Reaching Consensus in Opportunistic Networks

In large systems with many sensors and devices that collect information, the infrastructure costs can reach extremely high values, since all the nodes in the network need to be connected to a server so they can upload their data and have them processed. However, in order to reduce such costs, the sensors and devices can communicate with each other, exchanging data when in close proximity. Thus, an opportunistic network (ON) is created, where devices follow a store-carry-forward paradigm, deciding upon each contact whether the encountered node is the next hop for the data they are carrying. Besides their many advantages, opportunistic networks present several challenges, one of the most disputed being the achievement of consensus between the nodes of the network. The lack of a central entity which could gather information and validate messages leaves this task to the goodwill and collaboration between nodes. Consensus is required in order for the nodes to depend on and trust the data that they receive in the network. In this paper, we address the problem of reaching consensus in opportunistic networks. We propose an algorithm for consensus in ONs and show that it is able to correctly select several leaders for each community of devices, which then collaborate between each other to reach consensus. Furthermore, we present two scenarios where opportunistic networks can be employed and consensus is required.