Article Preview
Top1. Introduction
Delay Tolerant Networks (DTN) (DTN Research Group, www.dtnrg.org) provides communications between source and destination nodes where there is no direct path between any source and destination at a same time. In such environment, intermediate nodes are involved in order to receive messages from sources and carry them for a while until forwarding opportunities arise. Hence, messages are routed based on store-carry-forward manner. However, as the network may have limited resources such as buffer size, energy, and bandwidth, the network performance, namely delivery ratio, delivery delay and, overhead is affected by the following problems (Balasubramanian, Levine & Venkataramani, 2007). Firstly, if messages are flooded throughout the network, congestion may happen in the network which results in the network with high network overhead, low delivery ratio and large delays (Elwhishi, Ho, Naik & Shihada, 2013; Krifa, Barakat & Spyropoulos, 2012; Krifa, Barakat & Spyropoulos, 2008). On the other hand, in order to reduce the signaling overhead, number of replicas may be limited that in turn, due to low dissemination rate, decreases delivery ratio and increases delay.
Many routing protocols are proposed to address the issue of routing in DTNs (Elwhishi, Ho, Naik & Shihada, 2013). These routing protocols are categorized into two groups (Iranmanesh, Raad & Chin, 2014; Nelson, Bakh & Kravets, 2009) (1) Flooding-based and (2) Quota-based. In the first group, although nodes are able to replicate messages infinitely, these protocols may reduce the number of replications such that the network overhead decreases. It should be noted that when the network resources are not limited, despite of high energy usage (Juang, Wang & Martonosi, 2002), these protocols provide a high network performance such as high delivery ratio and low delivery delay. However, the overhead of these protocols is high and they suffer from low delivery ratio when the network resources are limited. Contrary to flooding based routing protocols where the number of replicas is dependent on the number of encounters, quota protocols limit the number of replicas for each generated message. These protocols reduce the network overhead and consume less energy compared to flooding based protocols. However, these protocols suffer from low delivery ratio and large delays due to low messages dissemination. In addition, without respect to the network’s capacity i.e., nodes’ buffer size and nodes’ service rate, the number of replicas for all messages is fixed. This means if a quota protocol generates a large number of replicas for each message when congestion happens in the network, the protocol similar to flooding protocols will suffer from high ratio of dropped messages. On the other hand, if the number of generated replicas is small and traffic is low in the network, the protocol suffers from low delivery ratio.