Network throughput maximization in unreliable wireless sensor networks with minimal remote data transfer cost

In this paper, we consider large-scale remote environmental monitoring (data gathering) through deploying an unreliable wireless sensor network in a remote region. The data monitoring center is geographically located far away from the region of the sensor network, which consists of sensors and gateways. Sensors are responsible for sensing and relaying data, and gateways are equipped with 3G/4G radios and can store the collected data from sensors temporarily and transmit the data to the remote data center through a third-party communication service. A service cost of using this service will be charged, which depends on not only the number of gateways employed but also the volume of data transmitted from each gateway within a given monitoring period. For this large-scale, remote, and unreliable data gathering, we first formulate a problem of maximizing network throughput with minimal service cost with an objective to maximize the amount of data collected by all gateways while minimizing the service cost. We then show that the problem is NP-complete and propose novel approximation algorithms. The key ingredients of the proposed algorithms include building load-balanced routing trees rooted at gateways and dynamically adjusting data load among the gateways. Finally, we conduct experiments by simulations to evaluate the performance of the proposed algorithms. Experimental results demonstrate that the proposed algorithms are very promising, and the obtained solutions are fractional of the optimum in terms of network throughput and the data service cost.