Optimizing Performance of Delay Tolerant Network (DTN) Routing Protocols in Post Disaster Response Networks

Abstract

To achieve performance optimization in opportunistic Delay Tolerant Networks (DTNs), most flooding based routing algorithms like Epidemic and Probabilistic Routing Protocol using History of Encounters and Transitivity (PRoPHET) transmit multiple copies of a message to several custodians. This results in quick depletion of valuable network resources. On the other hand quota based protocols, such as Spray and Wait, reduce routing overhead by limiting the number of redundant transmission of a message but achieve lesser throughput. This makes both these types of approaches directly inapplicable in a post disaster response network where maximum delivery ratio with minimum routing overhead is expected. Abstract Recent research suggests that with proper buffer management policies implemented at nodes, we can intentionally optimize any one or a number of routing metrics like message delivery ratio or delay etc. Similarly we can increase the message delivery ratios of flooding based schemes in resource constrained environments without increasing routing overhead. This makes them comparable to quota based protocols in terms of routing overhead while maintaining higher delivery ratios. In this thesis we assess the performance of flooding based protocols (e.g. Epidemic and PRoPHET) with several existing buffer management schemes in order to determine their impact. We also propose a novel buffer management scheme appropriate for a resource constrained environment like the post disaster response network, where communication is enabled through small hand held devices with limited buffer sizes and transmission capabilities. We further prove that to curtail the negative impact of buffer overflow in such congested and constrained environments, buffer management policies should carefully select the messages to be discarded. For this purpose we have proposed a dynamic threshold size based selection mechanism where the difference between the arriving message size and the available buffer space is calculated to determine the size of the message to be dropped, incase of buffer overflow. Due to the adaptive nature of our proposed scheme, it adds randomness to the message selection process. This gives each message equal chance of being successfully delivered. Simulations show that our proposed scheme named as Size Aware Drop (SA-Drop), significantly improves the performance of the basic flooding based DTN routing protocols by increasing their message delivery probability and by controlling routing overhead. When used with PRoPHET, our proposed policy outperforms Encounter Based Routing (EBR) a quota based protocol, specifically designed for post disaster response networks. This makes it suitable for enabling communication in a post disaster scenario.