Adaptive Optimal Buffer Management Policies for Realistic DTN

Delay Tolerant Networks (DTNs) are able to provide communication services in challenged networks where the end-to-end path between the source and destination does not exist. In order to increase the probability of delivery, DTN routing mechanisms may require nodes in the network to store and carry messages in their local buffer and replicate many copies. When the limited buffer is consumed, choosing the appropriate messages to drop is critical to maximizing the system performance. Current methods for this are sub-optimal or assume unrealistic conditions. In this paper, we propose an adaptive optimal buffer management scheme for DTN in the situations where the bandwidth is limited and messages vary in size. In our proposal, the mobility model is adjusted according to the nodes' historical meeting information, and the message dropping policies are designed to optimize certain network performance goals, such as maximizing the average delivery rate or minimizing the average delivery delay. Furthermore, we theoretically prove that the designed message dropping policies are optimal to achieve the best system performance.