The majority of wireless sensor networks are built on bespoke platforms, that is, custom designed and built hardware with a light weight software stack. There are a number of advantages to this approach. First, the ability to closely match and minimise the resource requirements (e.g., power consumption and communications protocols) to those that are suitable for the intended deployment. Second, as an entire hardware and software stack is often designed or at least optimised for each deployment, the latest advances can be quickly incorporated. However, this model generally requires the expertise of hardware and software engineers to design and build the system. In turn, this increases the cost and tends to shift the focus away from the initial science towards the development of the wireless sensor networks. This paper explores the utility and practicality of building wireless sensor networks based on commercially available embedded single board computing platforms using standard consumer operating systems. Our test bed was built using Gumstix computing platform, running a Linux Operating System (OS) with a java-based middleware coupled to low-cost scientific grade sensors. Test deployments have found this to be a highly versatile solution, able to leverage the flexibility of commodity hardware and software while maintaining reasonable utility.
Published in:
Intelligent Sensors, Sensor Networks and Information Processing (ISSNIP), 2011 Seventh International Conference on
Date of Conference: 6-9 Dec. 2011