Evaluation Of Ferry Based Opportunistic Message Dissemination With Deterministic Inter Contact Times

Abstract

Opportunistic Networks are characterized by intermittent connectivity and volatile topology. In such a network scenario, traditional Mobile Ad hoc Network (MANET) routing protocols cannot work efficiently. Hence, a novel way of message forwarding is required that operates even when an end-to-end path between a source and destination pair may never exist. Delay Tolerant Networks (DTNs) are an approach to alleviate network partitioning by working on the `store-carry-forward' paradigm [4]. Opportunistic Networks encompass DTNs and typically involve mobile devices that exchange data by taking advantage of their proximity to other nodes. Nodes in opportunistic networks may be heterogeneous with respect to their movement capabilities characterized by versatility of visits to different locations as well as speed of mobility. Special nodes called data mules have been used to ferry the data from restricted or less capable nodes to the message destination. In this thesis, we explore the implications of ferry based message dissemination in terms of average latency and delivery ratio performance metrics. We analyze two important metrics of measuring network performance, namely, inter contact times and contact frequency and their impact on performance metrics. We derive an expression for contact frequency per hour and show that a deterministic cyclic ferry network has a constant per hour capacity. Through simulations we show how an overlay network of agile ferries can be used to improve network performance and discuss the impact of such movement and routing heterogeneity on the network. Using the analogy of a deterministic cyclic city bus network, we show empirically how increase in number of travelers or regular nodes effects the network performance under routing based on deterministic ferry schedules.