Performance optimization in delay tolerant networks using backtracking algorithm for fully credits distribution to contrast selfish nodes

Abstract

Delay Tolerant Network (DTN) can be characterized as a heterogeneous network of nodes in which nodes mobility is high and resources are limited to transmit messages. In DTN, nodes use store, carry, and forward principle for delivering messages to the destination node. To forward messages through intermediate nodes may cause security issues in the network since there may exist a few selfish nodes. These nodes can affect the network's performance as they may drop messages due to their limited resources such as energy and storage capacity. To handle this kind of hazard, it is necessary to propose a mechanism that may decrease the degree of the selfishness of nodes and improve the network's delivery ratio by fully distributing credits to nodes. In this article, a credit-based mechanism has been proposed based on Combined Trust Value (CTV) of nodes in DTN. In the proposed mechanism, an agent is used to compute each node's trust value grounded of the number of messages relayed by sensor nodes. This trust value is used to distribute credits to the nodes in a distributed manner without any partiality with nodes. Backtracking approach is used to distribute credits to boundary nodes deserving credits but didn’t get credits by agent node. The proposed mechanism is implemented using ONE simulator, and the performance of the projected scheme is analyzed in comparison to existing techniques Dynamic Trust, SMART (Secure Multilayer credit based incentive Technique) and Credit-based. The results exhibit that the suggested mechanism is superior than existing techniques with reference to various performance metrics like 25% higher delivery ratio, 41% less overhead, 21% less average message delay, and 28% less packets dropped. The proposed mechanism might be helpful in scenarios where the degree of selfishness is high, and the distribution of credit follows a fully distributed approach rather than an existing partial distribution used in existing techniques.