1. Research into the hybridization of the PRoPHET and ERP network routing algorithms George Mason University INFS 612 (Spring 2013) Project Group 4: Richard Joy, Miriam Joy, Serena Mei, and Suyog Parajuli

2.  Research Problem  Background on ERP and PRoPHET  Research Approach  Results  Demonstration  Conclusion and Questions

3.  Delay Tolerant Network (DTN) addresses challenges in disconnected, disrupted networks without end- to-end connection  Over time, DTN routing protocols have developed in two areas: forwarding-based and replication-based  Replication-based algorithms have higher message delivery rates, but are resource hungry  Problem: Is there a way to combine two replication- based algorithms which result in the same delivery rate with lower overhead?

4.  Mobile Sensor Networks  Sensors with wireless connectivity deployed over a geographic area  Periodically transmit their findings to a base station, perhaps for analysis or permanent storage.  These sensors may be small and have limited communication range, implying that they are not always able to establish a connected path (leveraging other sensors as routers) back to base stations.  Disaster Recovery/Military Deployment:  People, in addition to sensors, are deployed over an area with limited wireless coverage (i.e., few, if any, base stations).  Interplanetary/Spacecraft Communications

6.  ERP is a “greedy” replication algorithm  Nodes continuously replicate and transmit messages to newly discovered contacts who do not already possess a copy  Very effective at delivering messages  Extremely inefficient for resources

7. S C1C1 C2C2 D C3C3 C2C2 C3C3 DS C1C1 Time = t 1 Time = t 2 >t 1

8.  PRoPHET is a “utility-based” replication protocol  Not indiscriminate in its replication; uses an adaptive algorithm  Set of probabilities for successful delivery  Algorithm updates whenever a node is encountered.  Nodes that are encountered frequently have a high delivery predictability.  PRoPHET statistics are shared between each of the communicating nodes.

9. Orbit and probability of encounterMessage S B C D S B C D S B C D S B C D 12 3 4 Message Source: SMessage Destination: D

11. DTN Listener DTN Connector Connection List Message Stats Message Manager DTNNode JBDC Connector Database In our environment, the nodes have particular orbits; connection probabilities are assigned to reflect this design 20 Nodes 4 connection partners per node with 25% probability 30 messages

12.  Ran a series of tests incrementing the threshold for switching to ERP (if it is the better solution)  Each test (each threshold) was 40 minutes long  Time-to-live for each message was 128 connections 00.10.20.30.40.50.60.70.80.9 100% PRoPHET 100% ERP 40 min

13. 0 = 100% ProPHET 1 = 100% ERP

14. ERP first selected by node

15. 1234012341 12342 123431234412345

16. 12346 12349 12350 12351 1234712348

17. 12352 1235312354 12355 12356 12357

18. 1235812359

19.  Routing Under Uncertainty  Is information on location of target node reliable or being distorted by environmental conditions? How do you adjust for this?  Resource Allocation  How do you properly elevate messages for priority handling?  Managing buffers and still maximizing delivery  Integrating a forwarding algorithm  Performance  Average latency in delivering messages;  Average amount of system storage and bandwith consumed;  Amount of energy consumed in transmission (particularly applicable to mobile devices serving as carriers for others)  Reliability  Sending back acknowledgements from message recipient  Security  New area of research into “reputation” protocol  How to track messages through the entire path  Encryption