Sensor Simulation System Article No.1
ByIn the context of the DEHEMS project, the research and development stages of various modules have underlined the need for a fast and scalable means of generating large amounts of high quality real time sensor data. This is why, for the better part of this year, our research team here at the Institute e-Austria Timisoara has been working on a sensor simulation system that successfully meets the data generation requirements (quantity and quality) imposed by DEHEMS.
In short, here are the key points we had to take into consideration whilst developing the DEHEMS Sensor Simulation System (or DS3 as we have named it):
-
the need to generate very accurate sensor data for the data-analysis engine and pattern matching algorithms, preferably based on previously defined energy consumption patterns;
-
the need to rapidly generate and flood the storage engine with large amounts of real time sensor data in order to simulate the data output of the real-life monitoring system;
-
the ability to wrap the simulated data in XML, JSON and other custom formats and rapidly switch between these formats;
-
hardware constraints: the system had to be scalable in order to take maximum advantage of the underlying hardware architecture and was required to perform well on standard PCs (i.e. simulate more than 30.000 households).
Our first intention was to adapt an existing sensor simulation system for our needs. Such types of simulators abound in Wireless Sensor Network (WSN) literature and we have found some interesting candidates like ns-2 and SENS. However, after a careful analysis of these frameworks, we have decided that adapting/using such a framework for the DEHEMS project was not a viable solution. This was due to the fact that these complex network simulation frameworks imposed drastic speed penalties when adapted for our specific goals.
As such, we have decided to develop our own multi-purpose sensor simulation system. Its design is centered around the Event-Driven Architecture (EDA) principle, multi-threading and a two stage mechanism for simulating spot energy consumption profiles. We will present a more detailed description of the architecture of DS3 in our future posts.
Research Institute e-Austria homepage: www.ieat.ro/
