Abstract
In this work we share lessons learned from our previous installations of wireless sensor networks and hope these lessons will help other research groups to install real-world applications more efficiently.First, we rediscovered the well-known KISS principle (Keep it Simple, Stupid), which holds true for real-world deployments of sensor networks.Second, we underestimated practical issues in outdoor installations, and it led to unexpected and time-consuming problems. For instance, we did not notice that the selected voltage regulator works reliably only in temperatures above zero degree Celsius, leading to packet losses and a long debugging process.Third, apart from extensive software tests before deployment, our protocols and applications include also self-healing instructions. They detect software bugs on run-time and restart motes if needed.With all these three major steps, we were able to run real-world sensor networks for several years without major problems.
Highlights
There are several ways to ensure that hard- and software is working as expected, such as various system-tests, different behavioral simulations, and testbeds
Researchers usually cannot carry out such extensive tests with a limited budget of research projects
In this paper we present lessons learned from our deployments: how to install and run reliably WSN regardless of limited testing resources
Summary
There are several ways to ensure that hard- and software is working as expected, such as various system-tests, different behavioral simulations, and testbeds. Scientific WSN include novel solutions and they resemble prototypes rather than products, due to limited testing and debugging resources Even such WSN prototypes should work reliably for a long time. We replaced our complex medium access control protocol [4], which included advanced features like clock drift prediction, with a simpler solution based on preamble sampling [5] Another problem with testing in that some bugs are hard to spot before the real deployment, because they do not occur under testing conditions [1]. Due to some minor bugs in the radio module and the routing protocol, not spotted during in-lab tests, our motes lost plenty of frames, leading to failures To attack such problems, we include self-healing mechanisms in our software, which detect software bugs on run-time and restart the affected mote.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
