Um protocolo de roteamento para redes sem fio adaptável por regras de aplicação

Abstract

The exchange of information among the layers of the communication stack has been frequently used in Wireless Sensor Networks (WSNs) to increase the performance of the network. Such interaction among layers enhances the reliability and eciency of the communication, despitethe severely restricted resources found on sensor nodes. However, such optimizations impose a huge burden on the developer, as an unsound decision might decrease the overall performance or even hinder the operation of the network. This dissertation presents a proactive routing protocol, called PROC (Proactive ROuting with Coordination). The protocol is optimized to static continuous data dissemination WSNs, thatis, every node periodically sends data towards the access point. To our knowledge, PROC is the first protocol devised specically to continuous dissemination networks which provides simple optimization mechanisms, allowing the applications to eortlessly optimize routing operation to their needs. This interaction takes place using the application rules, which allow PROC to adapt in runtime to any change in the environment. Routes are established in a two-phase process. In the rst phase, PROC determines which nodes will forward data, based on the information provided by the application rules. The second phase ensures that all nodes have valid routes. Unexperienced programmers might optimize the routes created by PROC, as no knowledge concerning the operation of the protocol is required. The protocol also includes fault-tolerance mechanisms that increase its resiliency against silentfailures. Simulation results showed that PROC increases the average network lifetime by up to 12% when compared to EAD and TOSB (a simplied version of the TinyOS Beaconing protocol). PROC also quickly recovers from failed nodes due to its fault-tolerance algorithms. PROC ishighly scalable, since the amount of route maintenance messages sent by each node is independent of node density or network size. urthermore, the protocol presents a low memory footprint as nodes only store information concerning one-hop neighbors. Finally, the implementation of PROC in the Mica2 platform, which is the most frequently used platform by researches in the eld, is available for download.