Performance Evaluation of Replication-Based DTN Routing Protocols in Intermittently Connected Mobile Networks Md. Sharif Hossen and Muhammad Sajjadur Rahim Department of Information and Communication Engineering University of Rajshahi Rajshahi–6205, Bangladesh sharif5613@gmail.com, sajid_ice@ru.ac.bd Abstract—Intermittently Connected Mobile Networks (ICMNs) are sparse wireless networks where most of the time an end-to-end path does not exist from source to destination. These networks are characterized by opportunistic connectivity, long and variable delay, asymmetric data rate and high error rate. Hence, they can be appropriately modeled as Delay-Tolerant Networks (DTNs). In this paper, the performance of replication- based DTN routing protocols, namely Epidemic, PRoPHET, MaxProp, RAPID, Binary Spray-and-Wait, and Spray-and- Focus has been investigated in an ICMN scenario against varying message generation rates and number of mobile nodes, respectively. The performance is analyzed on the basis of three metrics: Delivery Probability, Average Latency, and Overhead Ratio. Opportunistic Network Environment (ONE) simulator is used as the simulation tool for evaluating these performance metrics. The outcome of this work shows that for the ICMN scenario and given simulation setting, the best DTN routing technique is Spray-and-Focus, whereas Epidemic routing exhibits the worst performance in terms of all the metrics considered here. Keywords—delay-tolerant network; intermittently connected mobile network; opportunistic network environment simulator; epidemic; prophet; maxprop; rapid; spray-and-wait; spray-and- focus; message replication; routing protocol; simulation I. INTRODUCTION AND RELATED WORKS Delay- and Disruption-Tolerant Networks, also known as DTNs, enable transfer of data for intermittently connected mobile nodes. These networks are characterized by lack of connectivity due to node mobility and typically sparse topologies, network partitions, long and variable propagation delay, and high bit error rate of the communication channels [1]. These situations eventually lead to the lack of an instantaneous end-to-end path from source to destination. In such challenging environments, popular ad-hoc routing protocols such as Ad hoc On-Demand Distance Vector (AODV) [2] and Dynamic Source Routing (DSR) [3] cannot perform well for routing data successfully, as a continuous end- to-end path is essential before forwarding data packets through the network. Intermittently Connected Mobile Networks (ICMNs) [4] are featured by intermittent connectivity and temporarily broken links, which prevent the common ad-hoc routing protocols to transfer data from source to destination successfully [5]. When instantaneous end-to-end paths do not exist, the routing protocols adopt the “store-carry-and-forward” approach, which exploits the mobility of nodes to route data. In this approach, data is moved from the source to the next available node and stored there, waiting for an opportunity to forward the data. If present, mobile nodes can carry the stored data while moving, and search for opportunities to forward the data to other nodes towards the destination. Overall, these techniques provide eventual data delivery with a certain probability [6]–[8]. A common technique to maximize the delivery probability of the packets is to replicate data packets to different nodes so that at least one of the copies will successfully reach its destination with high probability [9]. Routing protocols that behave this way are called replication- based. On the other hand, the routing protocols that do not replicate packets are called forwarding-based [10]. In our previous work [11], the performance of several DTN routing protocols such as Epidemic, PRoPHET, and Spray-and- Wait was investigated through simulation in an ICMN scenario, and according to the simulation results, Binary Spray-and-Wait showed the best routing performance. In this paper, the performance of replication-based DTN routing protocols has been evaluated, as applied to ICMNs, with focus on Epidemic [9], PRoPHET [12], MaxProp [6], RAPID [13], Binary Spray-and-Wait (B-SNW) [14], and Spray-and-Focus (SNF) [15] on the basis of three metrics: Delivery Probability, Average Latency, and Overhead Ratio. The rest of the paper is organized as follows. Section II describes the replication-based DTN routing protocols under investigation. Section III depicts the simulation tool and setup. Section IV gives the analysis of the obtained results from our simulation. Finally, section V provides the concluding remarks and future directions of this work. II. REPLICATION-BASED DTN ROUTING PROTOCOLS UNDER INVESTIGATION DTN routing protocols are classified into two basic schemes: single-copy and multi-copy. In single-copy schemes, a single copy of each message is forwarded through the network, which is called forwarding-based routing. Multi-copy schemes forward several copies of the same message to the network, i.e., replicate messages, and hence they are called replication-based.