science ◮ Thesis: Comparative analysis of disruption tolerant network
routing simulations in the ONE and ns-3 [Mauldin, 2017]; Graduated Fall 2017
www.nps.edu 4 / 44
Naval Postgraduate School (NPS)
◮ Navy’s Research University ◮ Located in Monterey, CA ◮ ≃1500 graduate students: Military officers & DoD civilians
Center for Tactical Networked Communication Architecture
◮ 3 NPS professors, 2 NPS staff ◮ Sponsors: USN, USMC, NSF, NSA, ONR, DARPA, . . .
Focus:
◮ Network Survivability and Resilience ◮ Disruption Tolerant Networks for Tactical Environments
www.nps.edu 5 / 44
Goals for this work
◮ Experiment with new DTN routing primitives ◮ Have been working with the ONE simulator
www.nps.edu 6 / 44
Goals for this work
◮ Experiment with new DTN routing primitives ◮ Have been working with the ONE simulator
◮ Not satisfied with fidelity of results
◮ So we want to work with higher-fidelity simulator like ns-3 ◮ But ns-3 doesn’t currently have any DTN routing protocols ◮ How to start the transition?
www.nps.edu 6 / 44
Goals for this work
◮ Experiment with new DTN routing primitives ◮ Have been working with the ONE simulator
◮ Not satisfied with fidelity of results
◮ So we want to work with higher-fidelity simulator like ns-3 ◮ Epidemic is the simplest DTN protocol ◮ New protocols are benchmarked against Epidemic ◮ Sounds like a good place to start
www.nps.edu 6 / 44
Introduction Background Prior Work Implementation Evaluation Validation Conclusion
www.nps.edu 7 / 44
Epidemic Routing
Epidemic [Vahdat and Becker, 2000]
◮ Simplest possible DTN routing protocol ◮ Flooding based ◮ Every message is forwarded to every node in the network ◮ Very high message-replication overhead due to forwarding n
copies of each message
www.nps.edu 8 / 44
Mobile AdHoc Routing (Supported in ns-3)
Mobile AdHoc Networks (MANETs) allow nodes to multihop messages through the wireless network to their
router, forwarding messages from peers as well as their own messages.
◮ Use a routing protocol to distribute topology information
◮ Distance vector ◮ Link state ◮ Source routing
◮ May be proactive or reactive ◮ Requires end-to-end connectivity ◮ Frequent updates allow the network to tolerate mobility
www.nps.edu 9 / 44
Mobile AdHoc Routing (Supported in ns-3)
Mobile AdHoc Networks (MANETs) allow nodes to multihop messages through the wireless network to their
router, forwarding messages from peers as well as their own messages.
◮ Use a routing protocol to distribute topology information
◮ Distance vector ◮ Link state ◮ Source routing
◮ May be proactive or reactive ◮ Requires end-to-end connectivity ◮ Frequent updates allow the network to tolerate mobility ◮ Significant fraction of available bandwidth used for routing
messages
◮ Highly dynamic topologies may prevent convergence
www.nps.edu 9 / 44
Disruption Tolerant Routing (Not currently offered in ns-3)
Disruption Tolerant Networks allow nodes to multihop messages while tolerating disruptions in connectivity among its
channel connectivity, mobility, long & unpredictable delay, energy and power constraints.
◮ Make local forwarding decisions without globally consistent
information
◮ More information = better decisions ◮ Nodes buffer messages until next-hop is available ◮ Requires only hop-by-hop connectivity
www.nps.edu 10 / 44
DTN Network Stack
Internet 1 Internet 2 Internet 3 Network 1 Transport Layer 1 Bundle Protocol
Bundle Protocol Application
Network 1 Network 2 Transport 1 Transport 2 Bundle Protocol Network 2 Network 3 Transport 2 Transport 3 Bundle Protocol Network 3 Transport Layer 3 Bundle Protocol
Bundle Protocol Application
Figure: Bundle Protocol Network Stack. Adapted from [Scott and Burleigh, 2007]
◮ DTN’s use large buffers; less concerned with fastpath ◮ More concerned with flexibility/late binding ◮ Use mechanisms such as Store-and-Haul
(physically move messages closer to destination)
◮ Often implemented as application-overlay
www.nps.edu 11 / 44
DTN Simulators
◮ Abstract away network stack (e.g. the ONE)
◮ No propagation/loss model ◮ No MAC contention ◮ No message segmentation ◮ No network headers ◮ No control message overhead ◮ No transport layer
◮ Message is basic communication unit
◮ May be 10s of MB www.nps.edu 12 / 44
Network Simulators
◮ In this work we specifically consider two simulators,
as we migrate from one to the other ns-3 is a discrete-event simulator with detailed channel, MAC, network, and routing-layer (MANET) models. It is open-source and written in C++. The ONE is a discrete-time simulator designed specifically for DTN protocol simulations. It abstracts away details
and is written in Java.
