Improving fault tolerance support in wireless sensor network macroprogramming

• Ano de defesa: 2014
• Autor(a) principal: Nogueira, Guilherme de Maio
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Fault tolerance; Macroprogramação; Macroprogramming; Redes de sensores sem fio; Srijan; Tolerância a falhas; Wireless sensor networks
• Link de acesso: http://www.teses.usp.br/teses/disponiveis/45/45134/tde-03062015-214359/

Wireless Sensor Networks (WSN) are distributed sensing network systems composed of tiny networked devices. These systems are employed to develop applications for sensing and acting on the environment. Each network device, or node, is equipped with sensors and sometimes actuators as well. WSNs typically have limited power, processing, and storage capability, and are also subject to faults, especially when deployed in harsh environments. Given WSNs limitations, application developers often design fault-tolerance mechanisms. Although developers implement some fault-tolerance mechanisms in hardware, most are implemented in software. Indeed, WSN application development mostly occurs at a low level, close to the operating system, which forces developers to focus away from application logic and dive into WSNs technical background. Some have proposed high-level programming solutions, such as macroprogramming languages and frameworks; however, few deal with fault-tolerance. This dissertation aims to incorporate fault-tolerance features into Srijan, an open-source WSN macroprogramming framework based on a mixed declarative-imperative language called Abstract Task Graph (ATaG). We augment Srijans framework to support code generation for dealing with devices that crash or report meaningless values. We present our feature implementation here, along with an evaluation of the tool, demonstrating that it is possible to provide a macroprogramming framework with appropriate support for developing fault-tolerant WSN applications.