Divide-and-Conquer for Internet-of-Things: Discrete vs Continuous Groups

Abstract

Application of Divide-and-Conquer paradigm has been quite popular in addressing scalability issues in the design and development of protocols for large distributed systems. However, often in these applications, apart from appropriate formation of the groups, it is also essential for the neighboring nodes to smoothly interact with one another to achieve the final goal. Traditional discrete group-division strategies while successfully forms the groups, fails to achieve the later due to hard inter-group separation. In this work, we propose the concept of continuous-groups which enables a system to take advantage of Divide-and-Conquer without any explicit physical division of the network. While the design and implementation of discrete groups are quite straightforward, due to conflicting requirements, efficient realization of continuous-groups is quite non-trivial. In this paper, we report our study on the design and evaluation of an efficient strategy for creating continuous-groups and demonstrate its effectiveness w.r.t. many-to-many data-sharing in large systems.