Self-Organized Data Aggregation Techniques in Ferry Assisted Multi Cluster DTNs using a Novel Strategy of Evolutionary Game Theory

Abstract

Delay Tolerant Networks (DTN) can provide data connectivity for areas where internet cannot provide any end to end connectivity. In DTNs, store-carry-forward mechanism with the help of custody transfer technique provides reliable end-to-end data transfer where nodes which transfer data with custody called custodians. In this system, sometimes storage congestion may occur. This drawback can be improved by using special mobile nodes called message ferries. In such scenario, it is better to aggregate data to some custodians so that the message ferry can efficiently collect them and carries collected bundles to a base station referred to as a sink node. When there are several isolated clusters (formed by physically close wirelessly connected nodes), message ferry have to visit those clusters and collects bundles from custodians. When message ferry visits any individual cluster (referred to as intra-cluster visit or visit within cluster), sometimes it is not possible to visit and collect data from all nodes in that cluster within particular period of time. To resolve this problem, self-organized data aggregation technique is developed, where, with the help of the evolutionary game theoretic approach, the system can automatically select some limited number of nodes (custodians) called aggregators. As a result, the message ferry visits and collects bundles only from those aggregators in that particular cluster. In this technique, Aggregators are changed autonomously in each round. In self-organized data aggregation technique, the strategy (i.e., to become aggregator or sender) of selection of aggregators is modeled as a game in evolutionary game theory, where, each node will draw payoff after interaction. For strategy selection, Imitation Update Rule of evolutionary game theory was used previously. In this research, we compared the existing solution of Imitation Update rule with Birth-Death and Death-Birth Update Rules of evolutionary game theory and select the better update rules among them for selection of Strategy in Self-Organized Data Aggregation Technique for the cases of no retransmission and retransmission. For this, we proposed our mathematical model and after the numerical analysis, we have select Birth-Death Update Rule is as better update rule for the case of no retransmission and Death-Birth Update rule is as better update rule for the case of retransmission when degree is smaller and for higher degree, Imitation Update Rule as better update rule.