www.nps.edu 13 / 44
Introduction Background Prior Work Implementation Evaluation Validation Conclusion
www.nps.edu 14 / 44
Epidemic Simulation Models and Limitations
in the ONE
◮ Epidemic is part of ONE distribution ◮ Message-based ◮ No control messages
in ns-3
◮ Implemented [Alenazi et al., 2015]; not distributed with ns-3 ◮ No message support ◮ Summary vector limited to 1 packet ◮ Size of message buffer limited to number of packet IDs that fit
in summary vector packet
www.nps.edu 15 / 44
Introduction Background Prior Work Implementation Evaluation Validation Conclusion
www.nps.edu 16 / 44
Code Structure
IPv4 Routing Protocol Routing Class
+ Recv<Protocol> + RouteInput + RouteOutput + SendDisjointMessages + SendBeacons + SendMessageFromQueue + SendACK Packet Queue Class
+ Enqueue + Dequeue
+ Find
+ FindDisjointMessages + GetSize Queue Entry
+ AddPacket + GetPackets + SetIpv4Header + GetIpv4Header + GetExpireTime + SetExpireTime + SetMessageID + GetMessageID + GetMessageByteSize + GetMessagePacketTotal + GetCurrentPktCnt + GetPacketSize Packet Classes + Serialize + Deserialize + Print
Figure: DTN routing UML Diagram
◮ Derived from IPv4 Routing Protocol ◮ Based on Alenazi et al. code structure
www.nps.edu 17 / 44
Code Structure Cont.
Packet Queue
◮ Max size set in Bytes ◮ Manages list of Queue Entries ◮ A Queue Entry contains all packets belonging to one message ◮ Supports advanced buffer management (not used in Epidemic)
Routing Protocol
◮ Exchanges message vectors with neighboring routers ◮ Uses Message Queue to generate disjoint message set ◮ Sends disjoint messages to neighbor
www.nps.edu 18 / 44
Message Generation
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 64-bit Message Identification Number 16-bit Last Hop 32-bit Total Number of Packets 32-bit Packet Index
Figure: DTN Data Packet Header
◮ New application type (DTNApplication) generates messages ◮ Messages are segmented into packets ◮ Each packet contains a DTNHeader identifying the message to
which it belongs
◮ Messages are forwarded atomically, i.e. partially received
messages are discarded if connection is broken
www.nps.edu 19 / 44
Node Discovery
BEACON REPLY REPLY_BACK Messages
Figure: Epidemic Control Packet Exchange Sequence
◮ Nodes send beacon messages every BeaconInterval [5 s] ◮ Nodes hearing the beacon reply ◮ Originating node chooses one of the replies to reply to and
exchanges summary vectors
◮ Both nodes calculate their disjoint message set and exchange
www.nps.edu 20 / 44
Introduction Background Prior Work Implementation Evaluation Validation Conclusion
www.nps.edu 21 / 44
Mobility: Helsinki
Figure: Helsinki map, adapted from [Keränen et al., 2009] Table: Helsinki Parameters
Parameter Values Simulation Duration 12 hrs Number of Pedestrians 80 Number of Cars 40 Number of Trams 6 Pedestrian Speed 0.5 - 1.5 m/s Car Speed 2.7 - 13.9 m/s Tram Speed 7 - 19 m/s Pedestrian Pause Time 0 - 120 s Car Pause Time 0 - 120 s Tram Pause Time 10 - 30 s Movement Model Seed 1,2,3,4,5,6,7,8
◮ Default mobility model in the ONE simulator
[Keränen et al., 2009]
www.nps.edu 22 / 44
Mobility: Bold Alligator
Figure: Bold Alligator map Table: Bold Alligator Parameters
