MAGNA: Middleware for dynamic and resource constrained sensor networks

Abstract

Traditionally sensor networks have been typically enterprise architecture driven, where a dedicated server is used to collect data and record in a database for analysis and historical reference. While modern day computing capabilities allow drilling through megabytes of information, it is not always the most cost- effective solution. The other alternative is the use of smart sensors, whereby each node is capable of routing data to any destination using a HTTP server. This not only escalates the cost of each node but also imposes severe strain on the network - thus necessitating developers and architects to evaluate the impact of swarming and "lazy routing". The notion of the proposed middleware solution (MAGNA) is to bring about context-based processing and intelligent data capture and/or filter right up to the sensor nodes through the use of powerful "sub-net" controllers. The paper will explore the application of MAGNA as a platform capable of offering ambient intelligence in a typical sensor network environment, integrating everyday appliances and our life-style together. In doing so, two primary methodologies will be discussed on which MAGNA is essentially based, namely the soft systems methodology and the banking model approach.