A bus-based opportunistic sensing network

Abstract

Embedding sensors in urban buses is a promising strategy to deploy a city-wide Mobile Wireless Sensor Network (MWSN). The Internet of Things (IoT) paradigm can take advantage of bus mobility, achieving extended spatial coverage with fewer sensors as compared to a static setup. To overcome the limitations of IoT devices, buses can, in an opportunistic fashion, deliver data to Fog nodes located in bus stops. Fog nodes pre-process data and send it to the Cloud, which makes makes it externally available. The trade-offs are that urban buses only cover part of the city and that the frequency of the buses, and consequently of the data collection, is heterogeneous across the city. Additionally, buses may be unable to deliver the collected data on time due to the opportunistic communication to the Fog node. In this thesis, we present three main contributions. First, we propose a method to minimize delivery delays when there is a limited number of Fog nodes. In our second contribution, we propose a coverage metric to bus-based MWSNs and an optimization model to maximize coverage for a constrained number of participating buses. We also propose a more restrictive coverage metric, that takes into account the delivery delay and the measurement frequency of each sensed region, relating these metrics to different applications. We propose a metric for bus coverage contribution, showing that the importance of each bus depends on the applications the system serves. Finally, our third contribution is a prototype for SensingBus, a bus-based MWSN. We use real GPS traces from the bus fleet of Rio de Janeiro to validate our contributions. Among other remarks, our results show that if 16% of bus stops are equipped with Fog nodes, the maximum delivery delay is about 32 minutes. We also show that 32 buses can cover about 40% of the region covered by all the 6,075 buses of the fleet.