Parameter Values Simulation Duration 24 hrs Number of Marines 70 Number of Humvees 20 Number of Drones 2 Number of Helicopters 8 Number of LCACs 2 Number of Ships 3 Marine Speed 0.5 - 1.5 m/s Humvee Speed 13 - 22 m/s Drone 11 - 19 m/s LCAC Speed 11 - 19 m/s Helicopter Speed 125 - 167 m/s Ship Speed 1 - 4 m/s Marine Pause Time 0 - 60 s Humvee Pause Time 0 - 60 s Helicopter Pause Time 0 - 1800 s LCAC Pause Time 0 - 1800 s Ship Pause Time 0 s Movement Model Seed 1,2,3,4,5,6,7,8
◮ Replicates real military exercise ◮ Developed by Kevin Killeen for his thesis [Killeen, 2015]
www.nps.edu 23 / 44
Mobility: Omaha
Figure: Omaha Scenario Map Table: Omaha Parameters
Parameter Values Simulation Duration 12 hrs Warmup Time 1000 s Timestamp Resolution 0.1 s Number of Soldiers 44 Number of Ships 17 Soldier Speed 0.5 - 1.5 m/s Ship Speed 8 - 16 m/s Soldier Pause Time 0 - 60 s Ship Pause Time 0 - 300 s Movement Model Seed 1,2,3,4,5,6,7,8
◮ Approximates Omaha Beach landing on June 6, 1944 ◮ As-if DTN comms had been available ◮ New for this work
www.nps.edu 24 / 44
Traffic and Resources: Helsinki
Table: Helsinki Scenario Parameters
Parameter Values Simulation Duration 12 hrs Simulator Seed 719 Warmup Time 1000 s Timestamp Resolution 0.1 s Beacon Interval 5 s Base Radio Bandwidth (Mbps) 6 12 24 36 54 Tram Radio Bandwidth (Mbps) 58.5 117 234 351 526.5 Base Radio Transmit Range 10 m Tram Radio Transmit Range 1000 m Base Buffer Size (MB) 5 10 25 50 100 Tram Buffer Size (MB) 50 100 250 500 1000 Message Rate 1 / 25 - 35 s Message Size 0.5 - 1.0 MB Message TTL 5 hrs Hop Limit 50 Protocols Epidemic, Centroid, GAPR, GAPR2, GAPR2a, Vector
◮ A random node generates a 0.5 to 1.0 MB message to another
random node every 25 to 35 seconds.
www.nps.edu 25 / 44
Traffic and Resources: Bold Alligator
Table: Bold Alligator Scenario Parameters
Parameter Values Simulation Duration 24 hrs Simulator Seed 719 Warmup Time 10800 s Timestamp Resolution 0.1 s Beacon Interval 5 s Base Radio Bandwidth 12, 24, 36, 54 Mbps Humvee Radio Bandwidth 54 Mbps Drone Radio Bandwidth 6 Mbps Ship Radio Bandwidth 54 Mbps Base Radio Transmit Range 100 m Humvee Radio Transmit Range 3000 m Drone Radio Transmit Range 3000 m Ship Radio Transmit Range 10000 m Marine Node Buffer Size (MB) 5 10 25 50 Drone Node Buffer Size (MB) 5 10 25 50 Humvee Buffer Size (MB) 50 100 250 500 LCAC Buffer Size (MB) 50 100 250 500 Helo Buffer Size (MB) 50 100 250 500 Ship Buffer Size (MB) 500 1000 2500 5000 Marine Message Size 250 - 500 KB Humvee Message Size 0.5 - 1.0 MB Ship Message Size 0.5 - 1.0 MB Marine Message Rate 1 / 5 - 10 s Humvee Message Rate 1 / 10 - 20 s Ship Message Rate 1 / 25 - 35 s Message TTL 5 hrs Protocols Epidemic, Centroid, GAPR, GAPR2, GAPR2a, Vector
◮ Message rates are across all nodes of specified type
www.nps.edu 26 / 44
Traffic and Resources: Omaha
Table: Omaha Scenario Parameters
Parameter Values Simulation Duration 12 hrs Simulator Seed 719 Warmup Time 1000 s Timestamp Resolution 0.1 s Beacon Interval 5 s Radio Bandwidth 6, 12, 24, 36, 54 Mbps Radio Transmit Range 550 m Soldier Node Buffer Size (MB) 5 10 25 50 100 Ship Node Buffer Size (MB) 50 100 250 500 1000 Message Rate 1 / 25 - 35 s Message Size 0.5 - 1.0 MB Message TTL 5 hrs Protocols Epidemic, Centroid, GAPR, GAPR2, GAPR2a, Vector
◮ A random node generates a 0.5 to 1.0 MB message to another
random node every 25 to 35 seconds.
