LAUSR

Delay/Disruption Tolerant Networks target environments suffering from the instability or lack of end-to-end paths. Store-carry-and forward principle aims to sustain data sessions, and data replication to increase the probability of on-time delivery. However, these techniques require efficient scheduling and buffer management, to comply with limited resources availability (i.e., communication duration, storage). Multiple existing schemes aim to improve, or even optimize the resources usage. Nevertheless, their majority considers equally important application sessions. The few proposals considering different traffic classes, fail to provide real QoS guarantees. In this paper, we formulate the problem of maximizing the performance, subject to distinct QoS constraints (requirements) for each application class. We consider requirements related to delivery probability and delay. Then, we propose a distributed algorithm which: (i) guarantees satisfaction of the individual constraints, when this is feasible given the available resources, and (ii) allocates any remaining resources optimally, to maximize the desired performance metric. We first consider homogeneous mobility, and then extend our analysis to heterogeneous contact rates and sparse contact graphs, that better correspond to real life mobility. Simulation results, based on synthetic and real mobility scenarios, support our theoretical claims and show that our policy outperforms other existing schemes (i.e., ORWAR [1] and CoSSD [2] ).