Wearable technology and mobile platform are becoming more and more popular in clinical context, such as health care or monitoring and a lot of research has been conducted on data processing algorithms. However, very little work has been done on real time feedback data calibration which is critical for training purpose work. In addition, as raw data have been usually noisy, real time data calibration become more and more difficult. Therefore, this paper discusses a Binary Encoded Symbolic Representation (BESR) based calibration algorithm to provide feedback in real time. This new algorithm employs a new pruning method using a Distance Filter with Starting Symbol Requirement (DFSSR) to provide real time feedback. We apply the proposed work on breathing exercise training. Experimental results show that the BESR reduces storage to a factor of 160X compared with raw data storage. The new calibration algorithm achieves time complexity O(1) for each sample which makes it suitable for real time application.