www.nps.edu 27 / 44
Metrics
Message delivery ratio (MDR) Fraction of (unique) transmitted messages that reach destination.
destination.
message copy. Message-replication overhead Overhead due to creating multiple copies of messages. Comparable to overhead in the ONE simulator. Network overhead Includes message-replication overhead plus control messages and packet headers
model All metrics averaged over 8 runs with different RNG seeds 8 different mobility trace files per scenario
www.nps.edu 28 / 44
Protocols for Comparison against Epidemic
Motivation for this research is to provide Epidemic as comparison data point for more sophisticated DTN protocols, such as:
◮ Vector [Kang and Kim, 2008] ◮ Centroid [Rohrer, 2018] ◮ GAPR [Rohrer and Killeen, 2016] ◮ GAPR2 [Rohrer and Killeen, 2016] ◮ GAPR2a [Mauldin, 2017]
Given that context, we include results from these protocols for
non-resource-constrained scenarios, and lower bound for performance in resource-constrained scenarios.
www.nps.edu 29 / 44
Message Delivery Ratio
20 40 60 80 100 Buffer Size (MB) 0.0 0.2 0.4 0.6 0.8 1.0 Message Delivery Ratio (MDR)
GAPR2a GAPR Vector No Limit GAPR2 Centroid Vector Epidemic
Figure: Helsinki message delivery ratio, 95% confidence intervals; 54 Mb/s base radio
◮ Radio bandwidth had little/no effect on this metric ◮ Epidemic (as expected) requires large buffers
www.nps.edu 30 / 44
Message Delivery Ratio
10 20 30 40 50 Buffer Size (MB) 0.0 0.2 0.4 0.6 0.8 1.0 Message Delivery Ratio (MDR)
Vector No Limit Centroid GAPR GAPR2a Vector GAPR2 Epidemic
Figure: Bold Alligator message delivery ratio, 95% confidence intervals; 54 Mb/s base radio
◮ Radio bandwidth had little/no effect on this metric ◮ Epidemic (as expected) requires large buffers
www.nps.edu 30 / 44
Message Delivery Ratio
20 40 60 80 100 Buffer Size (MB) 0.0 0.2 0.4 0.6 0.8 1.0 Message Delivery Ratio (MDR)
GAPR2a Vector No Limit GAPR Vector GAPR2 Centroid Epidemic
Figure: Omaha message delivery ratio, 95% confidence intervals; 54 Mb/s base radio
◮ Radio bandwidth had little/no effect on this metric ◮ Epidemic (as expected) requires large buffers
www.nps.edu 30 / 44
Latency
20 40 60 80 100 Buffer Size (MB) 1000 2000 3000 4000 5000 6000 7000 Average Latency (seconds)
Vector GAPR2 Centroid GAPR2a Vector No Limit GAPR Epidemic
Figure: Helsinki message delivery ratio, 95% confidence intervals; 54 Mb/s base radio
◮ Most protocols are able to reduce latency with increased
resources
◮ Epidemic (as expected) requires large buffers
www.nps.edu 31 / 44
Latency
10 20 30 40 50 Buffer Size (MB) 1000 2000 3000 4000 5000 6000 7000 8000 9000 Average Latency (seconds)
GAPR2a GAPR2 GAPR Centroid Vector Vector No Limit Epidemic
Figure: Bold Alligator message delivery ratio, 95% confidence intervals; 54 Mb/s base radio
◮ For several protocols, messages are being delivered later,
rather than discarded.
