Exploiting unstable paths in urban-scale wireless sensor networks

Abstract

Wireless sensor networks (WSNs) are composed of small autonomous devices called sensor nodes, used to measure environmental phenomena. These sensor nodes are typically unattended battery-powered devices that typically have very limited storage, computation and communication capabilities. In particular, they have limited transmission range, and therefore their data need to be collected via multi-hop routes, thereby consuming energy at all intermediary nodes. Sensor nodes deployed near a gateway may experience high volumes of traffic and congestion, hence becoming a bottleneck in the network, and therefore may consume much more energy, resulting in network partitions. In urban environments, mobile devices such as smart phones, tablets and laptops are present. These devices generally have much better resources than sensor nodes in terms of storage, processing and communication capabilities. In particular, they may be able to establish long-range communication with distant servers via the Internet. While these devices also have a limited energy budget, they can typically be recharged easily. The presence of such mobile devices might be exploited to (i) reduce energy consumption at resource-constrained sensor nodes, (ii) bypass congested areas of the sensor network, (iii) re-establish connectivity in the case of network partition and (iv) improve end-to-end delays. The majority of the existing work that considers the presence of mobile devices in WSNs, however, does not exploit these devices to reduce traffic in the WSN. Furthermore, existing protocols that utilize mobile devices in WSNs do not consider all the aforementioned objectives or address how to conciliate them. In addition, in the MANET (mobile ad hoc network) community, the few protocols that redirect traffic towards more capable devices are not suitable for resource-constrained WSNs, due to their reliance on route discovery and repair protocols, and poor or non-existent congestion controls. Using mobile devices is difficult, because they may only appear for a short time, due to mobility and user control. For this reason, the paths established by such devices are inherently unstable, rendering route discovery a redundant task. Many existing routing algorithms do not exploit routes through such devices because of their unstable nature. Also, as these devices may not always be the closest ones to the destination, pure geographic routing algorithms typically do not take advantage of them either. This thesis explores the possibility of exploiting such unstable paths via more capable mobile devices to achieve the aforementioned objectives. More precisely, the thesis addresses the problem of finding the best next hop in a WSN in order to outsource the communication load to mobile devices while systematically balancing the achievement of congestion reduction at critical WSN nodes, energy conservation at resource-constrained nodes, reduction in end-to-end delays and fault tolerance with only acceptable degradation in the data delivery ratio due to mobility and disappearance. In particular, this research investigates how protocols and mechanisms can be designed in a way to keep processing intensive tasks on mobile devices, thereby making such protocols practical for resource-constrained sensor nodes. This thesis empirically shows that, in WSNs with mobile devices, just by placing the baseline discovery mechanism at more capable devices, the energy consumption of the sensor nodes reduces significantly, and by utilizing location awareness, the discovery time of the sensor nodes is improved. The empirical evidence also shows that by utilizing mobile devices appropriately, the communication load and congestion at resource-constrained critical sensor nodes is reduced, as well as the energy consumption and end-to-end delays. The results also show that in the case of network partitions, mobile devices provide alternative paths and increase the data delivery ratio.