Dynamic deployment of applications in wireless sensor networks

Over the past decades, the progress inWirelss Sensor Network (WSN) technology, both in terms of processing capability and energy consumption reduction, has evolved WSNs into complex systems that can gather information about the monitored environment and make prompt and intelligent decisions. In the beginning, military applications drove the research and development of WSNs, with large-scale acoustic systems for underwater surveillance, radar systems for the collection of data on air targets, and Unattended Ground Sensor (UGS) systems for ground target detection. Typical civil WSNs are basically not complex monitoring systems, whose applications encompass environment and habitat monitoring, infrastructure security and terror threat alerts, industrial sensing for machine health monitoring, and traffic control. In these WSNs, sensors gather the required information, mostly according to a fixed temporal schedule, and send it to the sink, which interfaces with a server or a computer. Only at this point data from sensors can be processed, before being stored. Recent advances in Micro-Eletro-Mechanical Systems (MEMS), low power transceivers and microprocessor dimensions have led to cost effective tiny sensor devices that combine sensing with computation, storage and communication. These developments have contributed to the efforts on interfacing WSNs with other technologies, enabling them to be one of the pillars of the Internet of Things (IoT) paradigm. In this context, WSNs take a key role in application areas such as domotics, assisted living, e-health, enhanced learning automation and industrial manufacturing logistics, business/process management, and intelligent transportation of people and goods. In doing so, a horizontal ambient intelligent infrastructure is made possible, wherein the sensing, computing and communicating tasks can be completed using programmable middleware that enables quick deployment of different applications and services. One of the major issues with WSNs is the energy scarcity, due to the fact that sensors are mainly battery powered. In several cases, nodes are deployed in hostile or unpractical environments, such as underground or underwater, where replacing battery could be an unfeasible operation. Therefore, extending the network lifetime is a crucial concern. Lifetime improvement has been approached by many recent studies, from different points of view, including node deployment, routing schemes, and data aggregation Recently, with the consistent increase in WSN application complexity, the way distributed applications are deployed in WSNs is another important component that affects the network lifetime. For instance, incorrect execution of data processing in some nodes or the transmission of big amounts of data with low entropy in some nodes could heavily deplete battery energy without any benefit. Indeed, application tasks are usually assigned statically to WSN nodes, which is an approach in contrast with the dynamic nature of future WSNs, where nodes frequently join and leave the network and applications change over the time. This brings to issue talked in this thesis, which is defined as follows. Dynamic deployment of distributed applications in WSNs: given the requirements of WSN applications, mostly in terms of execution time and data processing, the optimal allocation of tasks among the nodes should be identified so as to reach the application target and to satisfy the requirements while optimizing the network performance in terms of network lifetime. This issue should be continuously addressed to dynamically adapt the system to changes in terms of application requirements and network topology.