◮ Epidemic (as expected) requires large buffers
www.nps.edu 31 / 44
Latency
20 40 60 80 100 Buffer Size (MB) 1000 2000 3000 4000 5000 6000 7000 Average Latency (seconds)
Vector GAPR2 GAPR2a GAPR Vector No Limit Centroid Epidemic
Figure: Omaha message delivery ratio, 95% confidence intervals; 54 Mb/s base radio
◮ For several protocols, messages are being delivered later,
rather than discarded.
◮ Epidemic (as expected) requires large buffers
www.nps.edu 31 / 44
Message Replication Overhead; Total Network Overhead
20 40 60 80 100 Buffer Size (MB) 100 200 300 400 500 600 700 800 Message Replication Overhead Ratio
Epidemic Vector No Limit GAPR Centroid GAPR2a Vector GAPR2
Figure: Helsinki replication overhead
20 40 60 80 100 Buffer Size (MB) 50 100 150 200 250 300 Network Overhead Ratio
Epidemic GAPR Vector No Limit GAPR2a Centroid GAPR2 Vector
Figure: Helsinki network overhead
◮ Note difference in y-axis scale ◮ Network overhead is about 10% higher
www.nps.edu 32 / 44
Message Replication Overhead; Total Network Overhead
10 20 30 40 50 Buffer Size (MB) 100 200 300 400 500 Message Replication Overhead Ratio
GAPR Epidemic Vector No Limit GAPR2a Centroid GAPR2 Vector
Figure: Bold Alligator replication
10 20 30 40 50 Buffer Size (MB) 100 200 300 400 500 Network Overhead Ratio
Epidemic GAPR Vector No Limit GAPR2a Centroid GAPR2 Vector
Figure: Bold Alligator network
◮ Network overhead is about 10% higher
www.nps.edu 32 / 44
Message Replication Overhead; Total Network Overhead
20 40 60 80 100 Buffer Size (MB) 500 1000 1500 2000 2500 Message Replication Overhead Ratio
GAPR GAPR2 GAPR2a Epidemic Vector No Limit Centroid Vector
Figure: Omaha replication overhead
20 40 60 80 100 Buffer Size (MB) 200 400 600 800 1000 1200 1400 Network Overhead Ratio
GAPR GAPR2 GAPR2a Epidemic Vector No Limit Centroid Vector
Figure: Omaha network overhead
◮ Note difference in y-axis scale ◮ Network overhead is about 10% higher
www.nps.edu 32 / 44
Power Consumption
20 40 60 80 100 Buffer Size (MB) 0.820 0.822 0.824 0.826 0.828 Average Power Consumed per Node (W)
GAPR Epidemic Vector No Limit GAPR2a Centroid GAPR2 Vector
Figure: Helsinki per-node power cons., 95% confidence intervals; 54 Mb/s base radio
◮ Absolute values dependent on radio hardware, only relative
values of interest
◮ Correlates directly to number of packet transmissions
www.nps.edu 33 / 44
Power Consumption
10 20 30 40 50 Buffer Size (MB) 0.8190 0.8195 0.8200 0.8205 0.8210 0.8215 0.8220 0.8225 0.8230 Average Power Consumed per Node (W)
GAPR GAPR2a Vector No Limit Epidemic Centroid GAPR2 Vector
Figure: Bold Alligator per-node power cons., 95% confidence intervals; 54 Mb/s base radio
◮ Absolute values dependent on radio hardware, only relative
values of interest
◮ Correlates directly to number of packet transmissions
www.nps.edu 33 / 44
Power Consumption
20 40 60 80 100 Buffer Size (MB) 0.820 0.822 0.824 0.826 0.828 0.830 0.832 0.834 0.836 Average Power Consumed per Node (W)
GAPR GAPR2 GAPR2a Vector No Limit Vector Epidemic Centroid
Figure: Omaha per-node power cons., 95% confidence intervals; 54 Mb/s base radio
◮ Absolute values dependent on radio hardware, only relative
values of interest
◮ Correlates directly to number of packet transmissions
www.nps.edu 33 / 44
Introduction Background Prior Work Implementation Evaluation Validation Conclusion
www.nps.edu 34 / 44
Methodology
Goal: Characterize the same simulation on the ONE and ns-3
◮ Mobility produced on ONE, trace input to ns-3 ◮ Comparable traffic generation ◮ Reminder: The ONE abstracts away lower layers of network ◮ Does not model MAC contention, segmentation, network
