Delivery Ratio-and Buffered Time-constrained: Multicasting For Delay Tolerant Networks

Delay Tolerant Networks (DTNs) are a class of emerging wireless networks whichexperience frequent and long-duration partitions and long end-to-end delay. Many realnetworks fall into this category, such as wildlife racking and habitat monitoring sensornetworks, vehicular ad hoc networks, satellite networks, military networks, etc. Multicastservice supports the dissemination of data to a group of receivers, because most of thetime there dose not exist an end-to-end path from the source to the destination,multicasting in DTNs is a challenging work.The existing multicast routing algorithms for DTNs are mainly focused on improvingthe routing performance, but hardly consider the constraints existing in actual DTNmulticast applications. In this thesis, we investigate the constrained multicasting problems.Firstly, a pattern of constrained multicasting is defined, then based on the assumption thatthe pair-wise inter-contact time is exponentially distributed, we propose a Delivery Ratio-and Buffered Time-Constrained Multicast routing algorithm, named DBCM. DBCMconsiders the utility as delivery predictability within the maximum buffered time T andmakes routing decisions based on the utilities of current neighbors. To improve the contactutilization rate of DBCM, an enhanced scheme, called E_DBCM, is introduced. E_DBCMcan increase the delivery ratio without inducing additional overhead. Then， theperformance of DBCM is analysed theoretically in terms of message delivery ratio,average number of hops and end-to-end delay.Finally, the performance of DBCM and E_DBCM is evaluated by a series ofsimulations. In order to satisfy the assumption about node mobility, we also propose a newmobility model, named A_RWP, in which the pair-wise inter-contact time is proved to bedistributed exponentially. Simulation results show that DBCM and E_DBCM obviouslyoutperform the existing multicasting algorithms with respect to both delivery ratio androuting overhead. Furthermore, as a result of utilizing more contact opportunities,E_DBCM’s overall performance is better than DBCM.