An efficient algorithm for scheduling sensor data collection through multi-path routing structures

Abstract

Multi-path routing is essential to improve the robustness of sensor data collection in error-prone wireless communication environments. Besides the robustness against communication failures, both the energy efficiency and time efficiency are also of primary importance in sensor data collection due to the limited energy supplies of sensor nodes and the real-time nature of sensor network applications. Contention-free time division multiple access (TDMA) protocols have the potential to reduce the energy consumption and the latency of sensor data collection. To collect sensor data using a TDMA protocol, sensor nodes need to be assigned appropriate time slots for transmitting and receiving data prior to the data collection process. We note that the distributed TDMA scheduling process for sensor data collection incurs overhead costs of energy consumption and time latency. However, these overhead costs are usually overlooked, especially when multi-path routing is used to collect sensor data. In this paper, we propose an efficient scheduling algorithm for data collection through multi-path routing structures in wireless sensor networks. The objective of our scheduling algorithm is to reduce both the message complexity and running time of the scheduling process as much as possible. In addition, we also develop a method for deriving a lower bound on the shortest possible length of the data collection schedule that can be generated by any algorithm. The lower bound latency estimation offers a practical method to evaluate the efficiency of data collection schedules produced by scheduling algorithms. Extensive experimental results show that the proposed scheduling algorithm significantly reduces the number of messages transmitted during the scheduling process and the running time compared to an existing scheduling algorithm. The length of the data collection schedule produced by our algorithm is normally within 1.9 times of the lower bound estimate across a wide range of network settings.