Comunicação de dados baseada no escalonamento de trabalho de nós vizinhos em redes de sensores sem fio

Abstract

The best way to save energy in Wireless Sensor Networks (WSNs) is to have sensor nodes go to sleep whenever possible, making them periodically unavailable. On the other hand, this unavailability may affect network tasks such as data communication. Media Access Control protocols for these networks typically deal with the problem of scheduling the data communication when the next hop is sleeping. One of the approaches used by those protocols (proactive policy) is to share each node's working schedule with its neighbors and have sender nodes to schedule their transmissions to their next hop awaking moment. This sharing process is typically based on message exchanges, which increases the network energy consumption. Another approach (reactive policy) is to discover when the next hop is going to be awake, and, have a node to send a message at that moment. Although this approach eliminates the sharing cost, its discovery process introduces additional packet transmissions as well. Given the limitations of both polices, this thesis proposes three solutions to eliminate or minimize the sharing and discovery costs of a node's schedule. The first is a hybrid policy that combines both existing approaches. In this case, each sender node inserts its working schedule into its packet and each receiver node saves this information to use in future transmissions. The second contribution is a combined design involving the reactive policy and the receiver-based routing. This routing technique increases the number of next hop candidates and can minimize the cost of the reactive discovery. In this combined design, a sender node has to discover the working schedule of some neighbor node instead of a specific neighbor. Our last contribution is the Gossiping Using the Energy Map (GEM) protocol, a cross-layer design solution involving the network and MAC layers to perform broadcasting when sensor nodes frequently sleep (e.g., during 99% of their lifetime). In this case, when a sender node sends a packet, not all neighbors will be awake. Most of the solutions proposed for broadcasting in WSNs ignore that sensor nodes can frequently sleep. The main idea of this protocol is to create a broadcasting tree based on the network energy map (the information about the amount of energy available at each part of the network) and make each sensor node share information with only one neighbor node that belongs to this tree. Furthermore, this protocol explores the energy map to change the basic operation of nodes localized inside low-energy regions. Simulation results reveal that our solutions reduce the number of transmissions and energy consumption when compared with other solutions proposed in the literature and evaluated in this work.