Skip to Main Content
The Delay Tolerant Network employs a store-carry-forward paradigm to enable bundle delivery in intermittent connected environments. A router relays a message only when a proper contact opportunity occurs, which results in prominent congestion issues. An appropriate scheduling and dropping policy can alleviate this situation and maximize contact capacity. We investigate the impact of delivery probability and number of message replicas over the efficiency of scheduling and dropping policies in probabilistic routing, and present a new way to utilize contact probability and delivery predictability to direct the design of scheduling policies. We also design a simple and distributed method to approximate the global number of replicas. We integrate these results and present a new scheduling and dropping policy. Simulation based on map based movement has shown that the new policy gains a delivery rate very close to global knowledge based policies, and it outperforms all other existing scheduling and dropping policies.