Gathering data with packet-in-packet in wireless sensor networks

Abstract

In many wireless sensor network (WSN) mission-critical applications, a collector, e.g., a drone acting as a data sink, is required to gather data from all the source nodes in the area of interest. It is, thus, an all-to-one scenario. Data from each node (e.g., local temperature or its binary status) normally only consists of several bytes. Furthermore, often, data collection is expected to be done as fast as possible, especially for mission-critical applications or applications where a battery-powered sink is used (e.g., a drone). Thus, it is preferred that there is no preparation phase for the network to schedule or establish any routes in advance. Concurrent transmission (CT), a novel communication paradigm, has been shown to effectively achieve reliable and energy-efficient flooding in low-power wireless networks. With multiple nodes exploiting a receive-and-forward scheme, this technique requires no routing and works effectively in flooding-based networks, i.e., in one-to-many scenarios. Hence, CT-based communication is a good choice for fast data collection. However, a kind of global scheduling is essential for data collection in CT-based WSNs, in order to globally schedule a node that initiates a flooding in the network. In this article, to comply with the requirements of the aforementioned data collection application, we propose Packet-in-Packet (PiP), an energy-efficient paradigm requiring no global scheduling for timely data collection in low-power WSNs. PiP builds on concurrent transmissions and uses a packet concatenation capability to gather single-hop information in a best-effort manner. PiP does not require any pre-scheduling of traffic in the network, and thereby significantly reducing collection time. We compare PiP with a number of state-of-the-art protocols by extensive experiments in two testbeds, FlockLab and Indriya, respectively. Experimental results show that PiP achieves high reliability and low latency in all the scenarios in the real-world testbeds. Specifically, when the transmission power of nodes is set to 0 dBm, PiP takes only 287 ms (at least 1.8 × faster than the state-of-the-art protocols) to complete a 26-to-one data collection in FlockLab and 965 ms (1.1 × faster than a state-of-the-art protocol LWB) to complete a 42-to-one collection in Indriya, respectively.