As technologies evolve, mobile devices are capable of storing, communicating, and disseminating large data files across the network. Communicating large data is particularly challenging in Delay Tolerant Networks because of lack of connectivity and short contacts among network nodes. To simplify the routing, existing works often ignore the data size and duration of a contact. That is, once transmitted, the data always successfully arrives at the receiver. Thus, these protocols are not suitable in the data-intensive mobile era. In this work, we propose a large data routing (LDR) protocol that chops large data into chunks and performs forwarding at the chunk level over multiple successive contacts. We also build a probabilistic model integrating the inter-contact time, contact frequency, and contact duration. The model is used to compute the edge weights in a contact graph and multi-hop delivery probabilities among network nodes. Simulations using real-world mobility traces show that our scheme can achieve up to 53% higher delivery rate compared to other routing strategies.

随着技术的发展，移动设备能够在网络上存储，通信和分发大型数据文件。由于延迟缺乏连通性和网络节点之间的短联系，在延迟容忍网络中进行大数据通信尤其具有挑战性。为了简化路由，现有作品通常会忽略联系人的数据大小和持续时间。即，一旦发送，数据总是成功地到达接收器。因此，这些协议不适用于数据密集型移动时代。在这项工作中，我们提出了一种大数据路由（LDR）协议，该协议将大数据切成块，并在多个连续的联系人上以块级别执行转发。我们还建立了一个概率模型，该模型整合了接触时间，接触频率和持续时间。该模型用于计算联系图中的边缘权重和网络节点之间的多跳传递概率。使用现实世界中的移动轨迹进行的仿真表明，与其他路由策略相比，我们的方案可以将传输率提高多达53％。