Sincronização de tempo e alocação de recursos em sistemas da internet das coisas

Abstract

The Internet of Things (IoT) paradigm consists of the dynamic interconnection of autonomous objects in order to extend the Internet to the physical world. To meet application requirements, IoT architectures separate the system in three tiers: things tier, edge tier and cloud tier. Each tier must provide infrastructure services that help to leverage the applications performance while being efficient. In this thesis, two problems of the IoT systems infrastructure are addressed: the time synchronization on things tier and the resource allocation in edge tier. Time synchronization is the infrastructure service that aims to keep all nodes with a common view of the time. The choice of the reference node, which provides the reference time, is a strategy that can mitigate the synchronization error between the network nodes. In this thesis, an energy-efficient distributed time synchronization algorithm was proposed that aims to select the node that minimizes its distance to the other nodes in the network. The algorithm orchestrates the configuration of timers on sensor nodes, generating a mechanism for triggering the selection of the reference node without additional synchronization message exchanges. The resource allocation at the edge tier aims to choose physical and virtual available resources for use by the applications. In this thesis, a new distributed resource allocation algorithm is proposed that supports the heterogeneity of applications and devices of the things tier. The resource allocator explores the decentralized nature of the edge tier, promoting collaboration in the allocation process. It also provides an efficient use of resources while meeting the latency, heterogeneity and data freshness requirements of IoT applications, respecting their different priorities.