Standard deviation (SD)-based data filtering technique for body sensor network data

It has been observed that sometimes non-repeated, but redundant values are transmitted unnecessarily between body sensor network (BSN) and database server (DBS). The redundant value increases overhead without giving any conclusion. We have proposed a Standard Deviation (SD) based data filtering technique which improves the PDA performance by compacting the sensed data. We have introduced three categories (normal, critically below normal, and critically above normal) of the patient which are bound by predetermined boundaries. This span of boundary is known as 'Window Size'. This research work is restricted to controlling the size of windows so that the amount of sensed data being transmitted can be increased to utilise maximum network bandwidth. This window size variability is presented as 'elasticity'. The ultimate objective of collected data is to be analysed to identify patient's state so that medical expert may take appropriate actions accordingly. The proposed SD-based data filtering technique increase the size of window dynamically to accommodate more data without compromising the tone. Our experiments demonstrate that SD-based BSN data filtering can reduce the data up to 40%. The proposed SD-based technique is compared with 3-Sigma rule to show their applicability.