headers, transport, or even routing control message overhead
www.nps.edu 35 / 44
Delivery Probability Validation
Bold Alligator Helsinki Omaha
Protocols
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Message Delivery Ratio
ns-3 ONE
Figure: Message delivery ratio compared across simulators
◮ MDR is consistently lower in ns-3 ◮ Consistent with modeling control message overhead
www.nps.edu 36 / 44
Latency Validation
Bold Alligator Helsinki Omaha
Protocols
1000 2000 3000 4000 5000 6000 7000
Average Latency (seconds)
ns-3 ONE
Figure: Message latency compared across simulators
◮ Message latency is consistently higher in ns-3 ◮ Consistent with modeling control message overhead
www.nps.edu 37 / 44
Overhead Validation
Bold Alligator Helsinki Omaha
Protocols
50 100 150 200 250
Message Replication Overhead Ratio
ns-3 ONE
Figure: Message replication overhead compared across simulators
◮ Message replication overhead varies widely between the two ◮ Not predictable a-priori
www.nps.edu 38 / 44
Introduction Background Prior Work Implementation Evaluation Validation Conclusion
www.nps.edu 39 / 44
Takeaways
◮ We have implemented an effective DTN routing model in ns-3 ◮ Captures need for message-centric data exchange ◮ Integrates with existing network stack ◮ Consistent with performance expectations relative to the ONE ◮ Results indicate that “real” performance (e.g. overhead) can
not be estimated from ONE results
www.nps.edu 40 / 44
Future Work
◮ Validate against real implementation of Epidemic routing ◮ Further modularize code to provide base classes for future
DTN routing protocols
www.nps.edu 41 / 44
Thanks for watching!
www.nps.edu 42 / 44
Alenazi, M. J. F., Cheng, Y., Zhang, D., and Sterbenz, J. P. G. (2015). Epidemic routing protocol implementation in ns-3. In Proceedings of the 2015 Workshop on ns-3 (WNS3), pages 83–90, New York, NY, USA. ACM. Kang, H. and Kim, D. (2008). Vector routing for delay tolerant networks. In Proceedings of the IEEE 68th Vehicular Technology Conference, pages 1–5. Keränen, A., Ott, J., and Kärkkäinen, T. (2009). The ONE simulator for DTN protocol evaluation. In Proceedings of the 2nd International Conference on Simulation Tools and Techniques (SIMUTools), pages 55:1–55:10, New York, NY, USA. ICST. Killeen, K. M. (2015). GAPR2: a DTN routing protocol for communications in challenged, degraded, and denied environments. Master’s thesis, Naval Postgraduate School, Monterey, CA. Mauldin, A. N. (2017). Comparative analysis of disruption tolerant network routing simulations in the ONE and ns-3. Master’s thesis, Naval Postgraduate School, Monterey, CA. Rohrer, J. P. (2018). Geographic centroid routing for vehicular networks. In Proceedings of the Seventh International Conference on Advances in Vehicular Systems, Technologies and Applications (VEHICULAR), Venice, Italy. www.nps.edu 43 / 44
Rohrer, J. P. and Killeen, K. M. (2016). Geolocation assisted routing protocols for vehicular networks. In Proceedings of the 5th IEEE International Conference on Connected Vehicles (ICCVE), pages 1–6, Seattle, WA. Rohrer, J. P. and Mauldin, A. N. (2018). Implementation of epidemic routing with ip convergence layer in ns-3. In Proceedings of the 2018 Workshop on ns-3 (WNS3), Surathkal, India. ACM. Scott, K. and Burleigh, S. (2007). Bundle Protocol Specification. RFC 5050 (Experimental). Vahdat, A. and Becker, D. (2000). Epidemic routing for partially-connected ad hoc networks. Technical Report CS-200006, Duke University. www.nps.edu 44 / 44