A CONTEXT-BASED SOFTWARE RECONFIGURABLE SYSTEM FOR WIRELESS SENSOR NETWORK

Abstract

ABSTRACT Wireless Sensor Network application entails deploying thousands of wireless sensor nodes in unreachable locations. The inability to reconfigure each node in order to take on new tasks poses a serious challenge to the continued operation of the entire system. Several attempts have been made to address these challenges, of interest is one that exploits design-time knowledge of the application scenario dynamics to construct and implements a proactive runtime reconfiguration paradigm. However, It suffers two defects: the possibility of capturing all anticipated reconfiguration needs can be challenging, and the scarcely available memory space might not be sufficient to accommodate codes written to address these needs. Moreover, even if it does, there is the likelihood of redundant codes written to handle anticipated changes, which might never occur, and invariably taking up scarcely available memory spaces. This research work explores the use of context information to improve upon wireless sensor networks reconfiguration processes. The research’s aim is to develop a software system that dynamically reconfigures wireless sensor network operational functionalities optimally based on evolving application context. In order to demonstrate the benefits of the context based reconfiguration model, two contexts related input variables were used. The first variable is obtain using a metric tool (PDE) devised for extracting context information from the delta of two files (application related context). The second variable entails the battery energy level state of the sensor node taken as an operational demand related context. A robust inference engine was developed based on the inferred expert knowledge on memory related energy consumption pattern during the reconfiguration process. The pattern studied and presented explains how delta size and its orientation can influence energy consumption while reprogramming sensor nodes. The resulting output from the fuzzy logic system controls when and which one of the reconfiguration approaches should be implemented in order to prolong the battery life. The model's performance was evaluated on an OMNet++ simulation platform using pilot data obtained from a testbed composed of Microchips’ PIC32MX320F128H microcontroller and MRF24J40MB transceiver. In a network of six nodes, two were equipped with the developed model capability and the others were not. The overall energy expended as read, erase and write were obtained from each node for the purpose of comparison. Results obtained show that 65% of energy expended during the erasure procedure is saved in nodes that adopt the context based reconfiguration model. Similarly, 45% and 69% reduction in energy consumption were obtained for the read and write procedures respectively. The research work was able to emphasise the benefits of identifying, employing and managing the impact of contextual information (Application/operational related) during wireless sensor network reconfiguration procedure.